STRUCTURAL ECONOMIC CHANGE AND INTERNATIONAL MIGRATION FROM MEXICO AND POLAND

Abstract

In this article we use uniquely comparable data sets from two very different settings to examine how exogenous economic transformations affect the likelihood and selectivity of international out-migration. Specifically, we use data from the Mexican Migration Project to construct event history files predicting first U.S. trips from seven communities in the state of Veracruz, which until recently sent very few migrants abroad. Similarly, using data from the Polish Migration Project, we derive comparable event history files predicting first trips to Germany from four Polish communities, which also sent few migrants abroad before the 1980s. Our analyses suggest that the onset of structural adjustment in both places had a significant effect in raising the probability of international migration, even when controlling for a set of standard variables specified by other theories to influence migration propensity, such as the size of the binational income gap and various indicators of human and social capital.

In the last quarter of the 20^th century international migration became a truly global phenomenon and all developed countries became de facto countries of immigration, whether they cared to admit it or not (Massey et al. 1998). At present, roughly three percent of the world’s inhabitants live outside their country of birth, and this percentage is much higher in major immigrant-receiving nations such as Canada, the United States, and Germany (Zlotnik 1998, 2004). Moreover, non-permanent forms of transnational migration have rapidly increased during the last decades, thus adding further to the share of overall ‘migrants’ in a wider sense of the word.

It seems reasonable to argue that the recent upsurge in international migration stems from the transformation of the social and economic contexts within which mobility decisions of people are made. These changes in context in turn are driven by a powerful wave of globalization that crested during the 1990s, which constitutes the second wave of economic globalization in modern history (Hatton/Williamson 2006). Its foundations were laid by industrial nations in the wake of the Second World War. In an effort to avoid another global conflagration and promote prosperity, nations joined together to create a set of multilateral institutions such as the United Nations, the World Bank, the International Monetary Fund, the General Agreement on Tariffs and Trade, and most recently the World Trade Organization (Massey/Taylor 2004). These institutions enabled the recovery and growth of industrial economies during the 1950s and 1960s and facilitated the reemergence of international trade during the 1970s.

It was not until the 1980s, however, that the full potential for globalization could be realized because of the widespread abandonment of state-centered models of economic development (Massey/Taylor 2004). During the 1980s, Deng Xiaoping gained control of the communist party in China and used his power to move its economy away from state ownership and centralized planning toward markets and private ownership, leading to China’s reentry the global market economy during the 1990s (Marti 2002). At the same time, import substitution industrialization unraveled as a model for economic development in the Third World and political elites moved from statecentered to market-friendly policies. Finally, in the late 1980s a “velvet revolution” spread among nations of the former Soviet Block (Duberstein 2006) and led eventually to the fall of the Berlin Wall (Tames 2001), the collapse of communism in Eastern Europe (Goodwyn 1991), and the dissolution of the Soviet Union itself (Walker 2003).

Although studies by Liang (1999, 2001, 2004) have examined the migratory response to market expansion in China, few have examined the interrelation between international migration and structural economic change in Latin America and Eastern Europe. To date, most examinations of this relationship have been descriptive (cf. Iglicka/Sword 1999; Iglicka 2001; Jazwinska/Okolski 1996; Massey et al. 1987; Sipaviciene 1997). Thus it is hard to tell whether the migration dynamics observed in different context is driven by similar or different forces. In this paper, we adopt an analytic approach to study migratory responses to economic change in two very different national settings, Mexico and Poland. In each of these contexts we try to identify more detailed mechanisms that account for the link between the shift from state-centered to market-oriented economic regimes and the initiation of international migration. In doing so, we rely on comparable quantitative event history data gathered from household heads in Mexican and Polish communities.

Comparing Mexico to Poland provides an interesting test. Both countries experienced a notable increase in out-migration in recent decades and, although occupying very different geopolitical locations in the aftermath of World War II, their post-1945 dynamics of economic change display many parallels. These are briefly outlined in the first section (I). We continue by sketching basic mechanisms that can be derived from migration theory and that might account for the dynamics of recent migration (II). After describing our data sets (III), we document past and recent trends of emigration (IV). Then, we try to disentangle the mechanisms behind them (V). The paper ends with a discussion of similarities and differences that we find in our analyses (VI).

I. Structural Economic Change in Mexico and Poland

After World War II, both Mexico and Poland initially sought to use a strong, centralized state to achieve economic growth by creating protected economies that were isolated from global markets. In both cases, this state-led model of development worked well in the beginning, but its performance deteriorated as time went on. As the pace of technological change accelerated and systems of production diversified, the limits of a slow-moving, bureaucratic, state-centered strategy became more apparent and eventually both nations experienced a deep economic crisis that led to the imposition of economic reforms. Initially the reforms were small and grudging, but later they gave way to radical shifts toward privatization, market liberalization, and freer international trade.

1. Mexico

Like many developing countries in the period from 1945 to 1980, Mexico sought to achieve economic growth through a state-centered economic program known as Import Substitution Industrialization (ISI). Governmental authorities erected tariff and regulatory barriers to block the importation of consumer goods and then channeled capital to producers to meet the captive demand. Heavy industries were often owned by the government, along with basic industries such as telecommunications, electricity, and railroads (Bharat-Ram 1994).

ISI performed well as a development strategy for three decades, generating high rates of economic growth, significant capital accumulation and rising national income (Felix 1986). The model was particularly successful in Latin America (Portes 1997; Portes/Hoffman 2003). Despite widening inequalities and growing regional imbalances, countries throughout the region generally prospered as they urbanized and industrialized (De Ferranti et al. 2004). During the 1970s, however, economic growth under ISI stalled and conditions reached crisis proportions during the 1980s (Fiscia/Kovacs 1994).

Nations throughout Latin America systematically began to dismantle the state-centered apparatus of ISI during the late 1980s, lowering tariffs, eliminating regulations, privatizing state-owned firms, downsizing bureaucracies, floating currencies, and reducing controls on foreign investment, a set of policies that came to be labeled “the Washington Consensus” (Williamson 1990). In Mexico, a political crisis in 1968 led to massive expansion in state spending during the 1970s and ultimately to national insolvency and hyperinflation in the 1980s (Centeno 1994). In 1982 a new American-trained technocratic elite took power and began to undertake actions mandated under the Washington Consensus.

In 1986, President Miguel de la Madrid signed the General Agreement on Tariffs and Trade (GATT), opening Mexico to international trade and foreign investment. He also moved to cut social spending, reduce the size of the state, and balance the federal budget, initiatives that only accelerated under his successor, Carlos Salinas de Gortari. Under President Salinas virtually all state-owned firms were privatized and although it was not possible politically to sell the state oil monopoly PEMEX, he did successfully take on agrarian reform, phasing out agricultural subsidies and privatizing the once sacrosanct lands of peasants.

Perhaps the proudest accomplishment of the Mexican Revolution of 1910 was its expansive redistribution of farmland to peasants, especially during the 1930s. Large, private estates were confiscated by the government, broken into small communal plots, and given to local peasant cooperatives (Hart 1989). The redistributed lands, known as ejidos, were owned collectively by local communities. Although heritable rights of cultivation were granted to specific families in the ejido, who would then pass them on to their children, families were not allowed to rent, mortgage, or sell ejido lands (Markiewicz 1980). Until the early 1990s, half of all arable land in Mexico was located in the ejido sector. In 1992, however, President Salinas pushed through constitutional changes to permit the sale and rental of ejidos, initiating a sudden round of land privatization and property consolidation.

President Salinas then sought to institutionalize his various reforms by linking them to a treaty with Mexico’s powerful northern neighbor, and in 1994 Mexico joined the United States and Canada in the North American Free Trade Agreement. Under NAFTA, protective tariffs, quotas, and subsidies for agricultural goods were removed in phases, with the first set taking hold in 1998.

2. Poland

After World War II, the newly established communist regime of Poland tried to revitalize the post-war economy by replacing the market with a system of state ownership and central planning. As in Mexico, this state-led model was quite successful in the beginning, but by the end of the 1960s at latest, Poland had reached a point of economic stagnation (Balcerowicz 1995: 290). Beginning in 1970, the new Gierek team tried to overcome the structural problems by opening the economy somewhat and implementing an import-led policy (Poznanski 1996: 3 ff.). Massive imports of Western financial capital, consumer goods, and technology resulted in a period of rapid growth. But this, too, came to an early end. By the end of the 1970s, the system of borrowed prosperity collapsed like a house of cards. External debts had risen to alarming levels and forced a drastic decline in wages, consumption levels, and standard of living (Lipton/Sachs 1990: 103 f.). From 1979–82 Poland experienced a period of deep economic crisis.

During the 1980s two partial economic reforms (1981–82, 1987–88) were launched (Balcerowicz 1995: 292) and lead to a modest recovery (Lipton/Sachs 1990: 104). But already in 1985, new signs once again suggested that the country suffered not from occasional troubles that could be healed by soft reform programs, but from fundamental structural problems that required more drastic action (Poznanski 1996: 83 f.). As a result, in 1988 and 1989 the economy fell into a downward spiral of hyperinflation and collapse (Lane 1992: 10; Lipton/Sachs 1990: 109 f.). In the wake of these developments the communist Party experienced a crushing defeat in elections during 1989 and in November of that year a new Solidarity-led government took office.

The first task of the new government was to launch a radical economic reform designed by the economist and new minister of finance Leszek Balcerowicz and a team of international experts (the “Balcerowicz Plan”). They prescribed what has been aptly labeled as “shock therapy.” The new team essentially reset the economic clock to zero on January 1^st 1990, introducing a series of radical measures, such as full trade liberalization, massive privatization and the elimination of price controls. At the same time wages were deliberately kept low (Poznanski 1996: 169 f.). The net effect was a drastic decrease in production, real income and employment, yielding a severe economic recession that lasted much longer than anyone had expected. Notable signs of recovery did not begin to appear until the middle of the 1990s (Poznanski 1996: 170).

Although the economic shock and subsequent recession hit the majority of Poles hard, it turned out to be especially difficult on farmers and others employed in agriculture (Borzutzky/Kranidis 2005: 637). On the one hand, the hardships were directly attributable to the reforms themselves, such as the abrupt reduction of subsidies. On the other hand, economic turmoil also stemmed from Poland’s potential accession into the European Union and its necessary compliance with EU agricultural norms. Already by January 1991 the Association Agreement with the EU was signed, and it came into full force in 1994 (Borzutzky/Kranidis 2005: 633).

In 1997, Polish officials opened negotiations with EU authorities about the conditions of membership. The process of negotiation was accompanied by EU support programs such as Phare and SAPARD (Borzutzky/Kranidis 2005: 645), which could cushion some but not all of the dislocations associated with market liberalization as the share of employment in agriculture fell from 27 percent in 1993 to 19 percent in 2000 (Borzutzky/Kranidis 2005: 645). On May 1^st 2004 Poland, finally, became a member of the European Union, with only a few provisional restrictions to handicap its full integration into the common market. For many, Poland is regarded as a success story among transitory societies achieving fast and notable GDP growth (Keane/Prasad 2001). Since the early 1990s, however, this apparent success has also been accompanied by a considerable increase in income inequality.

II. Theoretical approaches to link structural change and international migration

International migration stems from choices made by people located within social contexts that vary across time and space. The context for any decision is defined by household and community circumstances, which are, in turn, affected by broader trends in the national and international political economy. In other words: the link between structural economic change and the rise in international migration can basically be conceived as a typical macro-micro-macro-transition (Coleman 1990: 8).

What are the more specific mechanisms within this general idea? To begin with, neoclassical economics has suggested the human capital model to be a core explanation at the micro level. It assumes that individual rational actors seek to maximize lifetime earnings and therefore invest in migration if the expected gains exceed the costs (Sjaastad 1962; Speare 1971). The strength of this model lies in the fact that it delivers an explicit and clear decision rule that allows one to derive precise hypotheses (about the age-selectivity of migration, for example). More importantly in our context here, it also allows us to predict theoretically the migratory consequences of structural economic change, as individual earning expectations are directly related to macro-economic conditions. In particular, the likelihood of migration should co-vary positively with the size of the wage gap between the sending and receiving countries. A prominent corollary is that migrants will be more positively selected with respect to human capital the higher the income inequality in the receiving country and the lower the inequality in the sending country (Borjas 1994).

However, the overt or hidden assumptions of the microeconomic have been subject to much critique (Massey et al. 1998) and more elaborated versions of the general cost-benefit approach have been suggested. Proponents of the theory of subjective expected utility (SEU) emphasize that additional, non-monetary utility terms may enter into the migration decision and that the calculation is often based on subjective rather than objective parameters (DeJong/Fawcett 1981). The SEU model allows one to account for the effect of other, non-economic factors that may systematically influence individual decision parameters. Most importantly, social capital theory can be married to the neoclassical cost-benefit-framework, with social ties to migrants operating to raise the benefits and lower the costs and risks of migration (Massey/Espinosa 1997). In this way, migration might become self-perpetuating over time, because each act of migration creates additional social capital that makes the migration of related others more likely (see Massey 1990), yielding a powerful feedback-loop well-known as “chain-migration” (MacDonald/MacDonald 1964; Massey/Zenteno 1999).

The new economics of labor migration has also challenged basic neoclassical assumptions (Stark 1991). It argues, first, that migration decisions are not made by single actors but collectively by larger units such as families or households. Second, it argues that these units seek not only to maximize expected income but also to minimize risks by diversifying the allocation of resources. This is seen as a means to overcome failures in local markets, such as futures, capital, and insurance. In addition to the labor market, therefore, other key markets become important in determining migration. Given that market failures are typical in developing and transforming societies, this kind of arguing delivers further theoretical justification for expecting a link between structural change and increased migration.

A further line of reasoning modifies models of migrant decision-making by tackling the problem of “hyper-rationality”, an implicit principle of all the foregoing theories. Building on Simon’s (1957) notion of “bounded rationality”, Kalter (1997) attempts to derive more realistic assumptions about the information processing capacity of actors. Given that information always comes at a cost, human calculations as a rule involve only a limited set of goals whose significance is driven by framing processes. In thinking about migration, for example, the set of potential destinations is usually restricted and often people do not seriously consider moving at all (Speare 1971: 130). Only under very specific conditions people switch frames, overcome the enormous hurdle of inertia, alter habitual behaviors, and put, as a result, migration on the decision-making agenda and engage in difficult cost-benefit considerations. Bounded rationality, thus, helps to explain why modest changes in contextual parameters are often not enough to trigger migration, even in the presence of substantial wage differentials, and why migration more commonly tends to begin only after “shock-like” events.

At the macro level, historical-structural theorists give supplementary arguments that can easily be accommodated with the different core models of micro-level decision-making sketched so far. They argue that societal transformations associated with the expansion of markets within and between nations change the context for decision-making by consolidating the ownership of land, capitalizing production, and monetizing exchanges, thereby creating new demands for cash that people seek to attain through migration (Portes/Walton 1981; Sassen 1988). Material networks of transportation and communication also expand in the course of market expansion to reduce the costs of expansion along certain pathways, thereby linking certain locations in core economies to peripheral zones of production and consumption (Sassen 1991).

III. Data Collection in Mexico and Poland

Among all areas of population data, the weakest is international migration. Unlike birth and death, migration is not a biological fact. It is a social fact expressed whenever a person crosses a meaningful social, legal, or geographic boundary. As a result, meaningful boundary-crossing is not an objective fact, but a subjective inference made by government officials. In addition to being hard to define, migration is also a repeatable event, unlike birth and death which happen once and only once in the course of a lifetime. For these reasons, in virtually all countries migration statistics are much less developed than statistics on birth or death. In the case of international migration, problems of measurement are compounded because much of the movement is clandestine and not subject to observation by state authorities.

In most countries, therefore, reliable statistics on the entry and exit of foreigners are scarce. Even when they exist, they are usually limited in the amount of detail they provide about the characteristics and behavior of immigrants. For these reasons, researchers have moved away from official data sources or standard census and have developed dedicated surveys to compile their own data on international migration, using a blend of different methods to undertake intensive studies of particular migrant-sending communities. Two such efforts are the Mexican Migration Project and the Polish Migration Project. New data gathered by these projects will be the basis for analyzing recent trends in international migration and each project is briefly described in this section.

1. Veracruz sample of the Mexican Migration Project (MMP)

Mexican data come from the Mexican Migration Project (MMP), which, since 1987, has annually fielded representative surveys in specific migrant-sending communities located throughout the nation (see Durand/Massey 2004). Although the project initially focused on communities situated in west-central Mexico, historically the heartland for migration to the United States (see Durand 1998), in recent years project investigators have made special efforts to expand data collection to newer sending regions that have only recently been incorporated into Mexico’s migratory outflow (Massey/Capoferro 2004). One such region is the state of Veracruz, and our analysis of MMP data focuses on surveys conducted in seven of its communities during 2003–2004.

Veracruz is a costal state located on the Gulf of Mexico that houses the country’s largest and most important port, the City of Veracruz. Although the narrow costal plane on which the city sits is tropical, the Sierra Madre Mountains rise rapidly as one moves inland and the topography changes. The communities, surveyed by the MMP, are located in middle altitudes near the state capital of Jalapa. Historically, this area has been devoted to the production of coffee and other crops on small communal plots. In addition to surveying a neighborhood in the city of Jalapa, MMP investigators also sampled six agrarian communities located in adjacent municipalities, ranging in size from 1 200 to 4 500 inhabitants. As Table 1 shows, all of these smaller communities specialized in agriculture, with the percentage of workers engaged in farming ranging from 62 percent to 75 percent, compared to just 5 percent in Jalapa. In keeping with their agrarian roots, these communities also display low levels of education, with the percentage of adults having less than 6 years of schooling ranging from 44 percent to 56 percent, compared with 19 percent in Jalapa.

Table 1.

Communities surveyed in Mexico and Poland

Country and Community	Population Size	Percent in Agriculture	Percent Low Educationa	Date of Survey	Size of Survey	Refusal Rate
Mexico
Jalapa	373 000	5,4	19,0	2004	201	7,4
Coyote	2 000	71,1	49,3	2004	101	3,8
Montes	1 600	61,9	44,0	2003	193	0,0
Lomas	1 200	73,2	51,6	2004	90	3,3
Mesones	2 800	75,3	55,0	2004	98	14,0
Alteño	4 500	75,2	56,0	2004	150	9,6
Santa Rita	1 700	71,1	49,3	2003	105	13,2
Poland
Jaraczewo	8 300	37,3	8,5	2005/6	160b	25,1c
Poznan	570 100	1,3	4,7	2005/6	159b	51,6c
Pawlów	15 100	27,9	32,0	2005/6	205b	16,9c
Kielce	210 000	1,3	4,4	2005/6	203b	26,6c

^aMexico: under 6 years of schooling; Poland: at most compulsory education.

^bFull interviews only (without screening).

^cRefusal rates refer to the first phase of the sampling (without screening).

The target sample size was 200 households in the city of Jalapa and between 100 and 150 in the surrounding agrarian communities. As indicated by the table, 201 households were surveyed in the city and 90–150 were surveyed in the other locations, yielding a total Mexican sample size of 938 households. Sampling frames were constructed by carrying out a house-to-house census within each community or neighborhood prior to the start of fieldwork. Refusal rates were generally modest, ranging from zero percent in the village of Montes to 14 percent in the town of Mesones. The interviewing was concentrated in the months of December and January, which are best to locate and interview seasonal U.S. migrants because most return to spend the Christmas holidays with their families.

All Mexican respondents were interviewed using ethnosurvey methods, a blend of ethnographic and survey techniques (see Massey 1987, 1999). The basic idea of an ethnosurvey is that these procedures complement one another, and that when properly combined, one’s weaknesses become the other’s strengths, yielding a body of data with greater reliability and more internal validity than would be possible to achieve using either method alone (Massey/Capoferro 2004). Information is solicited using a semi-structured survey instrument that yields a standard body of data but which allows interviewers flexibility to decide when, where and how to pose sensitive questions.

Within each household, interviewers gathered basic information about the social, economic and demographic characteristics of the head, the spouse, the head’s children, and other household members. After determining which members had ever been to the United States, they gathered basic data about the first and last U.S. trips, including date, duration, legal status, destination, occupation, and wage. Each household head was also administered a complete life history that included a yearly history of migration and border crossing. Systematic comparisons between data from the MMP and representative surveys have shown that the ethnosurvey yields a remarkably accurate and reliable profile of international migrants and their characteristics (Massey/Zenteno 2000).

2. The Polish Migration Project

The Polish Migration Project (PMP) follows the general ideas of the MMP gathering the same basic information. In 2005 it started to collect comparable data in four Polish communities. The selection of communities was purposive based on the following considerations: First, the connection with Germany through migration should be known to be relatively strong. Second, regions near to (but not directly at) the German- Polish border should be contrasted to regions that are further away. Third, rural communities should be contrasted to urban ones and fourth, the peculiarities of migration of Ethnic Germans should be avoided by excluding the respective regions. Based on these criteria at first two cities were chosen: Poznan, an industrial city in the West of Poland with 570 000 inhabitants, known for rather close connections to Berlin which is 270 km away, and Kielce, a city with 210 000 inhabitants, located in the South-Eastern part of central Poland. Despite its relative distance to Germany (475 km), the area around Kielce accounted for the highest numbers of seasonal workers in many recent years. The two rural contexts chosen were located nearby: Jaraczewo is a community located 70 km south of Poznan, consisting of 22 smaller villages with altogether 8 300 inhabitants, and Pawlów is a community located 40 km east of Kielce, consisting of 33 smaller villages with 15 100 inhabitants in total. As Table 1 shows in both, Jaraczewo and Pawlów, still a large share of the employed population is working in agriculture at the time of the survey. However, proportions do not reach the levels observed in of the Veracruz villages. Mean educational attainment is especially low in Pawlów where 32 percent of the adult population has at most compulsory education.

Although the PMP tried to comply with the design of the MMP, as far as possible, some deviations were necessary to cope with the peculiarities of Poland and context-specific restrictions. In contrast to the MMP, the survey was conducted in an almost fully standardized way. The target population was not household heads but all residents of the respective communities aged 18 to 65. Further, in order to increase efficiency, a stratified sample of migrant and non-migrants was constructed in each of the four communities. The criterion for being a ‘migrant’ is ever having worked in Germany since 1980.

To derive at the stratified samples the fieldwork was divided into two phases. In the first phase, cases were added from a randomized list of addresses until a number of about 100 interviews with non-migrants had been reached. In three communities the list consisted of a random sample of residents stemming from the electronic population register system PESEL. In Jaraczewo a list of households gathered in a random-walk procedure was used and the selection of the respondent was made with the last-birthday method. In the second phase following the lists further down, only short screening questions were asked and a full interview was only conducted if the target person was a migrant. This procedure was followed until either 100 migrants in total were reached or an upper limit of screening interviews had been finished. All in all, 727 full interviews were conducted which provide the basis for the following analyses. The marginal distributions of migrants in the first phase can be used to construct design weights in order to simulate a simple random sample if necessary.

IV. Migration prevalence in Mexico (Veracruz) and Poland

This section describes the rise in migration that accompanied the structural changes sketched above (I). After reporting briefly what is known from other sources we rely on the data of the MMP and the PMP to sketch recent trends in detail. In each context we look at labor migration to one adjacent nation, the most important in terms of migration numbers: to the United States in the case of Mexico (1.) and to Germany in the case of Poland (2.).

1. Mexico, Veracruz

Prior to the neoliberal era, Veracruz had never sent very many migrants to the United States. Until the 1980s, most migrants originated in the states of west-central Mexico, which had been sending immigrants to the United States since the early 1900s (Cardoso 1980; Massey et al. 1987, 2002). Before 1990, less than one percent of all Mexico-U.S. migrants originated in Veracruz (Durand/Massey 2003). However, the succession of economic shocks associated with the abandonment of import substitution industrialization dramatically changed circumstances in Veracruz, particularly the rural sector.

That the shift to neoliberalism went along with a rapid rise in out-migration to the U.S. is clearly evident in the Mexican Migration Project’s Veracruz surveys. Following the approach of Massey, Goldring, and Durand (1994) we used the date of each household member’s first U.S. trip to compute migration prevalence ratios for different years. The denominator of the prevalence ratio is the number of persons aged 15 or more during the year in question and the numerator is the number of persons aged 15 or more who had ever been to the United States by that year. For people in rural areas of Mexico, 15 is a common age of school-leaving and labor force entry. When multiplied by 100, the ratio gives the percentage of people from the community with U.S. migratory experience in any given year. It is computed using data on household members only, and indicates how widely migration has diffused throughout the community population.

Figure 1 plots migration prevalence ratios for males in each of the Veracruz communities from 1980 through 2004 along with key dates in the evolution of the Mexican political economy. As is visible, prior to Mexico’s entry into GATT, no community participated very intensively in migration to the United States. Up through 1986, no more than five percent of all males aged 15 or more had ever been to the United States. Around this date, the agrarian community of Santa Rita experienced a sudden and rather dramatic increase in migratory prevalence, with the male prevalence ratio doubling from 5 percent to 10 percent by 1988.

Figure 1.

Male migration prevalence in Veracruz communities

As part of his package of economic reforms, President Salinas privatized Mexico’s ejidos in 1992. After this date, out-migration from Santa Rita accelerated and the male prevalence ratio began a sustained increase that lasted through the year 2000, by which time around 25 percent of the town’s adult males had begun migrating to the United States. In other words, during the decade and a half from 1985 to 2000 the prevalence of migration among Santa Rita’s males increased by a factor of five, going from 5 percent to 25 percent in two distinct bursts, one following Mexico’s entry into GATT in 1986 and the other following the privatization of ejidos in 1992.

The privatization of the ejidos was also associated with the initiation of U.S. migration from other agrarian communities in Veracruz. After remaining roughly at, or below, the 5 percent from 1980 through 1991, after 1992 male prevalence ratio began to rise sharply in Lomas and Monte. In both cases, male migratory prevalence tripled, in seven years rising to around 15 percent by 1998. Although we see little evidence of a migratory response associated with Mexico’s formal entry in NAFTA in 1994, the first round of tariff removals under the treaty clearly raised the pace of out-migration and extended it to all of the other agrarian communities as well as to the city of Jalapa. Beginning around 1998, U.S. migration from the villages of Alteño, Lomas, Mesones and Coyote surged, in several cases quite dramatically. Prior to this date the migratory prevalence ratio among males had languished at around three percent. Afterward, it rose markedly in Lomas and Alteño, reaching 14 percent in the latter and 16 percent in the former.

In the communities of Mesones and Coyote, however, the surge in out-migration was nothing short of spectacular, rising from around 3 percent in 1996 to around 30 percent by 2003. In other words, in the short period between 1996 and 2003, at least a quarter all males aged 15 or more in these two communities began migrating to the United States. Early in the new millennium, communities throughout the state of Veracruz were participating heavily in regular out-migration to the United States. Even the urban center of Jalapa-home to the state government and a major university town had begun significant participation. By the year 2000, the male prevalence ratio in Jalapa for the first time exceeded 5 percent.

The foregoing analysis strongly suggests that the structural transformation of the Mexican political economy after 1986 was associated with a pronounced migratory response in the state of Veracruz, in a very short time transforming agrarian communities throughout the region from migratory backwaters that were only marginally connected to the transnational labor market into significant migrant-sending communities with high levels of male migratory participation, in many cases reaching levels that rival those in traditional migrant-sending areas in the states of west central Mexico. Very clearly, the social context within which Veracruzanos weighed the options of staying or leaving changed quite radically under neoliberalism.

2. Poland

Migration from Poland to German has a long history. It is estimated that from the late 19^th century until World War I about 750 000 Poles facing increasing rural poverty migrated into the German industrial metropolises (Pallaske 2001a: 10). World War I stopped this process and between the wars emigration of Poles to Germany occurred at rather low levels. The same holds true for the period from 1945 to 1980.¹ The 1980s began with a notable increase of emigration during Poland’s social and politic crisis, but this came to an abrupt end with the imposition of martial law. Owing to gradual liberalization late in the communist era, especially with regard to passport policies, numbers of emigrants began to rise again shortly thereafter, slightly at first, then growing more and more acute (Korcelli 1992: 293). During this time, tourist visas became the main gate for illegal mass emigration (Okolski 1998: 11). This process came to a climax in 1988. Soon thereafter, changing asylum policies and fixed quotas for late re-settlers in Germany led to a sharp decrease of permanent migration and numbers returned “back to normal” in the early 1990s (Pallaske 2001b: 125).

In the wake of these developments, temporary labor migration became the prevailing type of migration. It was facilitated by the growing demand for short-time work in Germany, especially for seasonal labor in agriculture and construction. During the 1990s and continuing into the new millennium, official numbers of temporary work grew steadily and for large portions of this period illegal forms of employment are even estimated to have exceeded the legal component (Okólski/Stola 1999: 19).

How has migration affected the selected Polish communities in the course of the above sketched structural economic changes? In order to answer this question we turned to the same method as in the Mexican case, introducing only some minor changes. Specifically, we computed migration prevalence ratios, but now for persons aged 18 or more. 18 is a more common age of school-leaving and labor force entry in Poland. The estimates rely only on respondents and not on other household members,² and include both males and females. To account for the stratification of the sample design, weights are used to derive correct estimates based on the population at risk.

Figure 2 shows the development of cumulated migration experience to Germany in the Polish communities over time as well as several milestones in the historical process of economic change. As is visible, two distinct patterns arise, one in the urban sector and another in the rural sector. The two cities saw notable emigration already by the 1980s. In 1989 more than 12 percent of the adult population in Poznan, and about 9 percent in Kielce, had some working experience in Germany. A closer look at the data reveals that about two thirds of all trips went to the former GDR. This form of ‘socialist brother help’ is a notable peculiarity of the Polish-German migration system (Miera 2007: 91 ff.).

Figure 2.

Migration prevalence in Polish communities

In the two rural communities the take-off point for migration was clearly 1989. During the period of economic collapse, shock therapy and subsequent recession migration prevalence rose considerably in Jaraczewo and Pawlów. During this time figures increased sharply also in Kielce, while in Poznan the gradient of migration prevalence began to decline. These trends continued in the first years after recovery. Thus, in the course of the structural changes necessary to move the economy toward capitalism, migration prevalence rose considerably in Kielce, Jaraczewo and was most pronounced in Pawlów. When Poland officially became a EU candidate country in 1997 about 17 percent of the adult population in Kielce, 15 percent in Poznan, 13 percent in Pawlów and 7 percent in Jaraczewo had worked in Germany at least once.

As Figure 2 illustrates quite clearly, the last phase in the economic transition towards EU accession in 2004 witnessed very different developments in urban and rural settings. While during this period almost no increase is apparent in Poznan and Kielce, the inclination to make a first trip from Jaraczewo or Pawlów rose even more sharply than before. In 2004 both rural communities had caught up with and even outpaced their urban neighbors in terms of migration prevalence. By the time that Poland became a member of the EU the prevalence ratio had risen to 25 percent in Pawlów and 15 percent in Jaracezewo, while the figure were 18 percent in Kielce and 17 percent in Poznan.

V. Analysis of first migration

1. Method and variables

The prevalence ratios in Figures 1 and 2 show that structural economic change was accompanied by increasing migration from Mexican and Polish communities over time, but they do not get at the specific mechanisms by which migration was initiated and why it expanded so rapidly. In order to address this issue we undertook a multivariate analysis of the determinants of first migration. Drawing upon data compiled by the MMP’s and PMP’s quantitative life histories, we followed each household head year-by-year from entry into the labor force up to the survey date and predicted a dichotomous outcome that equaled 1 if the person took a first trip (to the United States and Germany, respectively) in that year and 0 otherwise, excluding all years after the first trip from the analysis. Migration in year t is then predicted using logistic regression from a set of variables defined in year t – 1, yielding a discrete time event history analysis of first migration.

It is necessary to note that the analyses in this section, like Figures 1 and 2 above, rely only on respondents still living in the sending country. This might lead to a certain bias because permanent, not-(yet-)returned migrants do not enter the risk sets. In the Mexican case, we know that rising border enforcement since 1986 has lengthened trips and reduced rates of return migration, especially after 1993 (Massey et al. 2002). However, because migration from Veracruz is so recent it is composed mainly of older males, usually household heads, rather than entire families, who remain behind to report dates of departure and other characteristics, thus minimizing the bias. In the Polish case the bias will only be small: Both numbers from official Polish Statistics as also the PMP survey data (e. g. by looking at respondents’ close family members who moved abroad) indicate that the proportion of those who have left the four communities since 1980 to live permanently in Germany is extremely low.

The independent variables used in the analysis of out-migration are summarized in Tables 2a and 2b. Dummy variables representing each of the communities are used to control for community-level fixed effects. We also control for the respondent’s demographic background by measuring age and gender. More interesting theoretically are variables considered important by neoclassical economics. Our leading measure of human capital is education, which we measure categorically by dividing respondents into three groups: low, medium, and high education. In the Mexican data the low education category included those with under nine years of schooling, the medium category included those with 9–15 years of schooling and the high category include those with 16 or more years of schooling (i. e. college graduates). In the Polish data high education means a university degree, medium education means that the respondent has finished upper secondary school, while the category low comprises all other types with less years of schooling (primary school, vocational school).

Table 2.

a: Independent variables used to predict the likelihood of taking a first trip to the United States
Independent Variable	Mean/percent	Standard Deviation	Minimum	Maximum
Community
Jalapa	0,223
Coyote	0,099
Montes	0,204
Lomas	0,107
Mesones	0,101
Alteño	0,165
Santa Rita	0,112
Demographic Status
Female	0,158
Age	27,36	18,17	15	93
Educational Status
Low	0,707
Medium	0,213
High	0,080
Occupational Status
Not Employed	0,394
Farm Worker	0,412
Manual Worker	0,124
Non-manual Worker	0,070
Has Social Security	0,124
Social Capital
Spouse has U.S. Experience	0,003
Other Family U.S. Experience	0,228
Male Prevalence Ratio	7,019	6,828	1,0	21,5
U.S. Mexico Income Gap
Ratio of Per Capita GDP	3,883	0,477	2,88	4,42
Period
Pre-GATT 1980–1986	0,656
Early Neoliberal 1987–1992	0,116
Late Neoliberal 1993–2004	0,228

Table 2b: Independent variables used to predict the likelihood of taking a first trip to Germany
Independent variable	Mean/ percent	Standard Deviation	Minimum	Maximum	Mean/ percent (design- weighted)
Community
Jaraczewo	0,210				0,217
Poznan	0,218				0,215
Pawlów	0,287				0,291
Kielce	0,285				0,277
Demographic
Female	0,467				0,522
Age	34,116	10,964	17	67	34,705
Educational Status
Low	0,455				0,441
Medium	0,351				0,366
High	0,194				0,193
Occupational Status
Not Employed	0,517				0,512
Farm Worker	0,064				0,071
Manual Worker	0,195				0,175
Non-Manual Worker	0,223				0,243
Has Social Security	0,363				0,362
Social Capital
Spouse has Germany experience	0,040				0,038
Other Family Germany experience	0,067				0,060
Male Prevalence Ratio	9,950	6,629	0	25	9,859
Germany-Poland Poland Income Gap
Ratio of Per Capita GDP	3,160	0,323	2,6	3,8	3,158
Period
Late Communism	0,274				0,278
Collapse, Shock, Recession	0,153				0,151
Early Transition, PreEU	0,172				0,169
EU Candidate	0,348				0,347
EU Member	0,054				0,055

We measured each respondent’s access to local employment using a dummy variable that equaled 1 if the person was employed in year t – 1 and 0 otherwise. Among those who were employed, we included a classification by occupational status, dividing the sample into three categories corresponding roughly to occupational skill. Farm workers held jobs in the agricultural sector, manual workers included operatives, skilled and unskilled laborers and personal service workers and non-manual workers included those in sales as well as those in managerial, technical, or professional categories.³ We also included an indicator of whether or not the respondent’s job included social security coverage, thereby indicating employment in the formal economy.

To indicate relative access to migration-specific social capital, we include a dummy variable for whether or not a respondent’s spouse had been to the United States by year t – 1, and another indicating whether or not any member of the respondent’s immediate family (parents, children, or siblings) had migrated by year t – 1. We also include the overall migratory prevalence ratio in year t – 1 as a measure of the overall level of migratory experience that has accumulated in the community as a whole.

In order, to measure the incentives for migration emanating from the receiving society, we computed the ratio of per capita GDP in the receiving society (the United States or Germany) to that in the sending society (Mexico or Poland). These data were obtained from the Penn World Tables Version 6.2, which measure GDP in U.S. dollars adjusted for purchasing power parity (see Heston et al. 2006). Thus our measure of the relative gain to be had from migration controls for differences in relative cost of living across countries.

According to the means reported in Table 2a, during the typical person year observed from 1980 to 2004 in the Mexican sample the average respondent was 27 years old and around 16 percent were female. Most fell into the low educational category, with 71 percent reporting less than nine years of education. Only 21 percent and 8 percent reported medium and high levels of education, respectively. Unemployment was rather high, with about 39 percent of respondents not holding a job during the average person year (most of the jobless, however, were females). The largest occupational category was farm worker, followed by manual and then non-manual occupations and around 12 percent held formal sector jobs that were covered by the Mexican social security system. On average, per capita GDP in the United States was around 3,9 times that in Mexico, with a range from 2,9 to 4,4.

In the Polish data (see Table 2b) the mean age over all person years underlying the analyses is 34, thus somewhat higher than in the Mexican case because the event histories begin at age 18 rather than 15. Given that the sampling in Poland was on all adult residents rather than household heads and the fact that Mexican culture is quite patriarchal, the share of females is much higher among Poles (47 percent). Over all person years 46 percent of the respondents fall into the category of low education, 35 percent into the medium one, and 19 percent into the high one. In the majority of person years (52 percent) under consideration, the respondent was not employed the year before. Among those who were employed, most belonged to the category of non-manual workers, followed by manual workers. 36 percent of all respondents had social security coverage. Although the means in the first column of Table 2b are not design-weighted, the weighted means deviate only slightly and are given in the last column. On average, per capita GDP in Germany was around 3,2 and ranged from a low of 2,6 to a high of 3,8 over the range of years considered.

2. Determinants of first migration from Veracruz to the U.S

Table 3 presents estimates of three discrete-time event history models. The two columns on the left-hand side of the table estimate a model for all person years. It includes cumulative contrasts to capture the transition between three periods: the first variable indicates the switch from the pre-GATT period (1980 through 1986) to the early neoliberal period (from 1987 up to the privatization of the ejidos in 1992), the second variable the step to the late neoliberal period afterward, which includes the joining of NAFTA in 1994 and the removal of agricultural tariffs in 1998. Research by del Rey Poveda (2007) suggests that the ejido privatization of 1992 was a major dividing line in the evolution of economic conditions in Veracruz. Coefficients associated with these period variables generally confirm the significance of the shift in migration probabilities associated with the shift to neoliberalism. Compared with the pre-GATT era, the odds of migration rose very slightly 28 percent during the early neoliberal period before 1987 (e^0.250 = 1,28) but then jumped by a factor of nearly two and a half during the later neoliberal period following the privatization of communal farmlands (e^0.878 = 2,41).

Table 3.

Discrete time event history analysis predicting first trip to the United States by household heads for person years lived since 1980

	All Years		Years from 1980 to 1992		Years from 1993 to 2004
Independent Variables	B	SE	B	SE	B	SE
Community
Jalapa	–	–	–	–	–	–
Coyote	1,126**	0,417	1,203	1,301	1,125**	0,445
Montes	1,201***	0,373	1,331	1,179	1,207**	0,397
Lomas	0,529	0,499	2,214+	1,206	0,137	0,586
Mesones	1,028**	0,414	1,264	1,289	1,004**	0,441
Alteño	0,536	0,408	1,053	1,247	0,501	0,434
Santa Rita	1,233**	0,398	2,848**	1,099	0,766+	0,457
Demographic Status
Female	−1,679*	0,733	–	–	−1,570**	0,744
Age	0,117*	0,052	0,107	0,136	0,144**	0,064
Age Squared	−0,003***	0,001	−0,003	0,002	−0,003***	0,001
Educational Status
Low	–	–	–	–	–	–
Medium	0,072	0,290	−0,155	0,671	0,106	0,324
High	−0,973	1,015	−1,643	3,216	−0,742	1,019
Occupational Status
Not Employed	–	–	–	–	–	–
Farm Worker	0,686	0,503	1,671	1,118	0,500	0,566
Manual Worker	0,710	0,506	1,860	1,175	0,462	0,561
Non-manual Worker	0,714	0,538	2,231+	1,253	0,392	0,594
Has Social Security	−0,100	0,263	0,048	0,568	−0,104	0,269
Social Capital
Spouse has U.S. Experience	1,058**	0,543	1,582	1,125	1,034+	0,625
Other Family U.S. Experience	0,305	0,294	0,643	0,647	0,253	0,331
Male Prevalence Ratio	0,046**	0,017	0,430	0,366	0,042*	0,018
U.S.-Mexico Income Gap
Ratio of Per Capita GDP (PPP $)	1,102	0,680	0,675	0,931	1,144	0,850
Period (cumulative contrasts)
≥ 1987 (Early Neoliberal)	0,250	0,607	–	–	–	–
≥ 1993 (Late Neoliberal)	0,878**	0,347	–	–	–	–
Intercept	−12,835***	2,421	−13,190***	3,514	−12,105***	1,894
Log Likelihood	−1751,11		−334,47	−1333,15
Chi-Squared	172,04***		48,43**	155,11***
Number of Person Years	23229		12179		11050

⁺p < 0,10;

^*p < 0,05;

^**p < 0,01;

^***p < 0,001.

In the next two sets of columns, we show the model estimated separately before and after the year 1992, when Mexico’s ejidos were privatized. The middle two columns show estimates for the period 1980 through 1992. These data reveal no significant individual-level determinants of out-migration to the United States and few differences between communities with respect to the propensity to migrate internationally. There is no evidence, for example, of significant socioeconomic selection and only the two communities of Santa Rita, the earliest responder to structural economic changes, and Lomas, another agricultural village, displayed significantly greater propensities of out-migration. The only other variable to attain even marginal significance was being a non-manual worker, suggesting that apart from fixed effects associated with these two communities the process of out-migration was largely random (however, the number of female migrants was too small to estimate a reliable effect of gender).

The right-hand columns of Table 3 show the same analysis for person years observed after 1992. Compared to the earlier model, out-migration becomes considerably more selective with respect to theoretically expected variables. At the individual level, the odds of leaving show the expected curvilinear effect with respect to age, rising to peak in the young labor force ages before falling again and as one would expect, the process of emigration is also highly selective of males. Economic theory also hypothesizes selectivity with respect to human capital, but neither education nor occupational skill appears to predict out-migration, either before or after 1992. Likewise, at the macro level, migration is not significantly predicted by the GDP ratio, though the effect size is somewhat larger than in the earlier period. In general, the model suggests the relevance of other motivations for migration besides the desire to maximize earnings, such as the need to overcome failures in capital, credit, or insurance markets, as predicted by the New Economics of Labor Migration.

Holding constant the effects of age and gender, we also observe strong effects of social capital. Having a spouse with prior U.S. experience increases the likelihood of out-migration substantially, though because of the relatively large standard error the effect is only significant at the 10 percent level. The effect of migratory prevalence is more precisely estimated and more highly significant. As expected, the greater the number of males aged 15 or more in their community who have ever been to the United States in year t – 1, the greater the odds of taking a first trip in year t.

Even after controlling for this important community-level variable, four of the six community indicators attain clear statistical significance, suggesting that the likelihood of emigration was raised for all residents of these communities irrespective of their personal characteristics. Thus, external shocks associated with Mexico’s shift from import substitution industrialization to neoliberalism appear to have brought about a wholesale transformation of the calculus of migrant decision-making in rural communities Veracruz. Before the privatization of the ejidos in 1992, out-migration was limited to one or perhaps two communities and apart from these fixed effects was largely random and unselective. Afterward, out-migration to the United States not only increased, but became selective with respect to age, gender, and social capital while spreading to a majority of the rural communities around Jalapa.

3. Determinants of first migration from Poland to Germany

The results of comparable models analyzing the risk of taking a first trip from the Polish communities to Germany are given in Table 4. The first two columns on the left hand side give coefficients and standard errors for a model that uses all person years available between 1980 and 2004 (n=10548). These estimates reveal that the risk of becoming a migrant in that period is much lower for females and shows the typical age-selectivity expected from human capital theory. While there is no selectivity with respect to education when looking over the total time span, occupational status in year t – 1 turns out to be an important determinant. Being employed, especially in non-manual work, clearly reduces the risk of becoming a migrant. Other than human capital, social capital is also an important factor fostering out-migration. Access to migratory experience through close family ties and through migratory prevalence in the community both significantly raise the odds of taking a first trip.

Table 4.

Discrete time event history analysis predicting first trip to Germany by Polish residents for person years lived since 1980.

	All Years		Years from 1980 to 1992		Years from 1993 to 2004
Independent Variables	B	SE	B	SE	B	SE
Community
Jaraczewo	–	–
Poznan	−0,157	0,258	1,258*	0,510	−0,636+	0,355
Pawlów	0,168	0,200	−0,070	0,453	0,409	0,271
Kielce	0,237	0,244	1,215**	0,439	0,242	0,354
Demographic
Female	−0,618***	0,129	−0,991***	0,245	−0,416**	0,158
Age	0,213***	0,042	0,183*	0,092	0,183***	0,050
Age Squared	−0,003***	0,001	−0,003+	0,001	−0,003***	0,001
Educational Status
Low	–	–	–
Medium	−0,128	0,150	0,078	0,263	−0,229	0,184
High	0,120	0,193	0,553+	0,288	−0,156	0,266
Occupational Status
Not Employed	–	–	–
Farm Worker	−0,535+	0,289	−1,196	1,033	−0,438	0,309
Manual Worker	−0,339	0,335	−0,015	0,532	−0,411	0,435
Non-Manual Worker	−1,203***	0,331	−0,403	0,498	−1,476**	0,453
Has Social Security	0,244	0,314	−0,071	0,485	0,386	0,415
Social Capital
Spouse has Germany experience	0,114	0,325	−0,881	1,029	0,264	0,348
Other Family Germany experience	0,463*	0,214	0,143	0,490	0,595*	0,241
Prevalence Ratio	0,053*	0,021	−0,000	0,045	0,015	0,031
Germany-Poland Income Gap
Ratio of Per Capita GDP (PPP Dollars)	−0,239	0,402	1,187*	0,492	−1192**	0,458
Period (cumulative contrasts)
≥ 1989 (Collapse, Shock, Recession)	0,804**	0,289
≥ 1993 (Early Transition, PreEU)	−0,189	0,225
≥ 1997 (EU Candidate)	0,084	0,283
≥ 2004 (EU member)	0,772***	0,241
Intercept	−6,432***	1,436	−11,183***	2,233	−1,770	1,726
Log-Likelihood	−1275,1		−460,7		−788,1
Chi-Squared	212,4		90,3		143,8
Number of Person Years	10548		5198		5350

⁺p < 0,10;

^*p < 0,05;

^**p < 0,01;

^***p < 0,001.

The dynamics of emigration from Poland to Germany sketched in Figure 2, however, can not be explained by compositional effects related to human capital and social capital arguments alone. Strong period effects remain even after controlling for these variables. Hazard rates rose strongly and significantly with the collapse in 1989 and during the subsequent shock therapy. The period of recovery (1993–1996) led only to a modest decline and the baseline risk stayed almost the same in the phase of being an EU-candidate country (1997–2003). In 2004, however, the odds of emigration rose sharply once again as Poland joined the EU. Compared with the earlier period, the hazard rate more than doubled (e^0,772 = 2,2), the increase being significant even given the relative low number of person years available under EU membership. Note that this holds true controlling for the general income gap between Poland and Germany, which turns out to be of no importance in this model. It is also worth noting, that a continuous time variable, modeling more gradual change would not add significantly to the model and the cumulative contrasts would stay nearly unchanged when including it. Thus the increase in migration propensities appears to have happened indeed ‘shock-like’ rather than little by little in the course of the transformation process.

The process of structural economic change captured by the period variables not only had an additive effect on the occurrence of migration, but also fundamentally changed the relevance of other underlying mechanisms. This becomes obvious when contrasting equation estimates from the early period 1980–1992 to those from the later period 1993 onwards (see next two sets of columns in Table 4). As already indicated in Figure 2 the high risk areas switched from the urban to the rural contexts: controlling for the other variables residents of Poznan experienced greater risks of migration than those in the reference community of Jaraczewo during the first thirteen years under consideration, but significantly lower risks in the last twelve. Likewise, for Kielce the sign has changed (significantly) from being positive to negative.

Contrasting the two models also shows that from the first half to the second half of the transition period females became more involved into migration and that migration changed from being positively selective to being slightly negatively selective with respect to education and differences for both the gender coefficient and the high education coefficient between the two modes are significant (p < 0,05). When comparing the coefficients of occupational status indicators one also finds some interesting changes, but owing to the high standard errors one should be cautious not to over-interpret these. In addition to that, there seems to be a clear signal that social capital has become more important over time. Although the differences are not significant at 5 percent-level, the coefficients of all three indicators have moved in a positive direction from the period 1980–1992 to the period 1993–2004. Finally, there is an interesting and important shift in the influence the income gap. The coefficient is significantly positive until 1992, but significantly negative from 1993 onwards. This reflects the fact that the income was widening in the first period and narrowing in the second. Yet all of the time migration probabilities were steadily rising. This can, thus, in no way be attributed to the development of the general income gap.

As the analyses in Table 4 clearly indicate, the shift from Poland of being a communist country of central planning to a market economy integrated as a full member in the European Union has not only abruptly involved more and more residents into migration, but also fundamentally changed the selectivity of those who elected to become migrants.

VI. Conclusion: similarities and differences

Although Mexico and Poland occupied very different positions in the world’s post-war political economy, their trajectories of economic development and international migration display many salient parallels. In this analysis, we systematically compared both cases to discern similarities and differences in the basic mechanisms by which international migration is initiated, taking advantage of reliable longitudinal data gathered by the Mexican Migration Project and the Polish Migration Project, both of which provide rich labor and migration histories and measurement of key independent variables specified by prominent theories of migration. Above all, the Mexican and Polish data are, apart from minor country-specific adaptations, almost perfectly comparable. As a result, we were able to run exactly the same analyses for the Mexican-U.S. case and the Polish-German case and thereby deliver a unique contribution to empirical research on migration.

Our comparison reveals that in both settings period effects associated with the onset of structural adjustment remain significant in predicting out-migration, even when controlling for a set of standard variables usually associated with migration propensity. Thus shocks connected to the process of transition to the market appear to have an independent exogenous effect on the taking of working trips to the United States and Germany, apart from the standard mechanisms put forth by theorists to account for migration dynamics. This conclusion seems all the more justified as migratory prevalence in the community and the size of the binational income gap are already controlled, thus holding constant two key determinants that also vary over time. As might be expected, however, the residual period effects are stronger in Poland, which of course shifted from a full-blown socialist to a capitalist economy, whereas Mexico only went from a state-led to a private-led model of capitalist development.

The two contexts also exhibit a variety of further salient differences and the effect of structural economic change seems to play out somewhat different ways. First of all, the structural shift was associated with very different patterns of macroeconomic performance in each context. In Poland the period prior to 1992 was associated with a monotonically rising income gap with respect to Germany, while afterward the income gap steadily fell. As a result, the coefficient associated with the income gap was positive in the pre-adjustment period and negative afterward. In contrast, Mexico’s income with respect to the United States fluctuated before 1992, with the gap falling from 1980 to 1982, then rising from 1982 to 1989 before falling again until 1992, after which the income gap steadily increased. In other words, whereas structural adjustment in Poland was associated with a declining income gap that in Mexico was associated with a growing gap

Controlling for the GDP ratio and breaking the sample into two periods essentially eliminated the effect of migration prevalence in the before-and-after regressions for Poland. In Mexico, however, the more independent trajectories of prevalence and income yielded a significant effect of social capital after the onset of structural adjustment. In Mexico emigration grew more selective with respect to migratory prevalence in the community, whereas in Poland it did not.

Although emigration was not selective with respect to education in either country, before or after structural adjustment, in Poland emigration did grow more selective with respect to occupational skill in the wake of the market transition. Whereas there were no differences in the odds of out-migration among Poles before 1992, afterward nonmanual workers displayed markedly lower likelihoods of departure. Likewise, having a family member with German migratory experience became an important predictor of migration after adjustment, whereas it had no effect before. However, unlike the Mexican case, out-migration was selective with respect to age and gender in both periods.

To a greater extent than in Mexico, the effect of structural adjustment seemed to operate exogenously in Poland, yielding a massive shock that significantly raised the odds of out-migration. It also changed the characteristics of the migrants, involving females and lower educated much stronger in more recent years. Following the general reasoning of Borjas (1994), this shift in selectivity could have something to do with increasing income inequality Poland after the early 1990s. In Mexico structural adjustment promoted migration increasing the susceptibility of certain kinds of people to international movement, raising the odds of migration among working age men with access to social capital and increasing the rate of out-migration from certain specific rural communities. Whatever the differences in detail, however, our analysis clearly shows that structural transformations associated with the shift toward market development tend to produce out-migrants, at least in the short term. Rather than stemming from a lack of economic development, international migration often follows directly from development itself.