Income differentials and global migration in the contemporary world-economy

Abstract

Spanish

French

Shifting the unit of analysis from the nation-state to the world as a whole fundamentally changes our understanding of migration. Elsewhere, the authors have argued that ascriptive criteria centered on national identity and citizenship have long served as a fundamental basis of inequality in the world. Here, they develop a model that seeks to identify the main forces driving migration across the world-economy. They test this model by drawing on an original cross-national dataset on population flows. This allows them to more precisely identify country- and region-specific patterns of outgoing and incoming migration, and to assess the relative weight of specific variables (e.g., wage differentials between countries, the extent of income inequality and social mobility in sending and receiving countries, civil war, famine, geopolitical location, and migration policy regimes) in explaining these patterns. Finally, the authors discuss the implications of their findings for a more productive understanding of global social stratification and mobility in the contemporary world-economy.

Keywords

Migration social mobility social stratification world-economy

The single most important argument advanced by a world-systems perspective is the notion that space, as territorially bounded units of inclusion and exclusion (involving most importantly, but not solely, nation-states), has been a fundamental axis of global social stratification. If so, migration across such boundaries has been and continues to be a key strategy of social mobility. To pursue this point, we argue that wage differentials are of crucial importance in shaping migration, and that we should use this empirical observation as an opportunity to engage more fully in theorizing migration as a strategy of social mobility used by populations that seek to overcome barriers of exclusion.

There is a certain reluctance to study how migration relates to income differentials. For many, focusing on income differentials as a driving force of migration takes us into the dangerous terrain of emphasizing rational decision-making and market-oriented actions by individuals, when we should concentrate instead on identifying the ways in which migration flows are shaped by social networks and social processes, in ways that respond to the needs of capitalist accumulation rather than challenge existing inequalities. In other words, scholars often assume that an empirical issue (the extent to which migration flows respond to wage differentials) inevitably leads into the deployment of a given theoretical framework (as provided by neoclassical economics).

In this article, we argue that this is not the case. Focusing on the relationship between income differentials and between-country migration can help us in the task of reconceptualizing stratification as a global social process.¹ Rather than as bounded by national borders, a world-systems perspective moves us to understand social hierarchies, stratification, and mobility as fundamentally constituted by the creation of national boundaries and national citizenships. In this regard, nationality joins other relevant ascriptive criteria (e.g., gender, race, age, ethnicity) in shaping hierarchies and inequalities, inclusion and exclusion, on a global scale; and migration can be understood as a key strategy used by populations that seek to overcome such ascriptive barriers of exclusion.² If so, a better understanding of the relationship between income differentials and migration flows can be placed in a different theoretical framework than the one provided by neoclassical economics, to build instead upon a world-systems perspective that highlights the crucial role of agency, stratification, and mobility as driving forces of global migration.

Hence, rather than as a mere equalization of wage rates through the integration of markets, as is often posed from a neoclassical perspective, international migration can be understood from a world-systems perspective as one of the strategies through which some disadvantaged populations are able to actively engage in upward social mobility. From such a perspective, national boundaries can be understood as ‘a global wall erected by the rich industrial states to protect themselves from “invasion” by the world’s poor’ (Zolberg, 1999: 72). As indicated by the same author,

In retrospect, one can only wonder at the strangeness of a historical trajectory whereby most of the population of the world’s richest and most powerful countries, encompassing tens and eventually hundreds of millions of individuals, came to believe that they constituted mutually exclusive ‘natural’ communities, sharing a hallowed ancestry and destined in the divine scheme of things to share a common fate. (Zolberg, 2000: 512)

Indeed, the income gains derived from moving from a poor country into a wealthier one are often vastly superior to any effort to gain upward mobility within a given poor country. We argue that the extent of mobility gained through migration, in fact, serves to ‘flag’ the vast categorical inequality embedded in national borders and citizenships. In such a context, migration from poor to wealthy countries embodies a challenge to one of the most significant forms of categorical inequality shaping contemporary social stratification.

Data and methods

We have constructed a dataset that includes the size of migrant stocks by country of origin and destination in more than 200 countries (and territories); we appended to that a set of economic, geographic, demographic, and political explanatory variables that describe origin and destination countries independently (e.g., gross domestic product [GDP] per capita) and relationally (e.g., distance). Together, this dataset allows us to measure the relationship between income and income differentials on migration while also controlling for those variables that are usually emphasized as relevant in the sociological literature – such as the cost of migration, linguistic and religious homogeneities, and historical ties.

The data on migration used in this exercise come from the World Bank’s Global Bilateral Migration Database. Drawing on over 1000 national censuses and population registers, the database identifies country of origin for migrants in 226 host countries for five census rounds (1960, 1970, 1980, 1990, and 2000). The database was recently updated to 2010 (Global Bilateral Migration Database et al., 2011). In total, the dataset offers over 300,000 observations of bilateral migrant stocks. For some exercises we need estimates of net migration. To this end, we convert our data from a measure of stocks to a measure of flows.³

For our purposes, migrants are defined as individuals that are residing in a country other than their country of origin. We recognize that this definition overlooks important variations in the duration of stay and legal status of global migrants. That said, our relatively simple definition of ‘migrant’ is preferable to a more complicated operationalization of migrant status because (1) the data actually exist for the broader definition but not for a more targeted operationalization of migrant status; and (2) the definition effectively captures the key process we are trying to understand through this exercise: the movement of people who move away from their country of birth to go to reside in a different country.

We measure income differentials between countries using the GDP per capita (at exchange rates in most cases but also using purchasing power parity). In some models we include the GDP growth rate (as the logged ratio of the per capita GDP from the current time to the previous time period) and the unemployment rate. These measures allow us to assess, as often contended in the literature, whether migrants choose to leave countries with high unemployment for countries with low unemployment, high rates of economic growth, and a higher per capita GDP.

Our model controls for many of the variables that usually are cited as relevant to migration flows. Thus, it includes geographical controls, such as distance between countries, shared borders, and country size (Greenwood, 1975; Kim and Cohen, 2010; Mayda, 2005; CIA World Factbook, 2013). It also incorporates a number of demographic controls (Hatton and Williamson, 2005; Kim and Cohen, 2010). These include total population (World Bank, 2013), fertility rate (lagged 20 years; UNPD, 2011), population density (persons per square kilometer), infant mortality (IMR), and life expectancy (World Bank, 2013). Because of the high correlation between IMR, life expectancy, and GDP per capita, we created a ‘Health Index’ to measure the physical well-being of the population independent of the relationship between health and income.⁴

Our model controls for cultural homogeneity (Clark et al. 2007; Greenwood and McDowell, 1982; McLean Petras, 1981): it includes a measure of the extent of religious homogeneity between two countries (i.e., the probability that two randomly selected individuals from each country will be members of the same major religious group); of common language (a source/destination pairing receives a score of 0 if they share no languages and a score of 10 if they share a significant language or a significant language in the sending country is an official language in the receiving country); and colonial relationship (McLean Petras, 1981; Neumayer, 2005; Pederson et al., 2008) (we have coded country pairings using three dichotomous variables based on the relationship of the country of origin to the destination: colony, colonizer, and shared empire).

We include a number of political variables highlighted in the literature as relevant to migration (e.g., Morawska, 2009; Zolberg, 1981, 1999), such as a naturalization index measuring the difficulty of access to such status (US Office of Personnel Management, 2004); a measure of type of political regime (Hsu, 2008); a measure of prevalence of conflict (as classified by the UCDP/PRIO Armed Conflict Dataset; see Gleditsch et al., 2002); and a measure of fractured state status identifying pairings that were once united within the same political entity (these primarily include the states of the former Soviet Union but also include Yugoslavia, Czechoslovakia, Ethiopia/Eritrea, and Sudan/South Sudan among others; this measure helps us assess whether the bilateral migrant stock is greater within fractured states, but these pairings should also experience a relatively high rate of return migration after independence).

We tested a number of other variables that were excluded from the final analysis because while sometimes highlighted in the literature, they did not add substantially to the models or they introduced too many missing values. These include agriculture and manufacturing as a percent of GDP, economic recession (negative GDP growth from year to year), and shared continental landmass.

To estimate the relationship between these independent variables and the size of bilateral migrant stocks and flows, we begin with a gravity model:

S = G \frac{P_{s} P_{d}}{D}

(1)

such that the migrant stock (S) between countries depends on the ‘masses’ (population sizes) of those countries (P_s and P_d) and the distance between countries (D). G is a constant that scales the right side of the equation to fit the left, the intercept in linear modeling parlance (the use of gravity models can be traced back to Ravenstein, 1885 and 1889); the approach generates results that are easy to interpret and are tightly clustered across studies (e.g., Anderson, 2010; Kim and Cohen, 2010).

We can convert the gravity form to a linear equation by taking the natural log of each side, and we can add income and control variables to the model (X_i) to improve the fit. We solve this equation using ordinary least squares regression.

L N (S) = B_{P s} L N (P_{s}) + B_{P d} L N (P_{d}) + B_{D} L N (D) + \sum X_{i} B_{i} + e

(2)

In addition to testing for significance, we use standardized betas to compare the relative size effects of the various factors. All models are weighted to the population of the sending country.

Ultimately, our goal in this exercise is to quantify the impact of incomes in sending and receiving countries on the size of migrant flows while controlling for, but also identifying interactions between, a fairly comprehensive list of control variables. We recognize that this set of control variables does not account for every possible characteristic that might affect decisions of whether or not to migrate. While that might be a limitation of our exercise, we believe the control variables we have identified effectively capture most of the forces cited in the literature as being relevant to migration, and that our model indeed is robust, highlighting social patterns that would stand even if more specific (or individual) characteristics were taken into consideration.

Some might criticize our approach for being individualistic, and for presenting an economic-person model of migration. This criticism would be valid if it were not a bit absurd. We make no claims to predicting or even explaining decision-making processes by any given individual. To the contrary, our goal is to track and understand the social trend at a global level. What we present is the net outcome of billions of individuals weighing thousands of variables in complex ways to make life-altering decisions. We do not pretend to explain migration processes at an individual level. We do not know why any given person chooses to migrate, or how historical circumstances might affect some specific groups, but our exercise indicates that people tend to flow from poorer countries to richer countries, even when controlling for an extensive set of other considerations. To admit that economic incentives are at play is not the same as adopting a reductionist, homo economicus assumption of the human soul. We do not know why any individual is responding to economic incentives, but people are clearly responding to economic incentives, and in very large, significant numbers.

We anticipate that the gap in per capita GDP between sending and receiving countries is the most important factor driving international migration. This relationship is moderated by other factors, particularly the GDP per capita of the sending country and the size of the migrant stock.

Findings

We argue that international migration is a key strategy of upward mobility; many migrants seek to overcome the categorical inequalities of political exclusion across the world system. As such, we anticipate a significant tendency for migrants to leave lower-income countries for higher-income countries. We expect migrants respond to other potential mechanisms that can enable or prevent economic access in destination countries: e.g., linguistic and political barriers, special access granted to former colonies. Finally, migration requires a substantial investment and there is little in the way of credit market alternatives to cover those costs; migrants need access to the resources to cover the costs of migration, and potential migrants choose destinations that minimize the costs of migration.

Our findings, discussed below, provide strong support for our first anticipated result: income per capita of the destination country is a powerful explanatory variable in the size of international migrant flows. The analysis generally supports the other anticipated results, but with some important caveats that we discuss below. We begin by reviewing the main relationship between income differentials and between-country migration, and then proceed to show that this relationship is robust even after controlling for many other variables that are highlighted as relevant in the sociological literature.

Per capita GDP and emigration

It seems logical that, ceteris paribus, a migrant chooses the destination with the higher per capita GDP. But it does not hold that the countries facing the largest income gaps send the most migrants. ‘Lower levels of per worker GDP in the source country both strengthen incentives to leave and make it more difficult to overcome poverty constraints … due to fixed costs of migration and credit-market imperfections’ (Mayda, 2005: 4). Hatton and Williamson (2005: 67), for example, found that ‘rising incomes at home increased the emigration rate by releasing the supply constraint.’

Together, these two observations create contradictory trends. On one hand, a higher GDP per capita in the country of origin increases the number of individuals with access to the necessary resources to migrate. On the other hand, a higher GDP per capita in the country of origin reduces the incentive to migrate. The net effect is that emigration is most common from middle-income countries, whose residents have both the resources and incentive to migrate.

The relationship is captured in Figure 1. We used logistic regression and the first four degrees of logged GDP per capita to estimate the ratio of net emigration (by decade, 1960–2010) to total population. All four degrees of logged per capita GDP are significant. The line plots the predicted emigration rate against per capita GDP for 2000–2010. The result is an inverted-U, such that migration rates increase as incomes increase up to a point, and then fall. More specifically, individuals in countries with a per capita GDP around $8000 (e⁹, in 2000 US$) are the most likely to emigrate while those in the richest countries are the least likely to do so. This confirms findings elsewhere in the literature (e.g., De Haas, 2010).

Figure 1.

Emigration rates by per capita GDP_origin.

Table 1 considers the per capita GDP of both the sending and receiving country between 1990 and 2000. Countries are divided into three categories – low-income (< $1000 per capita GDP), middle (> $1,000, < $10,000), and high (> $10,000) (2000 US$). The table tracks the size of each flow as a percent of total migration during the decade, the emigration rates per 100,000 population in sending countries, the size of the flow versus the expected size of the flow based on the total population of sending countries and the global emigration rate, and the percent of emigrants from each set of countries going to low-, middle-, and high-income countries.

Table 1.

Migration by GDP per capita, 1990–2000.

Origin/Destination	Low	Middle	High	Total
Low
Net flow	6,005,320	6,534,458	8,031,136	20,570,914
% of total	10.5%	11.4%	14.0%	36.0%
Net flow/100K	202.9	220.8	271.3	695.0
vs. exp.	54.3%	59.0%	72.6%	61.9%
Row %	29.2%	31.8%	39.0%	100.0%
Middle
Net flow	2,589,782	5,933,002	17,700,000	26,222,784
% of total	4.5%	10.4%	30.9%	45.8%
Net flow/100K	244.3	559.7	1,669.8	2,473.8
Vs. exp.	65.3%	149.7%	446.5%	220.5%
Row %	9.9%	22.6%	67.5%	100.0%
High
Net flow	943,974	2,622,697	6,853,890	10,420,561
% of total	1.6%	4.6%	12.0%	18.2%
Net flow/100K	87.4	242.8	634.6	964.9
Vs. exp.	23.4%	64.9%	169.7%	86.0%
Row %	9.1%	25.2%	65.8%	100.0%
Total
Net flow	9,539,076	15,090,157	32,585,026	57,214,259
% of total	16.7%	26.4%	57.0%	100.0%
Net flow/100K	187.0	295.9	638.9	1,121.8
Vs. exp.	50.0%	79.1%	170.9%	100.0%

Low-income countries accounted for only 36.0% of global emigrants despite containing more than half of the world’s inhabitants. They produced only 61.9% as many migrants as expected based on the size of the population. Middle-income countries produced almost half of migrants and more than two times (220.5%) as many as expected based on the total population of middle-income countries. At 18.2%, high-income countries sent fewer than one in five international migrants between 1990 and 2000.

Middle-income countries, on the other hand, produced 45.8% of all migrants. The largest net flow between 1990 and 2000 was from middle-income countries to high-income countries, both in terms of the total figure (17.7 million) and as the rate per 100,000 people in sending countries (1669.8). Almost a third of migration between 1990 and 2000 involved individuals moving from a middle-income to a high-income country. In fact, migrants from both high- and middle-income countries were twice as likely to end up in high-income countries as low- and middle-income countries combined. Migration from lower-income countries was functionally different that from middle- and higher-income countries. Migrants from low-income countries were still more likely to end up in high-income countries than middle- or lower-income countries, but almost a third migrated to another low-income country. Their options are constrained by a lack of resources, information, and internationally inconsistent immigration restrictions (e.g., more liberal migration rules between European Union members).

For workers in lower-income countries, migration is a much more direct path to achieving wage convergence with workers in rich countries than riding economic growth in their own country. This point is illustrated in Figure 2. The chart tracks income differences for various groups between 2000 and 2010. Each black square represents a country, located using the per capita GDP in 2000 on the x-axis and in 2010 on the y-axis. If the per capita GDP of a country did not change during this period, the square would center on the diagonal; squares above the diagonal represent countries that experienced real growth in terms of GDP per capita during the period. The size of squares reflects the relative population of the country. The circles track migrant groups by source and destination country, plotting the per capita GDP in 2000 of the source country and the per capita GDP in 2010 of the destination country. The circles are sized according the relative size of each migrant flow.

Figure 2.

GDP per capita via national economic growth and migration, 2000–2010.

Over the 10-year period, the opportunities for broad upward mobility within a country, even accounting for the few cases with rapid economic growth, were limited. The potential for broad upward mobility through migration, though, is substantial. For example, India and China, the two largest squares in Figure 2, have clearly enjoyed growth in GDP per capita, but that movement is fairly insubstantial compared to the GDP per capita of rich countries. In this context, it makes sense that the large majority of migrants between 2000 and 2010 ended up in one of the world’s richest countries.

Estimating migrant stocks

Having established the bivariate relationship between GDP per capita and international migration, we test the significance of this relationship against our array of control variables. In Tables 2 and 3, we estimate bilateral migrant stocks from 144 sending to 149 destination countries in 2000 using the gravity model framework. We used the average in cases where a metric varies over time (e.g., unemployment rate), and coded regime type as the percent of years by each regime type. We report clustered standard errors, clustered on the sending country, for significance testing to account for potential dependence between these observations.

Table 2.

Estimating bilateral migrant stocks, 2000, OLS regression.

Bilateral migrant	Base + GDP per capita		Full Model 1		Full Model 2
Stock (logged)	Beta	S-Beta	Beta	S-Beta	Beta	S-Beta
Population
Origin	1.049***	0.502	0.860***	0.412	0.816***	0.378
	0.044		0.057		0.059
Destination	0.646***	0.300	0.685***	0.318	0.733***	0.352
	0.020		0.022		0.021
Distance	–1.636***	–0.306	–1.455***	–0.273	–1.464***	–0.276
	0.136		0.114		0.133
GDP per capita
Origin	1.334***	0.604	1.094***	0.496	1.228***	0.565
	0.080		0.133		0.160
Difference	0.940***	0.578	0.846***	0.520	1.008***	0.626
	0.037		0.081		0.109
Controls	No		Yes		Yes
Intercept	–21.150***		–17.557***		–19.116***
	1.859		2.239		2.330
R²	0.458		0.602		0.629
Pairs (N)	21,323		21,323		14,150
Origin countries	144		144		115
Dest. countries	149		149		124

Notes: *p < .05, **p <.01, ***p < .001; observations weighted to the population of the sending country; all non-indexed, non-dichotomous variables are logged; significance testing based on clustered standard errors on the sending country.

Sources: CIA World Factbook (2013), Gleditsch et al. (2002), Global Migration Database et al. (2011), Hsu (2008), UNPD (2011), US Office of Personnel Management (2004), World Bank (2013).

Table 3.

Control variables for Table 2.

Bilateral migrant	Full Model 1		Full Model 2
Stock (logged)	Beta	S-Beta	Beta	S-Beta
Coast
Origin	0.255	0.017	–0.011	–0.001
Destination	0.198*	0.023	–0.238*	–0.028
Shared border	1.237**	0.063	0.782	0.043
Area/10^6
Destination	0.015	0.008	–0.010	–0.006
Pop. density/10^3
Origin	0.091	0.004	0.091	0.004
Destination	–0.235*	–0.020	–0.545***	–0.052
Fertility rate
Origin	–0.105	–0.042	–0.154**	–0.062
Destination	0.084	0.042	0.243	0.123
Health
Origin	0.117**	0.065	0.083	0.046
Destination	0.222***	0.105	0.305***	0.134
Shared language	0.124***	0.189	0.119***	0.195
Shared religion	0.243	0.020	0.516	0.042
Colony	0.291	0.008	0.287	0.009
Colonizer	1.047**	0.029	0.877*	0.025
Shared empire	0.678***	0.078	0.747***	0.087
Fractured state	2.772	0.064	2.787	0.062
Naturalization
Destination	–0.007	–0.011	–0.024*	–0.039
Regime (vs. Soc. Dem.)
Dictatorship	–1.257***	–0.113	–1.104***	–0.095
Mil. dictatorship	–1.576***	–0.089	–1.819***	–0.099
Civil war	–0.617	–0.018	–0.965*	–0.020
One-party democ.	–1.468***	–0.051	–1.143***	–0.042
Communist	–4.280***	–0.320	–2.738***	–0.193
Islamic republic	–3.058***	–0.073	–4.225***	–0.066
European colony	–0.684	–0.018	0.390	0.011
Cons. democracy	–0.667***	–0.069	–0.330	–0.034
Failed state	–1.149***	–0.045	–0.359	–0.013
Conflict/10^3
Origin	–0.049*	–0.043	–0.025**	–0.022
Destination	–0.232	–0.068	–0.307	–0.094
Growth
Origin			14.248***	0.097
Destination			14.073***	0.091
Unemployment
Origin			0.025	0.026
Destination			–0.066***	–0.112

Notes: See Table 2.

We begin with the base model (distance and population), GDP per capita in the sending country and the GDP per capita gap between sending and receiving countries. These five variables explain almost 50% of the variance in logged bilateral migrant stocks. Further, our GDP per capita measures have the largest size effect of the five according to the standardized beta – that is to say, when we convert the independent variables to the same scale, a one unit change in GDP per capita in the sending country has a larger effect on migrant stocks than distance and the respective populations. The coefficient for the population of the origin country is close to 1; this implies that the size of this population does not have a unique effect on emigration rates, but given a standard emigration rate, the larger population will tend to produce more emigrants.

In the next iteration of the model we add a number of control variables (Table 3). The additional variables improve the fit of the model but not dramatically. The unexplained variance falls 26.6% from 54.2% to 39.8%. The results are generally consistent with our expectations with one major exception: the stock of migrants from former colonies is not significantly larger than in other pairings. This is a salient finding, as much of the literature generally emphasizes colonial relationships as highly significant to both migrant stocks and flows.

Shared language and good health in destination countries are very important in determining the size of migrant stocks, and communist states are particularly abhorrent to potential migrants. But the most important factor driving the size of bilateral migrant stocks, according to the standardized betas, is the gap in GDP per capita between sending and receiving countries. The size effect of GDP per capita of the sending country is also very large. The standardized beta of the per capita GDP gap is more than two and half times larger than for all control variables.

Gross domestic product in these models is valued using exchange rates. If migration is a strategy individuals deploy to improve their standard of living, GDP figures adjusted to purchasing power parity (PPP) might better capture the forces driving migration. We substituted PPP-adjusted GDP per capita for the sending country and the differential, one at a time and together (not shown). In each case, the explanatory power of the model fell slightly, suggesting that GDP per capita at exchange rates are, in fact, more relevant. We offer three explanations for this result. First, migrants do not always or immediately adopt local purchasing habits; so locally adjusted PPPs overstate cost of living differences. Second, migrants often use global inequalities to overcome local inequalities or constraints in their country of origin, either for family members through remittances or with the intention of returning home; they are more interested in global command over resources than local living standards. Third, adjusting to PPP is a complex and incomplete science, and doing so may introduce noise that reduces the variables’ explanatory power.

In the final model, we add per capita GDP growth and unemployment. Unfortunately, missing data cut the samples by a third, but results are consistent from Full Model 1 to Full Model 2. Per capita GDP growth and unemployment have the anticipated results: migrants target countries with low unemployment and prefer rapidly growing economies. Notably, even after controlling for economic growth in the sending country, GDP per capita in the sending country is still positively and significantly associated with emigration. The level of GDP per capita in sending countries, independent of economic growth trends, is many times more important in explaining emigration.

The models above measure only the unique effects of the independent variables on migrant stocks, but the effects of these variables can also depend on each other. An obvious example is that the opportunity cost of distance is greater for poorer potential migrants, so the effect of distance on the size of migrant stocks will be greater for migration from poorer countries. To test this relationship, we added the distance*GDPPC_origin interaction term to Full Model 1. We used the coefficients for distance, GDPPC_origin and the interaction term to estimate the net effect across a range of income levels for countries at two different distances. We used m=n resampling with replacement to estimate confidence intervals.

The vertical axis in Figure 3 measures the expected change in the logged migrant stock based on the distance and GDP per capita of the sending country against a baseline scenario of migration from a very poor country (GDP per capita = $148.41) to a distant destination (distance = 8100 miles). First, if we trace the bottom of the two lines, we find a strong correlation between the per capita GDP of the sending country and the expected size of the migrant stock: as the sending country gets richer, the expected size of the migrant stock increases. This line represents migration between two distance countries (e.g., Canada and Vietnam). The second of the two lines reflects the relative size of migrants stock between two countries with population centers approximately 150 miles apart – Honduras to El Salvador or Ghana to Togo. Beginning at very low levels of GDP per capita, we see that the expected migrant stock between these pairs is much larger than between the more distant countries. Also, the size of the migrant stock increases with the income of the sending country, but the correlation (the slope of the line) is much weaker, such that the gap in the expected size of the migrant stock by distance is much smaller at very high levels of GDP per capita. In other words, higher incomes in sending countries always facilitate migration, but especially across great distances where migration costs (such as transportation) are greater.

Figure 3.

Distance, GDPPC_origin and the size of the migrant stock.

We tested interaction terms for the per capita GDP gap by language, religion, colony, and naturalization. The coefficient for GDPPC_gap*language and GDPPC_gap*colony is positive and significant. This means that the effect of the income gap on migration is greater for former colonies and between countries with high linguistic homogeneity than in other cases. While it is possible that larger income gaps increase migration by increasing the appeal of linguistic homogeneities, it is more likely that linguistic homogeneities make it easier for migrants to exploit international income gaps.

The coefficient for GDPPC_gap*naturalization index is also positive. Because the naturalization index is negatively associated with migration, this positive coefficient means that the effect of stricter naturalization requirements have less impact on migration when income gaps are larger. GDPPC_gap*religion was not significant at the .01 level.

By social science standards the percentage of variance explained by our model is very high. The full model explains 60% of the variance in the logged size of bilateral migrant stocks globally.⁵ The critical observation, hence, is robust: a substantial share of international migrants recognize the limitations of their own geography, and choose destinations that allow them upward social mobility.

Discussion

In light of our findings, it is important to reiterate the observation made at the beginning of this article. Rather than as a mere equalization of wage rates through the integration of markets, as is often posed from a neoclassical perspective, international migration can be understood from a world-systems perspective as one of the strategies through which some disadvantaged populations are able to actively respond to global income inequality (here measured as GDP per capita) and engage in upward social mobility.

Hence, emphasizing the crucial role played by wage differentials in shaping migration does not mean the displacement of the social by the individual. The processes that respond to such wage differentials (as well as those that have generated the latter) are social at their very core. Constantly, Schumpeterian cycles of creative destruction create new labor demands in some places while reducing demand in others. Through migration, people often seek to escape the consequences of ‘destruction’ and seek to secure access to the benefits of innovation (as inherent in social mobility). But migration is a complex, expensive, risky, and, with the emergence of the nation-state, politically sensitive activity. When migration is limited, the benefits of innovation become geographically constrained – the very unequal global distribution of wealth reflects these constraints.

We have argued elsewhere (e.g., Korzeniewicz and Moran, 2009) that migration has come to constitute the single most immediate and effective means of global social mobility. For example, Malaysia and Mexico are receiving countries for migrants from Indonesia and Guatemala (respectively), and sending countries to Australia and the United States (again, respectively). In the case of Guatemala, anyone belonging to the poorest six deciles would gain upward mobility by gaining access to the average income of the second poorest decile in Mexico; likewise, 70% of the population of Indonesia would be upwardly mobile were they able to attain the average income of the second poorest decile in Malaysia. In the case of Mexico, the incentives are even more striking, as all but the wealthiest decile would find upward mobility in gaining access to the average income of the second poorest USA decile; and only the richest decile in Malaysia would not be worse off than the second poorest decile in Australia. Such disparities help explain why economic migrants often are willing to abandon a professional status in their country of origin to work in more menial positions in their country of destination. In a world in which nationhood continues to be central in shaping global stratification, ‘jumping’ categories by moving from a poorer country to a wealthier one is a highly effective strategy of mobility. Migration in fact represents a sort of Schumpeterian innovation that draws on the opportunities provided by market mechanisms to overcome exclusion and political barriers to entry.

Prevailing critical perspectives, drawing on the notion that the extension of markets is promoted by elites to gain power and privilege, and emphasizing the importance of voice as a means of enhancing income and status, might balk at the notion that migrants might challenge hierarchies by using market mechanisms in their own favor. But, pace Polanyi, in situations in which privilege is obtained through exclusion (rather than exploitation), challenges to exclusionary mechanisms by the promotion of more competitive markets can be an effective means of pursuing social mobility. In this sense, ‘excluded’ populations can exploit opportunities (paradoxically, as generated by exclusion) to competitively ‘push’ their way into markets, leading to market expansion from below. To the extent that national borders have become a primary mechanism of selective exclusion, migration, as a push for inclusion, represents a significant avenue of challenging exclusion and pursuing social mobility. Both migrants (in their crossing of borders constitutive of stratification) and non-migrants (when constrained by borders that hamper such crossings) reveal the relative boundaries and forces of social stratification and mobility to be truly global.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

Author biographies

Roberto Patricio Korzeniewicz (PhD, Binghamton University, 1989) is Professor of Sociology at the University of Maryland, College Park (USA), where he has taught since 1993. His book Unveiling Inequality (2009), co-written with Timothy P Moran, won the 2010 Best Book Award of the Political Economy of the World-System section of the American Sociological Association. He is the co-editor, with William C Smith, of Latin America in the World-Economy (1996), and Politics, Social Change, and Economic Restructuring in Latin America (1997). His articles have appeared in the American Journal of Sociology, Bulletin of Latin American Studies, Desarrollo Económico, Economic Development and Cultural Change, Hispanic American Historical Review, Latin American Research Review, Political Power and Social Theory, Revista Mexicana de Sociología and Sociological Forum. His current research focuses on global patterns of income inequality, social stratification and mobility, and on historical and current patterns of change among social movements in Latin America.

Scott Albrecht is a postdoctoral research assistant in the Maryland Population Research Center (MPRC) at the University of Maryland, College Park. Albrecht has published on US economic inequality and completed a dissertation on wealth and income concentration through US history. With Roberto Korzeniewicz, Albrecht’s articles include ‘Thinking globally about inequality and stratification: Wages across the world, 1982−2009’ (International Journal of Comparative Sociology, 2012) and ‘Global wages and world inequality: The impact of the Great Recession’ (in C Suter and C Chase-Dunn, eds, Structures of the World Political Economy and the Future Global Conflict and Cooperation, 2014).

References

Anderson

(2010) The gravity model. NBER Working Paper 16576.

Boatcă

(2015) Global Inequalities Beyond Occidentalism. London: Ashgate.

CIA World Factbook (2013) The World Factbook 2013–14. Washington, DC: Central Intelligence Agency. Available at: www.cia.gov/library/publications/the-world-factbook/index.html (accessed 15 July 2013).

Clark

Hatton

Williamson

(2007) Explaining US migration, 1971–1998. The Review of Economics and Statistics 89(2): 359–373.

De Haas

(2010) Migration and development: A theoretical perspective. International Migration Review 44(1): 227–264.

Gleditsch

Wallensteen

Eriksson

. (2002) Armed conflict 1946–2001: A new dataset. Journal of Peace Research 39(5): 615–637.

Global Bilateral Migration Database, World Bank Group and Özden

. (2011) Where on earth is everybody? The evolution of global bilateral migration, 1960–2000. World Bank Economic Review 25(1): 12–56.

Greenwood

(1975) Research on internal migration in the United States: A survey. Journal of Economic Literature 13(2): 397–433.

Greenwood

McDowell

(1982) The supply of immigrants to the United States. In: Chiswick

(ed.) The Gateway: US Immigration and Policies and Issues. Washington, DC: American Enterprise Institute, pp. 54–85.

10.

Hatton

Williamson

(2005) Global Migration and the World Economy: Two Centuries of Policy and Performance. Cambridge, MA: MIT Press.

11.

Hsu

(2008) The effect of political regimes on inequality, 1963–2002. UTIP Working Paper 53, 22 September. Available at: utip.gov.utexas.edu (accessed 25 March 2012).

12.

Kim

Cohen

(2010) Determinants of international migration flows to and from industrialized countries: A panel data approach beyond gravity. International Migration Review 44(4): 899–932.

13.

Klinthall

(2006) Immigration, integration, and return migration. In: International Symposium on International Migration and Development, Turin, Italy, 28–30 June 2006.

14.

Korzeniewicz

Moran

(2009) Unveiling Inequality. New York: Russell Sage Foundation.

15.

McLean Petras

(1981) The global labor market in the modern world-economy. In: Kritz

Keely

Tomasi

(eds) Global Trends in Migration: Theory and Research on International Population Movements. Staten Island, NY: The Center for Migration Studies of New York, pp. 44–63.

16.

Mayda

(2005) International migration: A panel data analysis of economic and non-economic determinants. IZA Discussion Paper No. 1590.

17.

Morawska

(2009) A Sociology of Immigration: (Re)Making Multifaceted America. London: Palgrave Macmillan.

18.

Neumayer

(2005) Bogus refugees: The determinants of asylum migration to Western Europe. International Studies Quarterly 49(3): 389–409.

19.

Pederson

Pytlikova

Smith

(2008) Selection and network effects: Migration flows into OECD countries 1990–2000. European Economic Review 52(7): 1160–1186.

20.

Ravenstein

(1885) The laws of migration. Journal of the Royal Statistical Society 48(2): 167–235.

21.

Ravenstein

(1889) The laws of migration: Second paper. Journal of the Royal Statistical Society 52(2): 241–305.

22.

Shachar

(2009) The Birthright Lottery: Citizenship and Global Inequality. Cambridge, MA: Harvard University Press.

23.

UNPD (United Nations, Department of Economic and Social Affairs, Population Division) (2011) World Population Prospects: The 2010 Revision. New York: United Nations. Available at: data.un.org/Data.aspx?d=PopDiv&f=variableID:54

24.

US Office of Personnel Management (2004) Citizenship laws of the world.

25.

World Bank (2013) World Development Indicators. Available at: data.worldbank.org/developers/apps/wbopendata (accessed 23 April 2012 and 2 May 2013).

26.

Zolberg

(1981) International migrations in political perspective. In: Kritz

Keely

Tomasi

(eds) Global Trends in Migration: Theory and Research on International Population Movements. Staten Island, NY: The Center for Migration Studies of New York, pp. 3–27.

27.

Zolberg

(1999) Matters of state: Theorizing immigration policy. In: Hirschman

Kasinitz

DeWind

(eds) The Handbook of International Migration: The American Experience. New York: Russell Sage Foundation, pp. 71–93.

28.

Zolberg

(2000) The dawn of cosmopolitan denizenship. Indiana Journal of Global Legal Studies 7(2): 511–518.