Introduction

Migration is an essentially multidisciplinary and interdisciplinary topic, drawing the attention of geographers, sociologists, economists, anthropologists, political scientists and so forth. Yet, as far as the ‘grand’ theories of migration are concerned—i.e. those theoretical accounts that seek to account for the more or less permanent movement of people—, there is a clear disciplinary prominence of economics: other relevant but arguably secondary theoretical contributions notwithstanding, this theoretical field has been traditionally characterized by a pronounced opposition between the standard neoclassical orthodoxy, on the one hand, and the historical-structural heterodoxy, on the other, which can be respectively regarded as derivative applications of neoclassical economics and (neo-)Marxist political economy. In the context of the traditional opposition between agency (i.e. independent individual action and choices) and structure (i.e. recurrent patterned arrangements that influence or limit the choices and opportunities of individuals) as the fundamental loci of explanation in social science (Barker 2005), the neoclassical approach would clearly fall on the side of the former, whereas the historical-structural perspective, as the name itself suggests, of course emphasizes the latter.

More recently, an alleged theoretical ‘third way’ has been put forth under the guise of the so-called New Economics of Labor Migration, or NELM, which has purportedly sought to redress both the excessive structural emphasis of the historical-structural perspective and the theoretical insufficiencies of the standard neoclassical theoretical framework. According both to its proponents (e.g. Taylor 2001) and to latter-day authors reviewing the literature (e.g. De Haas 2010; Hagen-Zanker 2008), the NELM has thus constituted a fundamentally new theory—one which effectively reconciled agency and structure and allowed for “a greater variety of outcome than would have been allowed from either the single aggregation of individual decision making or from the unidirectional imperatives of structures” (De Haas 2010:242).

This paper takes a different view and instead argues that the NELM is characterized by the same fundamental flaws as the standard neoclassical theoretical account, albeit in a more sophisticated information-theoretic clothing. In this respect, it constitutes migration theory’s own instance of what Wade (1996) has called “the art of paradigm maintenance”: bulwarking the central tenets of a theoretical body against rising contestation through peripheral concessions and readjustments. Such concessions and readjustments notwithstanding, however, it is argued that the methodological individualism that characterizes both the standard neoclassical theory and its NELM avatar structurally prevent them from constituting satisfactory theoretical accounts of migration. For that, one must look instead to the contributions of the historical-structural perspective, however much in need of a new ‘synthesis’ and a number of theoretical readjustments of its own this perspective may be.

In order to put forth these arguments, this paper is structured as follows: after this introduction, "Theories of Migration: The Neoclassical, Historical-Structural, and NELM Perspectives" section provides a summary presentation of the standard neoclassical theory of migration, the historical-structural heterodoxy, and the NELM. "The New Economics of Labor Migration: Beware of Neoclassicals Bearing Gifts" Section highlights why it is that the NELM should be regarded as a ‘reworked’ version of the neoclassical theoretical framework that is unable to shed the fundamental weaknesses of the latter, and discusses its inception in the context of the ‘information-theoretic revolution’ in economics. Then, "Towards a New Historical-Structural Synthesis" section argues in favor of a new and improved ‘historical-structural synthesis’ as a more satisfactory alternative to both the NELM and the standard neoclassical theory. Finally, "Conclusions" section sums up the main conclusions.

Theories of Migration: The Neoclassical, Historical-Structural, and NELM Perspectives

By and large, one thing which most surveys of the literature on the determinants of migration agree upon is the pronounced opposition between the “equilibrium” and “historical-structuralist” theories of migration (Wood 1982), which up until recently featured as the two main theoretical perspectives dominating this field. The standard neoclassical theory of migration (seminal formulations of which can be found in Ranis and Fei 1961; Sjaastad 1962; Todaro 1969 and Harris and Todaro 1970), with both its ‘micro’ and ‘macro’ components, is subsumed in the former and presented as heir to the tradition initiated by Ravenstein (1885 and 1889) and pursued by Lee (1966), namely as of the latter’s formulation of the ‘push-pull’ theory of migration. By contrast, the historical-structural approach to migration is typically presented as a rather loose set of theoretical propositions that focus primarily on the structural demand for migrant labor in advanced capitalist societies (Piore 1979; Castles and Kosack 1973; Nikolinakos 1975; Sassen 1988 and 1991) and on ‘world-systemic’ interpretations of the migration-inducing effects of the penetration of capital(ism) in peripheral areas (Petras 1981; Massey 1988). The third main theoretical perspective that has come to dominate the theoretical understanding of migration is the New Economics of Labor Migration, or NELM (Katz and Stark 1986; Stark and Bloom 1985; Stark and Taylor 1989), which has been presented (e.g. in De Haas 2010:240-1) as a “structurationist” and “more nuanced” third way between the agency orientation of the equilibrium theories and the structural emphasis of the historical-structural approaches. In addition to these theoretical explanations of the determinants of migration, other relevant theoretical approaches have sought to address more specific aspects of the migration process, or to apply to migration insights from more general or abstract approaches. Specially noteworthy among the latter are migration systems theory (e.g. Mabogunje 1970; Fawcett 1989), migration networks theory (e.g. Gurak and Caces 1992) and the ‘mobility transition’ approach (Zelinsky 1971; Skeldon 1990 and 1997). In the following pages, however, the main focus is on the first three of the aforementioned theoretical perspectives (neoclassical, historical-structural and NELM), as these are deemed more central both to the structure of this theoretical field and to the argument presented here.

The Neoclassical Orthodoxy

Theorizing on human migration as a topic in and of itself is usually traced back to Ravenstein’s (1885 and 1889) ‘”laws of migration”, an inductive attempt to draw general inferences from empirical data on migration patterns in the United Kingdom and across Europe. Ravenstein’s seminal work was to have little continuation in the subsequent half century, as shown by Everett Lee’s (1966:48) comment to the effect that “this century has brought no comparable [to Ravenstein’s] excursion into migration theory”. It was not until the 1950s and 1960s that new and more ambitious attempts to address the causes of migration were to be pursued again—including, in particular, the birth and rise of the neoclassical theory of migration. A number of theoretical developments were of special relevance in the emergence of the latter: Lee’s (1966) ‘push-pull’ theory of migrant agency; Lewis’ (1954); Ranis and Fei’s (1961) and Todaro’s (1969) seminal accounts of migration within the specific context of developing countries, which came to form the basis of the ‘standard’ neoclassical theory of migration; and Sjaastad’s (1962) ‘micro’ specification of migration as investment.

Lee’s aim was to develop “a general schema into which a variety of spatial movements can be placed” (Lee 1966:49), by putting forth a general hypothesis with respect to migrant agency that was to become known as ‘push-pull theory’. In his own formulation (id ibid:49–50), “the factors that enter into the decision to migrate and the process of migration may be summarized under four headings, as follows: i) factors associated with the area of origin; ii) factors associated with the area of destination; iii) intervening obstacles; and iv) personal factors”. The decision to migrate is thus the result of a (cost-benefit) comparison between the attractive and repulsive features of both areas, though this comparison and the enactment of its results are constrained by “natural inertia”, distance, information, personal factors, etc. (id ibid:51). Curiously enough, considering that the rational-choice foundations of neoclassical economics’ later formulations of migrant agency are often traced back to Lee’s paper, this author stresses (also in p.51) that “the decision to migrate is never completely rational, and for some persons the rational component is much less than the irrational”.

The next milestone in this theoretical trajectory was Arthur Lewis’ contribution. Although Lewis could hardly be considered a neoclassical economist, his account of the role of internal migration in the development process would be selectively extracted from his theoretical framework to form the basis of the neoclassical theory of migration. In reality, however, the most commonly cited of his articles in this context (Lewis 1954) did not seek to put forth a theory of the determinants of migration. Rather, its focus was on the economic duality characteristic of underdeveloped countries, in which a traditional sector characterized by the presence of redundant labor and a low capital-labor ratio exists alongside a modern sector characterized by higher levels of capital intensity. In this model, the different capital-labor ratios in the two sectors entail different marginal productivities of labor and different wage levels, which in turn encourage the workers in the traditional sector (specially “redundant” ones) to migrate to the modern sector (most often spatially concentrated in cities). This enabled Lewis to suggest: i) that the process of economic development inherently involves the absorption by the modern sector of the “surplus labor” from the traditional sector; and ii) that the competition undertaken by the migrant workers enables the modern sector to lower its wages to the level practiced in the traditional sector, thus explaining the high profits and capital rents that characterize the modern sector in these countries.

Then, by their own admission, Ranis and Fei (1961) drew heavily on Lewis’ work and explicitly reaffirmed that “development consists of the re-allocation of surplus agricultural labor, whose contribution to output may have been zero or negligible, to industry where they become productive members of the labor force at a wage equal (or tied) to the institutional wage in agriculture” (Ranis and Fei 1961:533). The main aim of their contribution was to incorporate “a [more] satisfactory analysis of the subsistence or agricultural sector” into Lewis’ model. At the risk of oversimplification, they did this by positing the existence, during the initial “stages” of development, of an institutional wage in the traditional sector set by non-competitive forces. As the absorption of labor by the modern sector proceeds, a point is reached when the marginal productivity of labor in the agricultural sector is equal to or greater than the institutional wage. According to these authors, that point marks not only the “full commercialization of the agricultural sector”, but also “a non-arbitrary criterion for an economy reaching the threshold of so-called self-sustaining growth” (id ibid: 537).

In the wake of both Lewis and Ranis & Fei came M. Todaro and J. Harris, whose model of rural-urban migration (Todaro 1969, Harris and Todaro 1970; Todaro 1976) came to be considered, and abundantly cited since, as the neoclassical model of migration par excellence. Though the purported aim was to account for “the chronic problem of unemployment and underemployment in almost every contemporary developing country” (Todaro 1969:138), this was done by explicitly putting forth a full-fledged “behavioral model of rural-urban labor migration”, in which “the percentage change in the urban labor force during any period is governed by the differential between the discounted streams of expected urban and rural real income” (id ibid:141). Because most migration potentially takes place from rural to urban areas, the migration decision-making process in this model is basically portrayed as a comparison between the discounted future streams of real rural income, which is known by the potential migrants, and the discounted future streams of the income that those migrants expect to earn by migrating to the cities (which is defined as the urban real income weighed by the probability of employment). An “urban-traditional” sector is additionally hypothesized, to which belong those rural-urban migrants that, at any given moment, have not yet been “selected from the pool” of similar migrants by the modern sector (id ibid:142)—thus accounting for urban unemployment and underemployment.

It is particularly hard to characterize Harris & Todaro’s model as either ‘macro’ or ‘micro’. On the one hand, it is presented as a “behavioral model of migration”; on the other, it seeks to account for ‘macro’ outcomes, and does so by aggregation of individual (optimizing) decisions. In that sense, it is indeed a typical methodologically-individualist neoclassical model, insofar as there is no room for forces or constraints operating at the structural level. The ‘macro’ implications, then, are that migration serves as an optimal resource-allocation mechanism, whereby labor is transferred from labor-abundant to labor-scarce regions and whereby factor returns are equalized.

Interestingly enough, the Todaro model came after a considerably less-cited, but arguably more realist, neoclassical behavioral model: Sjaastad’s (1962) “investment” model. This latter model proposes to “treat migration as an investment increasing the productivity of human resources” (id ibid:83, emphasis in the original), whereby workers incur in the “money and non-money” costs associated with the migration process, under the expectation of increasing the lifetime (money and non-money) returns to their provision of labor and choice of location. Note, however, that due to the difficulty of measuring or estimating the “non-money” costs and returns, as well as to the existence of “barriers to the free movement of labor” (id ibid:91), this author is careful not to derive a “macro” model of migration by simple aggregation. Ultimately, then, Sjaastad’s model may be regarded as closer to Lee’s (later) ‘push-pull’ model than to Todaro’s, despite its greater emphasis, compared to Lee’s, on the economic aspects of the decision-making process and its implicit rational-choice and perfect information assumptions.

When we examine the gradual emergence of the ‘standard’ neoclassical theory of migration, it is perhaps puzzling how Harris and Todaro’s model came to be considered its quintessential representative. Indeed, this model is arguably and essentially a theory of migrant agency, rather than a full theoretical explanation of migration processes. Moreover, it may be regarded as more oversimplifying and less sophisticated than other models of migrant agency put forth at roughly the same time (namely Sjaastad’s, also from a neoclassical perspective, or Lee’s). Furthermore, as concerns its ‘unrolling’ into a ‘macro’ model of the determinants of migration, it should be stressed that neither Lewis nor Ranis & Fei—in whose footsteps, by their own admission, Harris & Todaro followed—sought to put forth a theory of migration. Rather, their contributions were theoretical accounts of the process of economic development more generally, in which internal migration from the subsistence to the modern sectors was assumed to occur more or less automatically as a consequence of wage differentials. There is, however, a significant difference between an assumption in a given model and an actual model—a theoretical ‘leap of faith’ that, we would argue, was first made by Harris & Todaro never to be unmade. The transition from Lewis to Ranis & Fei to Harris & Todaro is therefore particularly interesting in that it arguably corresponds to a case of a model developing out of a simplifying hypothesis, itself selectively extracted from a model that sought to explain broader socioeconomic processes. Its rise to the status of ‘representative’ model is also interesting in that it suggests the often superior power and popularity of (overly) simple and (over-)simplifying ideas.

Whether in a more or less oversimplifying version, however, and whether in its ‘micro’ or ‘macro’ dimensions, it should be clear that the neoclassical theory of migration is nothing more nor less than a derivative application of neoclassical economics to the particular field of migration, with all of its characteristic features and assumptions: methodological individualism, optimization, rationality, hypothetical-deductivism, equilibrium.

Historical-Structural Perspectives

The main alternative to the neoclassical theoretical account of migration has traditionally been the historical-structural approach. It is not a unified theory, however, nor is there a ‘representative’ model analogous to Todaro’s. Indeed, as argued by Wood (1982:301), the historical-structural approach is “considerably more difficult to summarize”, due in part to the fact that it is to be found “in a variety of models”. Still, a number of features common to the various theoretical accounts within this perspective can be identified that make it possible to characterize this approach as a whole: i) its structural emphasis and lack of (even concern for) a theory of individual migrant agency; ii) the fact that migration, both internal and international, is regarded as part and parcel of broader processes of structural change (i.e. development), rather than as a “discreet element of social reality that can be subjected to separate investigation” (id ibid:301–302); iii) its typical recourse to an inductive and/or dialectical—in a word, historical—methodology, in contrast to neoclassical theory’s hypothetical-deductivism; and iv) its rejection of ‘equilibrium’ as a structuring principle.

This being said, it is certainly true that the main theoretical insights and propositions that make up the historical-structural approach emphasize different aspects of the migration phenomenon and it is only by bringing them together that the theory becomes a more comprehensive and integrated whole. One possible distinction is that between those authors that emphasize the processes characteristic of ‘migrant-attracting’ areas and those that focus on what takes place in ‘migrant-producing’ areas. The main argument of the former is that capitalist development in the advanced industrialized economies intrinsically requires a constant inflow of workers, which creates a structural incentive for immigration to occur. Castles and Kosack (1973) and Nikolinakos (1975), for example, call upon an explicitly Marxist lexicon and line of thinking to stress the role of labor migrants as an “industrial reserve army”, legally and culturally vulnerable to greater exploitation, which makes it possible for capital to undermine previous or potential conquests by the indigenous working class. Additionally, institutionally regulated in-migration serves to split the working class by giving large sections of its indigenous component the “consciousness of a labor aristocracy” (id ibid:6), thus reducing the likelihood and immediacy of counter-systemic uprisings.

In Piore’s account (1979; also Berger and Piore 1980), the focus is on the idea that the labor markets of advanced industrialized economies are characterized by an inherent dualism, whereby a primary sector characterized by skilled, stable and well-paid jobs exists alongside a secondary sector characterized by unstable, unskilled, poorly remunerated and often ‘3-D’ (dirty, dangerous and demanding) jobs. This economic dualism is regarded as inherently flowing from the duality of labor and capital as factors of production, whereby capital is regarded as essentially ‘fixed’ and labor as ‘variable’ (for capitalists typically respond to downward shifts in demand by laying off workers rather than capital). As a consequence of this, those workers that perform more ‘strategic’ tasks in more stable and capital-intensive production processes will typically be able to secure employment conditions characterized by higher wages and better working conditions, which typically also command higher social status. In other words, “they become more like capital” (Massey et al. 1993:442).

There are several reasons why migrants, especially long-distance ones, tend to take up a significant share of jobs in the secondary labor market. These include, on the demand side, the institutionally-determined granting of inferior legal and moral rights to (namely international) migrants, which facilitates their hiring and lay-off in a sector characterized by “flux and uncertainty” (Piore 1979:36). On the supply side, migrants are often overrepresented relative to indigenous workers because: i) the latter tend to reject secondary labor market jobs on the grounds of their working conditions and associated social status (a rejection made possible to many by the numerical requirements of the primary labor market), whereas migrants, at least short-term ones and those in the initial stages of their migration processes, are often ‘target-earners’ less concerned with their status in the host society; ii) in most cases, the substantive (pay and working) conditions available to migrants in the secondary labor markets are, in any case, significantly better than the ones to which they had access before migrating; and iii) advanced industrialized societies have, for various demographic, social and cultural reasons, largely exhausted the traditional sources of ‘lower-rank’ workers for the secondary labor market: women and youths (id ibid; Berger and Piore 1980; Massey et al. 1993).

In complementary fashion, those authors that have focused on the processes characteristic of migrant-producing areas have instead emphasized the fact that, rather than being a consequence of a lack of development, emigration from poorer areas is historically associated with the disruption brought about by development, and the emergence of capitalist relations, in those areas. Massey (1988), for example, argues that the processes of capital accumulation and modernization that constitute capitalist development, by necessarily entailing “three mutually reinforcing processes: the substitution of capital for labor, the privatization and consolidation of land-holding, and the creation of markets” (id ibid: 391), inevitably uproot large numbers of people from their traditional ways of life. It is these people, this author argues, that “constitute the source for the massive population movements that inevitably accompany development” (id ibid:384).

The distinction within the historical-structural perspective between authors focusing on migrant-sending and migrant-receiving regions is far from clear-cut, though: for example, Petras (1981) presents a world-systemic account that emphasizes the (rhythmic) historical and (hierarchical) geographical patterns that characterize the functioning of the global system of labor supply, whereas Sassen (1988 and 1991) has sought to bring the two aforementioned emphases together under a unified theoretical framework, while additionally highlighting the pivotal role played by “global cities” in the circulation of capital and labor. In any case, it is clear that the historical-structural approach to migration essentially constitutes a derivative application to the topic of migration of various Marxist and neo-Marxist insights on development. It explicitly dismisses as secondary the emphasis on individual migrant agency, considering instead that migration, as a social process, is fundamentally governed by, and best accounted for by reference to, broader political-economic processes at the structural level.

The New Economics of Labor Migration

The New Economics of Labor Migration (or NELM) emerged in the 1980s as an allegedly alternative theoretical framework for accounting for the determinants of migration. It purportedly sought to redress both the perceivably unsatisfactory character of the neoclassical theory of migration and the alleged lack of regard for human agency in historical-structural accounts. In so doing, it has been variously characterized as a “fundamental departure from past migration research” (Taylor 2001:181); a “fundamentally different theory of migration [...which constitutes...] the only migration theory that explicitly links the migration decision to the impacts of migration” (Hagen-Zanker 2008:13); and “a pluralist view on migration and development”, “inspired by Giddens’ (...) structuration theory”, which “sought to harmonize actor- and structure-oriented approaches” (De Haas 2010:241)—portrayals which we shall argue to largely miss the point. In any case, the theoretical insights of the NELM can, according to two of its major proponents (Stark and Bloom 1985: 173–6), be summarized under five main headings: i) the emphasis on relative deprivation as a determinant of migration; ii) the emphasis on the household as the relevant decision-making unit; iii) the emphasis on migration as a strategy to diversify risk and overcome market incompleteness; iv) the introduction of information-theoretical considerations in migration theory; and v) the interpretation of migration as a process of innovation adoption and diffusion.

The emphasis on relative deprivation as a determinant of migration was introduced by Stark (1984; see also Stark and Taylor 1989 and 1991). It rests on the hypothesis that potential migrants carry out interpersonal income comparisons with other people within their relevant social settings, and that it is these comparisons, along their wish to improve their relative positions within those settings, that constitute the relevant element in the decision-making process. This hypothesis constitutes an application to the field of migration of the theory of relative deprivation introduced by Stouffer et al. (1949, cit. in Stark and Taylor 1989), and it seeks to account for the fact that, in numerous empirical contexts, “migration rates are higher from villages where the distribution of income by size is more unequal” (Stark: 1984:475).

Also central to the NELM are the ideas that the relevant decision-making unit in the migration process is typically the household, rather than the individual, and that the decision by the household to have one of its members migrate to a different location can often be understood primarily as a way to hedge against risk and overcome market incompleteness (Katz and Stark 1986; Lauby and Stark 1988). In contrast to the case of the individual optimizing migrant assumed in the neoclassical microeconomic theory of migration, this theory of migrant agency holds that it is the household that optimizes, that there is an element of uncertainty and risk to future incomes (hence the household’s degree of risk aversion enters into the optimization exercise), and that migration by one or more members of the household often constitutes precisely a form of self-insurance against future income risk (particularly if the migrants’ future income is expected to be uncorrelated, or negatively correlated, with that of the rest of the household: Stark and Bloom 1985). In this way, it serves to hedge against, namely, the risks of crop failure, falling prices and unemployment (cf. Massey et al. 1993). Because these risks might alternatively be insured against through recourse to, respectively, crop insurance, futures markets and unemployment benefits, it follows that a greater propensity to migrate is to be expected from those areas where there is greater market incompleteness and lesser availability of formal and informal collective self-insurance schemes.

The incorporation of aspects such as incompleteness of information, risk, self-insurance or game-theoretical analyses of intra-household commitments (Stark and Bloom 1985:175) renders clear the information-theoretical character of the NELM. It also provides a theoretical framework in which to reframe the analysis of migration as a process of innovation adoption and diffusion. Thus, the speed of diffusion of the decision to migrate as an innovation, from the innovators and early adopters in a community through to the late majority and laggards (as per Rogers 1962), is in the NELM framework a function of the interaction between the risk-aversion properties of the potentially adopting households’ utility functions and the extent to which information conveyed by previous migrants reduces the uncertainty surrounding the migration option itself, conditioned by market incompleteness and the overall income distribution at the origin (Stark and Bloom 1985).

The reasons why the NELM has been portrayed by some as a successful attempt to overcome the alleged insufficiencies of the neoclassical and historical-structural accounts consist of: i) on the one hand, its greater sophistication and adherence to reality as a theory of migrant agency when compared to the neoclassical theory; and ii) on the other hand, the way in which it supposedly achieves a better balance between agency and structure than do the historical-structural perspectives. The next section argues that while the first of these two considerations is to a certain extent accurate, the second misses the point entirely, and that the NELM is in fact best regarded as a ‘reworked’ avatar of the neoclassical theoretical account of migration.

The New Economics of Labor Migration: Beware of Neoclassicals Bearing Gifts

The NELM was essentially a theoretical development from within the neoclassical paradigm and it arose out of a generalized lack of satisfaction with simpler models like Todaro’s. True, reference to Todaro (1969, 1976) or Harris and Todaro (1970) continued to be mandatory when mustering the theory, but it became increasingly apparent to anyone in the field of migration studies that models based on perfect rationality, optimization and methodological individualism in their extreme form were not only at odds with the ethnographic evidence with respect to the behavior of migrant and non-migrant households, but also singularly inept at explaining macro outcomes. In particular, why were migration flows in the real world nowhere near to exhibiting the strength and fluidity that one might reasonably expect given the extant absolute income differentials? The neoclassical approach was notably poor in explaining or predicting where and when migration occurred, but it was even more unsatisfactory in explaining where, when and why it did not occur.

The NELM is sometimes most associated with its call for shifting the locus of migration agency from the individual to the household, but in reality what is perhaps most fundamental about it is its information-theoretic character: migration is regarded as a rational way of hedging against risk in a world characterized by incompleteness of information and incomplete markets. It is not that migrants, particularly those in or from developing countries, behave any differently from other rational optimizers—no need for a distinctive microeconomics of migration and/or development whatsoever, nor to shed optimization and rationality as driving principles—, it is just that they are constrained by lack of information and market incompleteness. The idea of relative deprivation is itself also based on information: the income distribution based on which the implicit migration decision-making statistic is produced, and particularly its wider or narrower scope (namely, is the relevant reference group the village, the region, the country...?) is a function of the characteristics of the ‘information field’ accessed and mobilized by the potentially migrant household. Finally, the NELM explicitly accounts for migrant remittances, for which the basic neoclassical theoretical approach had no place.

Thus did the NELM indeed seem to bear gifts to those unsatisfied with the simpler and more extreme forms of the neoclassical paradigm: it acknowledged the anthropological importance of household-level decision-making dynamics (though it did not go to any lengths to open the ‘black box’ of those dynamics); it significantly relaxed the perfect-information assumption that was so clearly at odds with the reality of information flows shaping and reshaping migration systems through migrant networks and other sub-systems; and it allowed for a more satisfying explanation of macro outcomes (namely, the predominance of immobility and migration not hydraulically leading to worldwide factor equalization).

However, did it, as De Haas (2010:241) puts it, constitute a new, “pluralist” approach that reconciled agency and structure in a fundamentally new way, namely one akin to Giddens’ theory of structuration (which, it might be worth recalling, essentially aimed at rescuing structure through stressing its relative indeterminacy)? Certainly not. Like the ‘old’ neoclassical theory, and unlike the historical-structural perspective, the NELM remained resolutely methodologically-individualist, although the relevant ‘individuals’ are in this case households, rather than persons. Crucially, no allowance is made for dynamics or constraints at the structural level (with the exception of market incompleteness) shaping migration dynamics, nor for individual migrant agency interacting with that structural level in such a way as to be constrained by it, but sometimes give rise to changes to it, as in the theory of structuration. Nor did it in any way allow for migration to be analytically linked with broader processes of social transformation such as proletarianization and capital accumulation. The NELM is in fact as intrinsically built on methodological individualism and optimizing rationality as the ‘old’ neoclassical theory: it just moves the latter’s schizophrenic worldview a step closer to reality by allowing for ex-ante (and unaccounted for) market incompleteness; by positing the second-order rationality of institutions (namely, the institution of migration) in such a context; and by entrusting the household with decision-making privileges. Incidentally but crucially, it does the latter in a way that completely disregards both conflicts of interest within the household in general, and patriarchal power relations in particular—a lack of regard that has been thoroughly criticized for example by feminist critiques of economics (e.g. Kramarae and Spender 2000). Thus, in its inability to shed what is most central to neoclassical economics—and most inimical to a realist view of the world—, the NELM is simply “old wine in new bottles”, as argued by Fine (2006) with respect to other instances of the same trend across economics, namely the post-Washington Consensus and the new trade economics.

That such a way of seeking to rescue the fundamental tenets of the neoclassical theory of migration would arise was probably inevitable. It emerged at a time (the 1980s) when the information-theoretic revolution was sweeping across the economics discipline, as argued by Stiglitz in the following way (1994, cit. in Fine 2006:8):

“During the past fifteen years, a new paradigm, sometimes referred to as the information-theoretic approach to economics ... has developed ... . This paradigm has already provided us with insights into development economics and macroeconomics. It has provided us with a new welfare economics, a new theory of the firm, and a new understanding of the role and functioning of financial markets.”

Stiglitz does not mention, but certainly could have, that it also provided a new theoretical approach to migration under the guise of the NELM. What is interesting to see is the extent to which Fine’s (id ibid) criticism of the “new paradigm” applies equally well to the NELM:

“(...) the analysis is extended, as before, to non-economic factors, but in ways that are liable to be more palatable both to non-economists and opponents of neo-liberalism. (...) the market imperfection approach is better able to claim some hold on the realities of contemporary capitalism. (...Yet) despite all this, the theory continues to rest on the assumption of homo economicus, only one whose rationality is bounded both by imperfect information and historically-evolved non-market factors”.

It is also interesting to look at this particular development in migration theory from the broader perspective of economics imperialism, or the colonization by neoclassical economics of subject matters previously left to other social sciences. By virtue of its own historical trajectory and the objective characteristics of its subject matter, migration studies has traditionally been a field in which neoclassical economists were never in a position of dominance. Even though, for lack of other ones, the dominant models of the determinants of migration have always been mainly economic models, the truth is that most sociologists, anthropologists, geographers or others dealing with migration never had much time or patience for the gross over-simplifications and the level of detachment from reality of the ‘old’ neoclassical theory. Instead, they would much more quickly concede, with Massey et al. (1998:281), that “the leading theoretical treatment of the forces that promote emigration from developing countries is world-systems theory”, or the historical-structural approach more generally. Thus, migration could (and can) be regarded as an active and important front in the war waged by economics’ imperialism—and the NELM represented the deployment in this particular front of the same weapon (information-theoretic economics) being used elsewhere in this war.

It is hard to assess the extent to which this strategy was successful in this particular instance. Within the field of migration studies, it is likely that the historical-structural perspective is still dominant, at least among its most reputed authors—again, a fact to which the prevalence of social scientists other than economists is not alien. However, replacement of the (Harris-)Todaro models with the NELM as the leading theoretical explanation of migration among those endorsing a methodologically-individualist, rational-choice stance (despite the general lack of acknowledgement of the fact that this is what it essentially constitutes) has enabled the latter to stay afloat in view of what seemed to be an irrevocably discredited position. Wars, including the one being waged by economics imperialism, are not always made up of victories—in some battles, they involve damage limitation.

Towards a New Historical-Structural Synthesis

This paper puts forth the argument that the neoclassical theory of migration in any of its avatars is over-simplifying, unrealistic, and incapable of allowing us to come to terms with the reality of past and present migration flows. Like with many other social processes, the most relevant determinants, trends and constraints to migration are to be found at the structural level, and cannot be derived through aggregation—hence the theoretical superiority of the historical-structural approach. However, the latter is arguably in need of a new and improved synthesis, given: i) it being scattered throughout a variety of partial models and explanations with emphases on different elements of the same process; ii) its neglect of relevant insights from “alternative” theories that are in fact logically consistent with the historical-structural perspective; and iii) its neglect of crucial insights from classical Marxist theory, by virtue of its predominantly neo-Marxist (dependency and world-systems) influences.

This section briefly outlines what such a new and improved historical-structural synthesis might look like. First of all, it would acknowledge that the study of actually-existing migration flows must rely on the historical method, rather than on the deduction of outcomes from hypothesized assumptions regarding individual behavior. This is also a direct consequence of the fact that many of the central processes involved in migration emerge at the structural level. Then, even the most casual glance at the historical evidence leads to the conclusion that migration in the past two to three centuries has been inextricably linked to the emergence, development and expansion of capitalism: at the local level, under the guise of the massively revolutionary process of urbanization; at the global level, and especially in the post-WWII period, through the predominance of the flows from the ‘labor frontier’ to the various ‘cores’ (Skeldon 1997). If we take national states as our units of analysis, we additionally find that these processes take place in rather homologous ways from one state to the next, subject to their specific position in the hierarchy of the world system. Thus, a number of insights from the mobility transition approach, most commonly associated with Zelinsky (1971) but especially in the version put forth by Skeldon (1997), can certainly be fruitfully incorporated into a historical-structural synthesis.

It is our contention that labor migration is best regarded as a global and multi-scalar system of supply of labor-power that makes it possible to meet the requirements of capital accumulation, for the most part in spatially concentrated fashion (in cities, macro-regional centers, the global ‘core’). The specific ways in which these ‘abstract’ requirements are met depend on a series of concrete and complex circumstances, which have been quite aptly systematized by the systemic and network approaches (Mabogunje 1970; Fawcett 1989; Gurak and Caces 1992; Kritz and Zlotnik 1992). Here, in fact, the latter two approaches add significantly to the explanatory power of the accounts of some historical-structural authors, whose emphasis on labor flowing in the opposite direction of capital is perhaps excessive. True, the historical evidence seems to indicate that the disruption brought about by the emergence of capitalist relations is likely the most important cause behind the creation of potentially migratory populations. However, the emergence of capitalist social relations of production is not a simple function of the exogenous inflows of capital (Brenner 1977). Moreover, other factors and ‘systems’ come into play in determining the emergence, scale and direction of concrete migration streams, such as language and cultural factors, the relative autonomy of the state system, or the ways in which the individual agency of migrants, especially through migration networks, can both consolidate existing structures and, at certain times, introduce limited but relevant changes to them.

In order to arrive at a more fully comprehensive theory of labor migration (meaning one that accounts for, and analytically links, the micro and the macro as well as agency and structure), the final component missing would then be a theory of individual migrant agency. Of course, this becomes a relatively secondary issue as soon as one acknowledges that individual behavior is largely constituted, constrained and overridden by structural factors. What is more, it is also a sub-component with respect to which one can only expect further relevant insights to emerge through further detailed and methodologically sound ethnographic research, and in which there is no reason to expect that a universally valid ‘hierarchy of individual determinants’ can be identified. It is quite obvious that expected future income usually plays into the decision-making process, as does hedging against risk, relative deprivation, and the introduction of sudden disruptive changes in terms of livelihood. It is also clear that the locus of decision-making is usually a complex one, in which individual aspirations and decisions interact with household and community dynamics. However, one must be careful to also take account of more subtle issues of, for example, social status, tradition (e.g. temporary migration as rite of passage), or curiosity about the world. Overall, then, when it comes to the specific issue of migrant agency, it seems that Lee’s ‘push-pull’ formulation, despite being very much a truism, certainly constitutes a better starting point than the oversimplifications of the neoclassical ‘micro’ model.

Another aspect in which the historical-structural is arguably in need of theoretical improvement has to do with its treatment of the consequences of migration for the areas of origin of the migrants—a topic (“the migration-development nexus”) which has been receiving growing attention by virtue of the contrasting effects of the “brain drain”, on the one hand, and ever-increasing worldwide remittance flows, on the other. Largely due to the central and lasting influence upon many historical-structural authors of the neo-Marxist (namely dependency and world-systems) lines of thought, the historical-structural perspective has arguably failed to address this issue in a sufficiently complex and dialectical manner. Instead, the tendency has been to assume more or less automatically (e.g. in Petras 1981 and Sassen 1988) that, by virtue of the unfavorable hierarchical position in which migrant-producing areas participate in the global system of labor supply, that participation must inevitably be deleterious. Basically, migration is regarded as constituting yet another way through which “the surplus is siphoned off from the periphery to the core” (Papademetriou and Martin 1991:ix)—the “brain drain” providing sufficient evidence of this. However, assuming that the impacts of migration upon capitalist development in the areas of origin of the migrants are a direct function of the ‘net value transfers’ associated with those exchanges amounts to incurring in what Brenner (1977) has dubbed “neo-Smithian Marxism”: the idea that each and every social formation will put the surplus produced and extracted in it to an ‘optimal’ use in terms of capital accumulation, regardless of the extant social relations of production. The historical-structural perspective would therefore be well advised to do without this simplistic understanding of (capitalist) economic development, and to seek to come to terms with the complex reality of the interaction between migration and development in each particular context through a more classically Marxist focus on the social relations of production.

Conclusions

Until the 1980s, the field of migration theory was dominated by the rivalry between the neoclassical theory(ies) of migration and the historical-structural theoretical accounts. Arguably due both to its inherent explanatory superiority and to the prevalence in this field of a majority of social scientists other than economists, the latter gradually came to occupy the main role as the leading theoretical explanation of this phenomenon. Then, in the 1980s, a third alternative theoretical approach came forth: the New Economics of Labor Migration. It purportedly sought to occupy a sensible and realistic middle ground between the structural emphasis of the historical-structural perspective and the agency orientation of the neoclassical theory, and was hailed by many for succeeding to do so.

Yet, this paper has argued that the NELM has in fact been little more than an avatar of the neoclassical approach in which only marginal concessions and changes were made, while the core (rationality, methodological individualism, lack of regard for structural trends and constraints) remained untouched. That this new avatar arose in the 1980s and assumed a largely information-theoretic form must inevitably be linked to the processes then taking place more generally in the relationship between mainstream economics and the other social sciences. Indeed, we have argued here that, in the specific case of migration theory, the NELM constituted the way in which neoclassical economics, unable to attain yet another conquest for economics imperialism, at least sought to ensure damage limitation.

The NELM did, and does, indeed bear some gifts to migration scholars unsatisfied with the ‘standard’ neoclassical approach, but it must still be rejected as theoretically unsatisfactory. The main theoretical alternative—i.e. the historical-structural approach, with its Marxist theoretical underpinnings (albeit at times loosely so)—constitutes a much more powerful instrument for grasping the reality of past and present migration flows and systems. Yet, this paper has argued that it, too, is in need of a new and improved synthesis.

Such a new and improved historical-structural approach would bring together the insights from those historical-structural authors (e.g. Piore, Castles and Kosack, Nikolinakos) who focused on the structural demand for immigrant labor in the geographical loci of capital accumulation, as well as from those (e.g. Sassen, Massey) who highlighted how the mass creation of migration-prone populations is linked to the social disruption linked to the introduction of capitalist social relations of production in previously non-capitalist contexts (that is to say, migration largely constitutes a geographical manifestation of proletarianization and capitalist development). Such a synthesis would additionally acknowledge the empirical regularities highlighted by the largely a-theoretical mobility transition approach (Zelinsky, Skeldon) and, most importantly, it might seek to bridge the micro-macro gap through consideration of the increasingly complex and concrete phenomena emphasized by the migration systems and migration networks approaches (e.g. Mabogunje, Fawcett, Gurak and Caces). Finally, it would shed its “neo-Smithian” neo-Marxist components, most visible and harmful in the treatment of the consequences of migration upon capitalist development in the areas of origin of the migrants, and replace them with a more subtle (and dialectical) attention to the interrelationships between migration, the social relations of production and capital accumulation.