This was Font/Pitch 1,10 - Off.This was Font/Pitch 2,12 - On.                               

Chapter 14 of C. Chase-Dunn, Global Formation: Structures of the World-Economy, 1998 Lanham, MD: Rowman and Littlefield

                      Theory Construction

    Recent attacks upon structuralism and formal theorizing in the social sciences have occurred within the context of a global ideological shift toward the right.  The renewal of nineteenth-century glorifications of the free market and entrepreneurial profits by neo‑conservatives has been surprising, but even more curious is the Left's drift to the center.  This has occurred in many forms ─ the emphasis on modernizing the state sector and increasing its productivity, the shift toward markets and profit incentives in the "socialist" states, and Eurocommunist support for nationalism, state capitalism, and austerity.  We know that academia and world politics are not separate islands, and so it is tempting to assert a connection between the renewed emphasis on methodological individualism and the shift away from analyses of structural and institutional forces in both political ideology and social science. 

    Exactly what this connection is I cannot say.  The Zeitgeist seems to affect us all, and all of our activities.  I am not arguing that all those who have attacked Marxist structuralism are tools of the capitalists.  Rather I want to make the case that the building of a formal structural theory of capitalist accumulation remains an important goal for social science and a potentially valuable contribution to the creation of a more humane, egalitari­an, and balanced world society, and one that is less likely to send our collective experiment with life up in smoke.

    In my view the structuralism of Louis Althusser and his students has been attacked on the wrong grounds.  The effort to formulate a structural theory of capitalist development, now understood at the world‑system level, is a necessary effort if we are to collectively get control of this headless horseman, this driverless (or multidrivered) behemoth which is almost certainly hurtling us all toward the precipice.

 

A Defense of Theory

 

E. P. Thompson (1978) has made perhaps the most influential attack on  Althusserian structuralist theory in his essay "The Poverty of Theory."  Thompson argues that men and women are more than simply occupants of struc­tural positions, merely agents of social forces.  They are historical actors who actively negotiate and struggle to create their own lives and to change (or maintain) existing social institutions.  A structuralist theory is condemned because it implies a mechanical view of human beings hopelessly caught and scuttled along by social forces beyond their control.  This debilitates purposive action and promotes fatalism.  Structural theories are also themselves ideological weapons by which power is legitimated, and this power is used to oppress and exploit people.  Stalinism and the Third International are suggested as a pertinent example.

    As a substitute for structuralism Thompson puts forth his own method of historical analysis that emphasizes the authorship of culture by individuals and classes engaging in struggles to survive and create a better world.  The historical contingency of outcomes is emphasized, as are the elements of consciousness and intentional organization.

    Thompson has been joined by a large number of other academic Marxists who have made similar arguments with respect to particular subject areas, or in general philosophical and methodological terms.  The critique of Marxist structuralism has succeeded in mobilizing a broad rejection in favor of historicist, voluntarist, particularistic (area or local) studies.  The growing popularity of the deconstructionist critique, which demolishes all theories as textual tools of political power, is also quite evident.1

    Here I will formulate my own critique of Althusserian structuralism, and I will argue for a renewed effort to build a formal theory of the deep structure of capitalism, rather than a rejection of all theories.

 

The Historicist/Structuralist Continuum

 

First I want to pose the existence of a continuum between two extreme metatheo­retical positions in the philosophy of social science and locate various intermediate positions on that continuum.  The end points are what have been called nomothetic versus ideographic analysis.  There is a long history of contention within all the social science disciplines between these two very different stances.  Indeed many of the disciplines exhibit a somewhat cyclical pattern of variation in terms of the popularity of approaches along this continuum (e.g. Harris, 1968).  Nomothetic analysis attempts to formulate general laws that explain the regularities, patterns, and forms of change exhibited by a phenomenon.  In social science the most completely nomothetic formulations assert that a single ahistorical model can account for all human social systems, large or small, primitive, ancient, or modern.  Talcott Parsons and his followers represent this position, contending that the idea of social structure (composed of normatively defined statuses and relations) and a list of systemic necessities can be usefully applied to two‑person dyads, small groups, organizations, national societies, and global systems.

 

    Ideographic analysis, the other end of the continuum, focuses on what is unique about a person, a time period, or a locale.  Rather than asking what cases have in common, it stresses their differences.  It paints in rich detail, endeavoring to evince the mentality, both cognitive and affective,of a historical setting.  This is what I am calling historicism.2

 

     While this effort constitutes a valid and valuable exercise in the humanities, many of its proponents assert that a more generalizing approach imposes a false philosophy of physical science on human beings.  Beings which can alter their own behavior intelligently and in ways difficult to predict are not billiard balls, it is contended.  Social scientists respond that even "thick description" implicitly makes comparisons and utilizes generalizations in order to make its narrative statements intelligible.3

 

     Perry Anderson (1980) has formulated a valuable response to E. P. Thompson's attack on structuralist theory which rescues much of what is valuable in the Althusserian apparatus while acknowledging the dialectical  nature of structural determination and voluntaristic action.  Anthony Giddens (1979) has addressed this issue in great detail and his work is perhaps the most systematic effort to resolve the problem, but one quickly tires of the discussion of structure, agency, and action at a purely abstract level.  I am not agreeing with the historicists that every theory or piece of research must contain people, places, and events, as Charles Tilly (1984) has argued, but the absence of a specified context behind Giddens's discussion leaves one wonder­ing.  The discussion of structures, agency, power, ideology, and social change at such a completely abstract and general level sounds like so much talk.  But perhaps I am only revealing my own predilection for a stance near the middle of the continuum. 

    In between the extreme poles of total generalization and pure description are an infinite number of possible combinations of the two, not only with the quantitative mix varying, but also with the scope and nature of content assigned in different ways.  Many sociologists have argued that "grand theory" is vacuous (e.g. C. Wright Mills, 1959).  On the other hand, "theories of the middle range" applied to particular contexts are alleged to be more scientifi­cally valid and socially useful (Merton, 1957).

    Fernando Henrique Cardoso and Enzo Faletto (1979) introduce their important study of dependency in Latin America with a description of their "historical─structural" method, which emphasizes the differences among several qualitative types of dependency situations and the possibilities for maneuver within the structural constraints which emanate from the core (see also Bennett and Sharpe, 1985:9‑13).

     Immanuel Wallerstein has characterized his focal unit of analysis as "historical systems," a term which neatly captures the dialectical antinomy  between ideographic and nomothetic analyses.  Wallerstein's approach emphasizes the interaction between the historicity of socio‑economic systems and their deep structural or essential elements.  In this he is similar to Marx.  Marx criticized the ahistorical generalizations of classical political economy which ignored the unique qualities of different modes of production.  Assump­tions about timeless human nature, he pointed out, obscure the social origins of institutions and the qualitative transformations which occur in the development of human societies.  Rather than attempting to model all socio‑

ec­onomic systems, Marx focused on capitalism, an historical mode of production with its own unique logic of development and contradictory tendencies.  But whereas Marx attempted to formulate his deep structural model explicitly in the volumes of Capital, Wallerstein's remarks about theory and his somewhat casual approach to theoretical specification reveal a semi‑phenomenological attitude.  The statement of editorial policy in the Braudel Center's journal, Review, refers to "the transitory (hueristic) nature of theories."  Wallerstein's penchant for narrative first and theoreti­cal discussion second, and his ambivalent and contradictory record regarding the definition of his own concepts, place him further toward the historicist end of the continuum than Marx or Althusser.

     Althusser and his followers have been criticized for excessive struc­turalism and the allegedly  associated sins listed in the discussion of E. P. Thompson above.  But I do not think these are the main problems with Althus­ser's approach.  A structural theory of the tendential laws of capitali­sm, rather than implying a deterministic universe in which political action is necessarily futile, is  rather a guide to the weak links and openings for positive socialist politics, or it ought to be.  There is no contradiction here because no one claims that all social action is determined by structural forces.  A Marxist structural theory, as Engels (1935) long ago pointed out, is an effort to say how historical forces are moving and to indicate what is possible for "scientific socialism" within that context.  Both structural theories and historical accounts can be used to legitimate political power.  A good (or true) theory can be used for bad ends, but this possibility is not an argument for ripping down all theories.

 

    Althusser and Balibar's approach (Balibar, 1970)is valuable precisely because they make the distinction between:

    1 the mode of production ─ an abstract level of analysis which specifies essential structural tendencies of a  socio‑economic system, and

    2 the level of the social formation ─ a concrete and directly observable level of historical events and social institutions which may contain more than one mode  of production as well as more purely conjunctural features which combine to determine historical events.

The distinction between the deep structure and the conjunctural level of historical events is contained within the overall analysis.  Althusser and his students are more explicit about the content of the deep structural level, and they put more emphasis on its causal importance than does Wallerstein or other scholars who are closer to the historicist end of the continuum. 

    Of course it is not only a matter of the relative emphasis of structure and conjuncture, but also the substantive content of the distinction.  Among the structures, Parsons makes norms and values the master variables.  Exactly how one understands the distinction between "base" and "superstructure" (or between essence and epiphenomena) is of great importance to the substantive content of a theory.  In chapter 1 I have proposed a reformulation of Marx's model of capitalist accumulation which not only changes the framework of analysis to the level of world‑systems, but also argues that certain elements such as state‑formation, nation‑building, and class formation (which Marx consigned to the conjunctural) should instead be incorporated into the model of deep structure.  Here I am not defending those substantive theoretical decisions, but rather the prior decision to continue the project to produce a  structural theory.

    My own criticism of Althusser et al. focuses, not on the structuralist project but on the content of the theorization (see chapter 1), and also on the failure of the Althusserians to concern themselves with the confrontation between theoretical formulation and empirical research.  The philosophy of praxis employed by the Althusserians defines the confronta­tion with the empirical world in terms of political activity.  I do not agree that political practice is a substitute for systematic comparative research designed to distinguish among contending theoretical formulations.  Rejecting systematic comparative research as a bourgeois method, the Althusserians became mired in a scholastic world of textual interpretation, rationalistic deductive analysis, and political debate.  Many of those Marxists who were concerned with the real world of ongoing social movements outside of the immediate context of French political debates turned away from structuralism.  Thus Manuel Castells, himself an early proponent of Althusserian structuralism, embraced historicism in the form of a kind of populist romanticism (see Molotch, 1984), throwing out theory with the bath. 

    My defense of comparative empirical analysis is contained in the  following chapter.  Here I only want to point out that deductive reasoning, the logical exposition and critique of theoretical concepts and  propositions, is only half of the process of scientific theory production.  The other half is inductive empirical research.  The central tendency of American sociology has been the opposite error ─ research without theory.  As Arthur Stinchcombe (1968) and many others have pointed out, research can only distinguish between theories which predict different things, and so it is important to formulate contending theories in ways which make it clear what they do and do not imply

about the empirical world.4  Admittedly this demanding ideal of how theory construction ought to proceed is rarely followed, and I have not been able to completely follow it in this book.  Nevertheless, archaic as it may seem, it is one of the main justifications I offer for my theoretical effort, and for the review of comparative research results.

 

Ontology

 

In many discussions with students and colleagues in sociology I have discovered the curious assumption of a connection between spatial scale and the level of abstraction.  Many seem to assume that the world‑system is abstract, whereas an individual person is concrete.  I have also observed this error in the published works of distinguished social scientists (e.g. Tilly, 1984:14).  But on reflection everyone will admit that there is no  necessary connection between size and abstraction.  The sun is not more abstract than the earth.  I am not more abstract than an ant.  The world‑system is not directly observable to the eye, and this may be part of the reason why some people think of it as more abstract than smaller levels of  analysis.  But neither is the earth visible as a whole to most eyes.  And yet we would all agree that it, and many other things such as atoms and the solar system, are concrete entities.  The psychology of visual perception has established that we employ ideas even in our perception of the immediately visible.  But this does not mean that everything is equally abstract.  Though I need the concept "chair" to see the chair, I and most philosophers of knowledge believe that a chair is more concrete than is (say) beauty.

    In social science most of us believe that individuals really exist and that they are more concrete than classes or nations.  Since Lukács we have been cautioned to beware the reification of society.5  But John W. Meyer, a consumate structuralist, has pointed out that we more often reify the in­dividual.  As we all know from our undergraduate courses, the self is socially constructed and biological individuals are conceptualized very differently in different kinds of societies.  But let us assume that in­dividuals exist.  It does not follow that larger levels of analysis are less concrete.  At the most concrete level the world‑system is composed of all (or nearly all) of the people on earth, and the material interconnections (direct and indirect) among them.  It is large, but it is not abstract, or not more abstract than other, smaller, objects of social science analysis.

    But scale does affect the patterns that are observable.  When we use a telescope we see very different things than when we use the naked eye or a micro­scope, even when these are pointed at the same "reality."  There is little point in arguing about which are the real concrete patterns or objects.  Rather, if we are writing history we may choose to focus on one or another scale of analysis for its own sake, as we choose between different aesthetic styles.  But if we are making science we will wish to understand the causal processes by which patterns change, and we will ask which level of analysis accounts for more variation in a designated outcome variable (explanandum).  This last effort may require the study of different levels of analysis simultaneously, a matter that is discussed in the next chapter.

 

Models of the World‑system

 

World modeling is an enterprise which has proceeded largely unconnected with the world‑system perspective.  Here we will comment on the possible relations between these two projects and designate three types of models which are part of the effort to construct a theory of world‑systems.

    World modeling is a data‑gathering and simulations effort by different groups of scholars.  Much of this work is valuable because it focuses on the world economy as a whole as the unit of analysis and it uses empirical data to forecast trends.  The theoretical models used to created the simulations have varied greatly from project to  project, but none of the main projects has utilized the kind of theory developed in this book, a theory which uses Marx's  accumulation model of capitalist development as its starting point.  I would concur with Patrick McGowan (1980) that, despite the very different paradigms employed by world modelers and the world‑system perspective, these two projects can be useful to one another.

 

Levels of Specification

 

I would like to distinguish between three levels at which we may specify models of the modern world‑system.  The first can be called a descriptive model.  This type of model specifies the relationships in time between the several cycles and trends which are features of the world‑system that vary over time.  Such a model is implied by the discussion of cycles and trends in chapter 2 and several authors have presented such temporal models (e.g. Chase-Dunn, 1978:170; Hopkins and Wallerstein, 1979:496-7).  This sort of model does not specify causal relations among variable features.  Rather it simply predicts regular

temporal relations among different variables.  This descriptive level of analysis is rarely ever presented without some discussion of causal relations.  It is a valuable theoretical effort in its own right to make explicit hypothe­ses about temporal regularities because it facilitates the posing of empirical questions.

    The second type of model is a specification of the causal relations among variables which are features of the whole world‑system or features of subunits, such as zones, nation‑states, etc.  An example of such a model is given in figure 13.2 and a series of similar examples are contained in McGowan's (1985) critique of Bergesen and Schoenberg's (1980) study of cycles of colonialism.  Procedures for testing these causal models are discussed in the following chapter.  Their value is that they make explicit our arguments about what causes what, and therefore they can be helpful in evaluating different theoretical arguments about processes of the world‑system. 

    A third type of theoretical specification is one which formulates a  theory of the deep structure or the main engine which drives the world‑system.

Marx's theory of the capitalist mode of production is such a model.  It posits the existence of structural tendencies which can account for the long-run dynamics of growth and reproduction of a socio‑economic system.  The specifica­tion of such a deep structural model may be formalized in a number of different ways.  Part of Marx's model has been formalized as an axiomatic theory of logically related statements by Nowak (1971). Morishima (1973) has mathematically specified major aspects of Marx's accumulation model.  Models of the deep structure need not be completely formalized, but formalization makes assumptions clearer and makes it easier to see the empirical implications of a set of central theoretical statements.

    It should be pointed out that the language of deep structure versus surface‑level appearances does not require the nominalist philosophy of Hegelian idealism.  We need not assume that there is "really" an unobservable essence beneath the complexities of empirical appearances.  The model of the deep structure is like a map.  It is a simplification of the territory that helps us get where we want to go, to explain and predict as much as possible.

 

 It is desirable that the map be as simple as it can be, while still remaining helpful.  Here again we find the continuum between historicism and completely ahistorical theory.  Historicism copies the territory in rich detail, without an attempt to simplify.  On the other hand, the map of completely general ahistorical theory is so simple that only the most analytic features are drawn, e.g. the grid of longitude and latitude.  These features will provide a rough guide for all locations, but will not provide sufficient information to be helpful for most particular purposes.6  Thus it is not so much a question of the real existence of the deep structure but rather the usefulness of the model for explanation, prediction, and action.  And in this sense there are varying degrees of truthfulness rather than an absolute truth.

 

Concept Formation

 

Cardoso (1977) argues that concepts should not be rigidly defined and/or converted into one‑dimensional variables, because the social reality being studied is a dynamic, contradictory reality which is oversimplified by such precision.  This objection may be partly based on Marx's notion that theoreti­cal concepts in the social sciences should be reflective of the relational and contradictory character of social processes themselves (Marx, 1973).  I think that this is an important methodological idea, but it should not prevent us from making tentative definitions clear in order to see how they may be useful in explaining empirical reality.  Operationalizing a concept does not permanently commit us to either the definition or the particular indicator we employ to measure it.  Cardoso is right to point out that we should be aware of the assumptions behind converting a concept like dependency into a one‑­dimensional variable.  Indeed, the evidence of cross‑national research confirms that dependency is a multidimensional phenomenon (see chapter 9).

    Much of the concern about the implicitly static and mechanistic nature of formal causal models (e.g. Bach, 1977) seems to be the result of a misundersta­nding of the logic of causal analysis.  All non‑experimental research is an attempt to infer underlying causal processes from data over which we have little control.  The type of variables used (qualitative, metric, linear or curvilinear, multidimensional or not) and the logic and nature of the causal relations that are implied in a particular model are up to the theoretical imagination of the researcher.  It is true that the most commonly tested models assume one‑way causation and linear relations among variables, but these assumptions are by no means necessary to causal modeling.

 

Dialectics and Contradiction

 

Cardoso (1977) and Bach (1977) argue that conventional causal imagery should not be applied to dependency and world‑system processes because these  processes are dialectical and contradictory.  Presumably these authors are making a claim about objective reality; if we can be clear about what is meant by a dialectical process, there is no inherent reason why a dialectical model cannot be specified and tested.

    Many students of social structure prefer to use dialectics as an heuristic aid to thinking about processes of social change.  As such, the general notions of contradiction, opposition, and qualitative transformation can be quite useful for interpreting complex historical situations, and this heuristic aid is not at all incompatible with causal propositions of a more conventional kind.  When we assert that there is a negative causal relationship between, for example, dependence on foreign investment and economic growth (Bornschier and Chase‑Dunn, 1985), we do not deny the interactive and reactive nature of the relationships between transnational corporations, peripheral states, and peripheral workers.  What we are asserting is that on the whole, over many cases, in the long run, ceteris paribus, the more dependent a country is on foreign capital the slower it is likely to develop economically.  The

pos­sibility that there may be exceptions, or that some countries may be able to successfully combine foreign investment with a certain kind of growth, does not disprove the general contention.   Proposi­tions of this kind can easily be combined with a dialectical heuristic.

    More problematic is the specification of formal dialectical propositions within a model.  If we contend that a particular process or relationship is dialectical it is possible to formally specify the meaning of this assertion.  Although it is unpalatable to many dialecticians, we can translate the notion of contradiction into conventional causal logic.  Thus contradiction may be understood as:

    ─ causal vectors that affect a dependent variable in opposite ways;

    ─ variables that affect one another (reciprocal causation); or

    ─ a variable that negatively affects itself over time (negative   feed­back). 

It is true that most causal models in social research employ only fairly  simple assumptions about feedback, reciprocal causation and interaction, but much more complex forms of causation have been successfully modeled with  fairly standard mathematical tools. 

    Most serious dialecticians are not satisfied with such a simple trans- lation of the central notions of dialects (contradiction, opposition, and  qualitative transformation) into conventional causal logic, however.  They argue that Aristotelian logic precludes the simultaneous existence of  opposites, so a new logic and mathematics is required.  Work has begun to create a formalized dialectical logic and mathematics (Rescher, 1977; Alker, 1982). This type of work may eventually allow us to construct dialectical models that clearly specify the meaning of opposition, contradiction, syn­thesis, and qualitative transformation in ways that make them empirically disconfir­mable.  This line of work was suggested by Erik Wright (1978:chapter 1). He elaborated a number of "modes of determination" which are particularly appropriate to the task of modeling the causal structures of Marxian theory.

  

Critical Theory, Determinism and Political Practice

 

Structuralist theory, formal modeling, and quantitative comparative studies have been attacked by those who argue that dependency and world‑system studies should expose and condemn the structures of domina­tion.  Critical theory and deconstructionism want to expose positive social science as bourgeois ideology, and the argument is made that both the philoso­phy of knowledge and the comparative methodology used in normal social science are actually class‑based mystifications.  Cardoso (1977) contends that quantitative comparative research is necessarily uncritical because it must wear the mantle of "value‑free" objective science.

    My general defense of quantitative comparative methods is contained in  the next chapter.  Here I want to argue for the political value of a renewed effort in the realm of structural theory.  If we already had an adequate  theory of capitalist accumulation and sufficient understanding of the dynamics of world‑system processes and the nature of the transformation of modes of production, then we could go on to particular applications of this knowledge in social struggles.  But the current "crisis of Marxism" reveals that certain basic problems about the nature of capitalism and socialism have not yet been resolved.  That is why it is politically sensible to exert effort toward the reformulation of a positive theory of capitalist accumulation.  If we could simply accept Marx's model and apply it to the world of the twentieth century then we would not need to produce a new (or revised) theory.  A structural theory, however, is more than a call to action.  It is a model of how the socio‑economic system actually works, its dynamics of growth and competition, and the inherent contradictions which produce pressures for transformation.

    Bach (1977) and Cardoso (1977) argue that the construction of formal models implies a deterministic view of development which ignores the voluntary efforts of individuals and classes to resist and to alter the structures that are exploiting and underdeveloping them.  On the contrary, models in social science are most usually probabilistic rather than deterministic, thus to allow for the complexity and indeterminacy of human behavior.  Knowledge of the likelihood of a social outcome given certain conditions is a potential contributor to human freedom.  We do not become more determined by under-

stand­ing the laws of nature or the tendencies of social systems.  On the contrary, we become freer.  For example, models of dependency that show the size and nature of average effects on a national economy caused by an increase (or decrease) in dependence on foreign investment can be useful in helping policy‑makers in peripheral countries avoid the negative consequences of exploitation by transnational corporations.

    There is also the question of the uses of causal models and quantitative research for informing political practice.  There are many political organiza­tions and movements that might make good use of the results of comparative research on world‑system processes.  But just as policy‑makers and planners cannot apply models mechanically to every situation, so in politics (especially in politics) practice is more an art than a science because of the complex and conjunctural nature of the task.  This means that, as in medicine, the results of research must be used by informed and artful practitioners.

    As I see it the main use of a structural theory of the capitalist world‑economy is its potential to help us distinguish social changes and  political forces which reproduce capitalism from those which contribute to its transformation, and to identify "weak links" where political efforts at  transformation can bear the most fruit.  If the twentieth century is the beginning of a period of transition to a socialist world‑system, but the extant "socialist" states are functional parts of the capitalist world‑economy (Chase‑Dunn, ed., 1982b), we need a clearly specified theory of capitalism to help us distinguish newly emerging forms which contribute to the transforma­tion of the system from those which reproduce, further expand and deepen it.  And our new theory should also have implications for the question of agency in the building of socialism.