STÖR
Dependence and Dependencia Theory: Notes Toward Precision of Concept
and Argument                                                                                                   BUSifl ®
Raymond D. Duvall
International Organization, Vol. 32, No. 1, Dependence and Dependency in the Global System (Winter, 1978), 51-78.
Stable URL: http://links.jstor.org/sici?sici=0020-8183%28197824%2932%3Al%3C51%3ADADTNT%3E2.0.CO%3B2-V
International Organization is currently published by Cambridge University Press.
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/cup.html.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.
JSTOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support@jstor.org.
http://www.j stor.org/ Sat Sep 10 03:20:34 2005
Dependence and dependencia theory: notes toward precision of concept and argument
Raymond D. Duvall
The concept of dependence is used in several different scholarly traditions to refer to aspects of relational asymmetry in international and transnational relations. In three such traditions, dependence refers to three quite different kinds of concepts with the result that possibilities are restricted for fruitful dialogue about dependence across these scholarly traditions of dependencia theory, systematic empiricism, and formal, analytical theory. To aggravate this problem of multiple "languages," there are two basic conceptual notions generally associated with the term dependence. These two meanings have clearly distinct implications for the nature of a theory of dependence, the character of entities dependent on one another, and the assessment or measurement of dependence. Thus, if the "language" gap is to be bridged and fruitful dialogue is to occur among different scholarly traditions, attention must be directed to the basic conceptual meaning of dependence in each tradition. Dialogue between systematic empiricism and dependencia theory is possible if empiricists recognize the fundamentally historical and historicist character of the particular substance of dependencia theory. These principles are exemplified here.
The term "dependence" is heard frequently in discussions of contemporary international and transnational relations. This popularity is, in part, a result of a widespread acceptance and voguish adoption of theories of contemporary capitalist imperialism, of which the term is a common element. Indeed, an important part of imperialism theory is denoted by the term, that is, dependencia, or dependency, theory which is
Raymond Duvall is a member of the Political Science Department at the University of Minnesota. This paper is an outgrowth of a project on the empirical evaluation of dependencia theory under the direction of Bruce M. Russett at Yale University. Duncan Snidal, Steven Jackson, Scott Ross, Hayward Alker, James Caporaso, Brian Job, Gary Wynia, Douglas Bennett and John Freeman offered helpful suggestions for developing the thoughts presented here. Portions of the paper were presented to the Committee on Conceptual and Terminological Analysis at the Tenth World Congress, International Political Science Association, Edinburgh, Scotland, August 1976.
52
International Organization
attributable primarily to Third World scholars critical of some of the alleged consequences of global domination by developed capitalist metropoles. But the popularity of the term dependence is not limited to scholarship critical of developed capitalism, for it is often used, as well, in the language of social scientists who are part and parcel of the capitalist center. This is due somewhat to an effort by these scholars to come to grips with theories of imperialism, but more generally it is due to an increasing awareness of problems of the First World such as a dependence on petroleum produced abroad and a monetary interdependence among the major capitalist economies.
Popularity and frequency of use suggest that the term is believed to represent a concept or category beneficial for understanding modern international relations. But to be of much value as an aid to scholarly understanding, i.e., as an analytical concept, the term must have a degree of specificity of meaning sufficient to convey information about some phenomenon. To some extent this criterion clearly is satisfied. There is general agreement that the term dependence refers to asymmetric properties of the structure of relationships among social entities. But it has been frequently charged, especially by First World social scientists,1 that there is a lack of real precision, of conceptual clarity, in the particular asymmetric structural, or relational, properties that are denoted by the term. Thus, the term dependence is more a general cue, a "script,"2 than a precise "data container."3 It does not refer unambiguously to a single concept.
^anjaya Lall, "Is 'Dependence' a Useful Concept in Analyzing Underdevelopment?" World Development Vol. 3 (November 1975): 799-810. Philip O'Brien, "Dependency: The New Nationalism?" Latin American Review of Books Vol. 1 (1973): 35-41.
2The notion of "scripts" that I use here comes from contemporary traditions in psychology and linguistics in the analysis of natural language and knowledge and belief systems. "A script is an elaborate causal chain which provides world knowledge about an often experienced situation.. .. (S)cripts are predetermined sequences of action that define a situation. Scripts have entering conditions (how you know you are in one), reasons (why you get into one), and crucial conceptualizations (without which the script would fall apart and no longer be that script)." Roger C. Shank, "The Structure of Episodes in Memory," m Representation and Understanding: Studies in Cognitive Science, Daniel G. Bobrow and Allan Collins, eds. (New York: Academic Press, 1975), p. 264. "Script based knowledge is mundane information which tends to lie in the periphery of consciousness. . .. Scripts describe situations which are frequently encountered with negligible or predictable variations. People in situational scripts are seldom conscious of script related actions." Wendy Lehnert, "What Makes SAM Run? Script Based Techniques for Question Answering," in Theoretical Issues in Natural Language Processing: An Interdisciplinary Workshop in Computational Linguistics, Psychology, Linguistics, Artificial Intelligence, Roger C. Shank and Bonnie L. Nash-Webber, eds. (Cambridge, Mass.: June 10-13, 1975), p. 59. "[T]he 'consequence' relation for weaving themes together [into scripts] is much more vague and potentially idiosyncratic than was the 'causal' relationship for linking A and S-atoms [i.e., defining concepts]. Therefore we expect a great deal of variation from one belief system to another in how themes are combined into scripts.... There is a great deal of room for idiosyncracy, and we doubt that there is any way... to generate 'correct' scripts by an automatic procedure." Robert P. Abelson, "The Structure of Belief Systems," in Computer Models of Thought and Language, Roger C. Shank and Kenneth Mark Colby, eds. (San Francisco: W. H. Freeman, 1973), p. 333, emphasis in original. Thus, to say that the term dependence is a script is to say that it is an important organizing device for knowledge about social relations, but that as such it is not a precise analytical concept because its "meaning" is too complex, idiosyncratic, and unrefined.
3The term "data containers" is taken from Giovanni Sartori, "The Tower of Babel," paper presented at the Ninth World Congress of the International Political Science Association, Montreal, Canada, August
Dependency and dependencia theory
53
Moreover, in several published criticisms by First World scholars, it has been claimed that ambiguity is not restricted to the term dependence, but is also characteristic of the several arguments entailed in the body of theory denoted by that term.4 According to this line, dependencia theory is not marked by a high degree of precision in the statement of hypothesized relationships among concepts (including, perhaps, a concept of dependence).
If these claims are valid, dependence and dependencia theory are not apt to become important aspects of international relations scholarship in any but a faddish sense even though the class of phenomena to which they generally refer may be important and enduring aspects of the global social system. In this respect, the fate of integration theory provides, perhaps, an instructive lesson.5
The problem, however, is not simply an ambiguity of explicit meaning or a vagueness of precise empirical referent for the term dependence or for dependencia arguments. Rather, there is a much more fundamental problem than one of denotation. Scholars are literally speaking past one another in the sense that the term has very different conceptual and theoretican connotations in different contexts or conventions of usage. That is, the term dependence is not simply a "script" or general cue rather than a "data container," but more importantly, it is a "script" for very different scenarios, or "frames."6 Arguments about dependence are different in kind as well in particular content.
1973. Sartori argues that rather than developing a useful language with which to construct knowledge claims, social scientists are creating tumult because very little attention is paid to the extent to which terms have a shared or common meaning. For Sartori, terms become concepts useful for empirical science when they become precise "data containers" upon whose contents there is widespread professional agreement. That position is in opposition to one which recognizes the importance to knowledge, even scientific knowledge, of conceptual scripts as per the previous footnote.
4C. Richard Bath and Dilmus D. James, "Dependency Analysis of Latin America: Some Criticisms, Some Suggestions," Latin American Research Review, Vol. 11, No. 3. (Fall 1976): 3-54. Ronald H. Chilcote, "Dependency: A Critical Synthesis of the Literature," Latin American Perspectives, Vol. 1, No. 1 (1974): 4-29. Raymond D. Duvall and Bruce M. Russett, "Some Proposals to Guide Empirical Research on Contemporary Imperialism," The Jerusalem Journal of International Relations, Vol. 2, No. 1 (Fall 1976): 1-27.
5In a recent piece by a leader in the development of integration theory, it is claimed that that theory is now obsolescent because some major assumptions on which it was based are no longer empirically valid. See Ernst B. Haas, "Turbulent Fields and the Theory of Regional Integration," International Organization, Vol. 30, No. 2 (Spring 1976): 173-212. Haas' interpretation is much too generous, I believe. Integration theory is obsolescent because international relations scholarship is faddish—intellectual problems are defined as much in response to current events as to gaps and anomalies in established bodies of knowledge. The reason for this is well known; "established bodies of knowledge," systematized as theories capable of revealing important gaps and anomalies, are largely lacking. Integration theory, concerned with phenomena that are still important today, failed ever to become an established body of knowledge (i.e., a systematic body of theory) in large part because there never developed a substantial scholarly agreement about the precise conceptual properties of integration, nor a clear specification of the arguments which constituted integration theory. An important but insufficient effort was made by Hayward R. Alker, Jr., in "Integration Logics: A Review, Extension, and Critique," International Organization, Vol. 24, No. 4 (Autumn 1970): 869-913.
6The notion of "frames" is related to the notion of "scripts" presented in footnote 2 above. "Here is the essence of the frame theory: When one encounters a new situation (or makes substantial changes in one's view of a problem), one selects from memory a structure called a frame. This is a remembered framework to be adapted to fit reality by changing details as necessary. . . . We can think of a frame as a
54
International Organization
The main premise of this paper is that unless and until the nature of the different "frames" of reference is clearly understood scholarly dialogue about dependence has severely limited possibilities. Most scholars allegedly participating in a general dialogue about dependence are not in fact doing so because there is not yet a clear understanding of the nature and consequences of the different "frames" of reference for the term.
This paper is an attempt to provide a greater clarity so that a more useful dialogue can be promoted. It proceeds in three stages. In the first section, there is a discussion of the use of the term in different "language" traditions based on different orientations to knowledge. This provides a primary basis of distinguishing distinct referential frames. The second section is an analysis of the term dependence at a general, abstract level. The intent is to show that a secondary basis for distinguishing frames of reference cuts across and is intertwined with the first, with important consequences for the possibility of dialogue within and between different "language" traditions. In these first two sections the nature of the term dependence for dependencia theory is a special concern. Finally, in a third section, an attempt is made to represent some of the important arguments of dependencia theory in a fashion consistent with the nature of the term dependence as it is used in that body of theory but in a form acceptable to a different "language" tradition, that of First World scholars committed to a rigorous empirical social science and a logic of falsification. The hope is that this section will point to the possibilities of a genuine and fruitful dialogue.
Dependence in different scholarly languages
One can distinguish at least three scholarly traditions in which the term dependence is currently being used with some frequency in reference to international or transnational relations. Each of these traditions is based on a different orientation to knowledge. In that sense, each offers a different "language" for the analysis of dependence.
network of needs and relations. The 'top levels' of a frame are fixed, and represent things that are always true about the supposed situation. The lower levels have many terminals—'slots' that may be filled by specific instances or data." Marvin Minsky, "A Framework for Representing Knowledge," in Theoretical Issues in Natural Language Processing: An Interdisciplinary Workshop in Computational Linguistics, Psychology, Linguistics, Artificial Intelligence, Roger C. Shank and Bonnie L. Nash-Webber, eds. (Cambridge, Mass.: June 10-13, 1975), p. 118. "Underlying the organization is a belief that meanings cannot be reduced to any set of pure 'elements' or components from which everything else is built. Rather, a person categorizes his experience along lines which are relevant to the thought processes he will use, and his categorization is generally neither consistent, nor parsimonious, nor complete. A person may categorize a set of objects in his experience into, for example 'chair,' 'stool,' 'bench,' etc. If pushed, he cannot give an exact definition for any of these, and in naming some objects he will not be certain how to make the choice between them. This is even clearer if we consider words like 'truth,' 'virtue,' or 'democracy.'" Terry Winograd, "A Procedural Model of Language Understanding," in Computer Models of Thought and Language, Roger C. Shank and Kenneth Mark Colby, eds. (San Francisco: W. H. Freeman, 1973), p. 168. My point is that dependence is just such a "frame."
Dependency and dependencia theory
55
First are the dependencia theorists, or dependentistas, who are largely, but not entirely, Third World scholars, especially Latin Americans. But the geographical or national origin of scholars is not the distinguishing feature of dependencia theory.7 Rather, it is a set of questions, a perspective on those questions, and a concomitant orientation to knowledge about them that distinguishes the tradition. In particular, dependencia theorists focus on the economic, social, and political transformations, especially "distortions," of "peripheral" societies that result from, or are reflections of, the incorporation of those societies into the global capitalist system. The orientation taken by dependencia theorists is said to be historical, structural, and dialectical in that "The aim ... is to describe social process in terms of a 'complex whole,'. .. whose structural movement derives from the contradictory forces which sustain it."8 With this orientation, dependentistas have avoided, indeed, have explicitly rejected, a concern for conceptual precision and measurement:
Of course, the ambiguous and contradictory characteristics of actual history—which constitute muscle and meat for dialectical analysis—will disappear as a methodological consequence of this type of formal procedure which drowns history in 'rigorous' concepts. By rigorous what is meant is precisely the lack of opposition and dynamism which gives life and movement to social structures.9
Moreover, the claims to knowledge, the arguments, which constitute dependencia theory are not intended to be general formulations applicable to a wide range of situations and contexts. Instead, they are offered as descriptive statements whose validity is contingent on particular context, as indicated in the following quotations:
It is a mistaken approach, for instance, to test hypotheses about the effects, let us say, of associated-development on income distribution in situations of dependency based on the penetration of multinational corporations in the sector of mass consumption production, constructing the "independent variable," using aggregate data. . . drawn from economies based on enclave types of dependency, national export economies, and industrialized economies con-
7Several Europeans and students of Africa, at least, have made substantial contributions to the literature on "peripheral capitalism" with exactly the same concerns as the Latin American scholars who are universally recognized as the dependentistas. See, for example, Dieter Senghaas, "Multinational Corporations and the Third World: On the Problem of the Further Integration of Peripheries into the Given Structure of the International Economic System," Journal of Peace Research, Vol. 12, No. 4 (1975): 257-74; Eva Senghaas-Knobloch, "The Internationalization of Capital and the Process of Underdevelopment: The Case of Black Africa," Journal of Peace Research, Vol. 12, No. 4 (1975): 275-92; Samir Amin, Accumulation on a World Scale: A Critique of the Theory of Underdevelopment, 2 vols., (New York: Monthly Review Press, 1974); Ernest Mandel, Late Capitalism (London: NLB, 1975); Walter Rodney, How Europe Underdeveloped Africa (London: Bogle L'Ouverture, 1972).
8Fernando Henrique Cardoso and Enzo Faletto, "Preface to the American Edition," Dependency and Development in Latin America. (Berkeley: The University of California Press, forthcoming). The quotation is from p. 18 of the typewritten manuscript.
9Ibid, pp. 22-23 of the typewritten manuscript.
56
International Organization
trolled by multinational corporations, all together, as if they belonged to the same historical continuum. No statistical result extracted from such an amalgam can prove or disprove any statement on dependence because the basic assumption of such a formal methodology dissolves the differences in situations of dependency and these are key notions for the historico-structural approach.10
In any case, it would not be correct to substitute what we have made, that is, the dialectical analysis of concrete situations of dependence, by a formal theory.11
In this language tradition, which places primary emphasis on "concrete analysis" and detailed, historical, descriptive, contextually-bound knowledge claims about processes and conditions of the structural transformation of peripheral capitalist societies, the term dependence has a very special role. That role is to delineate the general context in which the knowledge claims apply—to denote the appropriate realm of the subject matter. In this sense, dependence is a label for a general frame of reference, rather than a term which applies to a particular conceptual referent within that frame. The various definitional statements for dependence that are provided in the dependencia tradition, then, serve less to denote a precise analytical concept to be utilized in theoretical statements than to establish the general context to which knowledge claims apply.
The most frequently cited definitional statement is one provided by Theotonio Dos Santos:
By dependence we mean a situation in which the economy of certain countries is conditioned by the development and expansion of another economy.. .. The relation of interdependence between two or more economies. . . assumes the form of dependence when some countries . .. can expand and can be self-starting, while other countries . . . can do this only as a reflection of that expansion, which can have either a positive or a negative effect on their immediate development.12
For the moment, two things are important to note about this statement. First, dependence is a situation. It is not a variable property of a country or relationship to be attributed to that country or relationship in greater or lesser degree. Rather, it is a certain form of relationship in the context of which other phenomena such as economic expansion and development, which are variable properties, are important and are the subjects of study. Second, dependence does not explain, or account for, development. On the contrary, Dos Santos communicates quite clearly that having knowledge that there exists a situation of dependence for a country does not, by
10Ibid, p. 23 of the typewritten manuscript.
"Fernando Henrique Cardoso, " 'Teória de la Dependencia' o Analisis de Situaciones Concretas de Dependencia?" Revista Latinoameričana de Ciencia Politica, Vol. 1 (December 1970): 414, emphasis in original. The translation is with the help of Bruce M. Russett.
12Theotonio Dos Santos, "The Structure of Dependence," American Economic Review, Vol. 60, No. 2 (May 1970): 231, emphasis added.
Dependency and dependencia theory
57
itself, permit any conclusion about the development of that country. Rather, in the context of dependence, development is to be explained by other phenomena (not specified in this definitional statement). This means that, for Dos Santos, at least, the term dependence delimits a context of inquiry, not a concept in theory.
But Dos Santos is not at all unique in this regard. Consider Cardoso's statement that "dependence, as we characterize it, is no more than the political expression on the periphery of the mode of capitalist production when this mode is taken up to international expansion."13 Senghaas, too, offers as a definitional statement a delimitation of general context much more than a specification of meaning for a particular substantive concept.
The economies of the Third World... thus [represent] a reflex reaction to the developmental direction of the world economic system which is decisively determined by the capitalist industrial nations. Therefore, it is still correct to term... the integration of the peripheries into the world market as structural dependency; for... the reproduction dynamics of the metropoles are transformed into the structure of the peripheries themselves (with considerable deforming repercussions), whereas no comparable process takes place in the opposite direction.14
In the dependencia "language" tradition, then, the term dependence is intended to connote a general "frame" rather than a precise "data container." A charge that dependence does not refer unambiguously to a precise concept and does not have a precise empirical referent is not a charge of much relevance to this tradition. Cardoso uses the term "totalizing" to refer to a concept whose meaning is precisely specified, primarily by denotation, and which is applied universally when denotative criteria are satisfied (i.e., an empirically useful scientific concept in the conventional Hempelian sense). He then rejects such a role for the term dependence. ' 'To pretend to elevate the notion of dependence to the category of totalizing concept is a paradox. .. . [W]e did not want to make a theory of dependence appealing to a totalizing notion of dependence."15
But to conclude that, as a result, dependence as used in the dependencia tradition is meaningless would be entirely erroneous. It is meaningful even as a scientific term because its connotation of context, of referential frame, is quite clear and unambiguous. In particular, dependence refers to a context of differentially or asymmetrically structured reflections of the processes of capitalist production and reproduction at the international level. A situation so characterized is a situation of dependence. The knowledge claims which constitute dependencia theory are about aspects of such situations. Thus, in the dependencia tradition dependence is meaningful primarily in delimiting quite clearly the context of theory; in general, it is not
13Cardoso, p. 406.
14Dieter Senghaas, "Introduction," to a special issue, "Overcoming Underdevelopment," Journal of Peace Research, Vol. 12, No. 4 (1975): 249-50, emphasis in original. 15Cardoso, p. 405 and p. 414, emphasis added.
58
International Organization
meant to be meaningful as an empirical concept within that theory (i.e., appearing in knowledge claims as a variable property of relationships or countries).16
Two implications of this point seem worthy of mention. First, the often waged criticism that dependencia theorists pay insufficient attention to Soviet imperialism and the dependence of Eastern Europe on the Soviet Union is fundamentally irrelevant and misdirected. The referential context is different, and hence, according to dependencia theorists, a different set of processes is apt to be involved and a different set of knowledge claims is apt to be validly applicable. Critics might wish that a more general and more inclusive context had been delimited such that the theory would be equally applicable to both Soviet and American imperialism. But to make such a wish is not to assure that the two different dependence contexts could be handled at anything approximating the same level of abstraction at which dependencia theory has been cast. A dependencia theory that included Soviet imperialism as an appropriate referential context would be a very different, a much more abstract body of theory. One might believe that that would be preferable but I know of no non-arbitrary criterion by which one could make that judgment. Certainly the phenomena referred to in dependencia theory as currently constituted are pervasive and important in their own right.17
Second, the almost universal representation of dependencia theory by North American scholars as a theory about the relationship between dependence and development (or underdevelopment) is an unfortunate and misleading representation.18 It is reasonable only to the extent that it implies that in a context of dependence certain processes affect the rate, the direction, even the possibility of economic expansion, or development. That is, development is affected by processes X,
16There are exceptions of some importance. One reads in dependencia theory of the cultural dependence of the Ivory Coast on France, or of the technological dependence of Brazil and Mexico. In such instances, the term dependence is a variable property of countries and is a concept m the theory. I ignore such usages at this point because I want here to emphasize the primary implication of the distinct "language" tradition of dependencia theory, namely that in an orientation to knowledge that is wholistic and particularistic, the clear determination of general referential context is very meaningful. In fact it is of primary importance, because otherwise any knowledge of particular wholes is entirely arbitrary and unique. A primary importance of dependence, then, is to refer to the general context that is delimited. Other uses of the term in the dependencia tradition are discussed in the second section of this paper.
17I speak of dependencia theory as being "currently constituted" because I believe that the basic perspective, the orientation to knowledge, and even the set of questions that have marked it persist as very important, especially, but increasingly less so, among Third World social scientists. There is some sentiment today that dependencia theory is dead, at least among Latin Americans. That position is tenable only by defining very restrictively the set of questions that characterize dependencia theory. Because somewhat different questions (e.g., the role and character of the peripheral capitalist state) attract more attention today is not to say that dependencia theory has been abandoned and another developed. Rather, it is to recognize that the purview of dependencia theory has been increased.
18Such a representation is offered or implied in Robert R. Kaufman, Harry I. Chernotsky, and Daniel S. Geller, "A Preliminary Test of the Theory of Dependency," Comparative Politics, Vol. 7, No. 3 (April 1975): 303-30; David Ray, "The Dependency Model of Latin American Underdevelopment: Three Basic Fallacies," Journal of Inter ante rican Studies and World Affairs, Vol. 15 (February 1973): 4-20; Christopher Chase-Dunn, "The Effects of International Economic Dependency on Development and Inequality: A Cross-National Study," American Sociological Review, Vol. 40 (December 1975): 720-38; Patrick J. McGowan, "Economic Dependence and Economic Performance in Black Africa," Journal of Modern African Studies, Vol. 14 (March 1976): 25-40.
Dependency and dependencia theory
59
Y, and Z, given a context of dependence. But such is not generally the implication given by most representations of the theory. The common implication is that development is affected by dependence, in which case the latter is transformed into a (central) concept in theory, and hence, is nothing more than a variable property of countries or of relationships among countries. Such a representation reflects a fundamental misunderstanding of the meaning of dependence for dependencia theory and distorts tremendously the nature of that theory.
The fact that such representations are commonplace is due, I believe, to the dominance in North American scholarship of a second "language" tradition. In particular, a commitment to a rigorous empirical social science has led to various attempts to increase the analytical scope and precise empirical content of the term dependence by developing measurement models for it.19 The tradition of rigorous empiricism is distinguished by a commitment to precise measurement criteria and on that basis alone is distinguishable from the dependencia tradition. Unfortunately, the distinction does not end there because empirical scientists have generally been guilty of ignoring that degree of precision of meaning which is provided by dependencia theorists for the concept of dependence. Instead, they have started with more common-sensical meanings of the term, and have given the impression that measurement rules developed in accord with those meanings are all that matter. Precision is provided by, and meaning is captured in, operational criteria. The result is a distortion of the primary meaning of the term as it appears in dependencia theory. A construct intended to refer to general referential contexts, to delimit the bounds of theory, becomes in this "language" tradition a concept to be measured, a variable property of countries (or other entities). In particular, dependence here refers to the extent to which value attainment is controlled externally and especially as the control is concentrated in a single entity, rather than to a situation of structured asymmetry in the consequences of international processes of capitalist production and reproduction. The two are related, but clearly distinct, kinds of concepts.
It is not my intent to condemn the enterprise of devising precise measurement models for a concept of dependence. Rather, it is simply to argue that in developing precise measurements one is talking about and dealing with a different type of concept than the primary concept of dependence in dependencia theory.20 In itself, that is not a bad; the same term may profitably be used to refer to quite different kinds of concepts. But it is a bad if it is not recognized and kept clearly in mind that different kinds of concepts are being referenced. Unfortunately, that seems most frequently to be the case because many rigorous empiricists claim to be developing
19This tradition is well represented by James Caporaso, "Methodological Issues in the Measurement of Inequality, Dependence, and Exploitation," mTesting Theories of Economic Imperialism, Steven Rosen and James Kurth, eds. (Lexington, Mass.: Lexington Books, 1974) pp. 87-114. It is also represented by Duvall and Russett, who presented an empiricist position that I now reject in some part.
20Again I use a modifier, this time "primary," to make it clear that I recognize the periodic secondary usage in dependencia theory of dependence as a variable property concept in theory. This secondary usage creates a real ambiguity, and justifies a desire to increase precision through the development of explicit measurement criteria. It does not justify the position that the measured dependence is central to dependencia theory. It is not!
60
International Organization
the requisite bases for systematic tests of dependencia theory. That is, they imply that the dependence which they measure is the dependence that is central and primary to dependencia theory.21 But it is not. Thus, most of the concern among North American social scientists with testing dependencia theory by measuring dependence as the central concept of that theory is misdirected and is a nonsensical enterprise. It is this fact that stimulated Cardoso to point to the "poor consumption" of dependencia theory in the United States through the attempts at creation of a "totalizing" concept of dependence.22 Two "languages" are being spoken. Conversations in one have very little direct bearing on conversations in the other.
Largely removed from this "dialogue" but not irrelevant to it is a third tradition. This is the tradition of abstract analytical theory, in which a few social scientists from both the First and Second Worlds have attempted to incorporate the term dependence into their formal theoretical vocabularies.23 To my knowledge, there is no allegation in this tradition that formal theorists are talking about the same thing as are dependencia theorists. Thus conceptual confusion is unlikely and it is not necessary for me to belabor distinctions. Suffice it to say that a third distinct kind of concept of dependence is at issue here, namely, a variable property of relations between purposeful or goal-directed systems. Dependence refers to "a relation in which one system affects. . . —by specific means (rewards and punishments)—the [system and components needs satisfaction] of another system."24 This is dependence as relational control. As such it is not a measurable property of social entities such as countries as it is for rigorous empiricists. Nor is it a general frame of reference for particular processes that relate countries to one another and affect outcomes for them as it is for dependencia theorists. It contains an element of each, but encompasses neither.
The term dependence, then, is used today to refer to at least three distinct kinds of concepts relevant to the study of international and transnational relations. On the one hand, it is used by a group of scholars who are concerned with an wholistic descriptive analysis of historical processes of social-structural transformation to demarcate the general context of their inquiry. Adjectives are used in this tradition
21Examples are the works cited in footnotes 18 and 19 above. The problem is the now common logical error that because dependencia theory is concerned with the notion of dependence, whenever one talks about, measures, or analyzes some particular aspect, form, or dimension of dependence, one is thereby reflecting on dependencia theory. That is simply not true.
22Fernando Henrique Cardoso, "The Consumption of Dependency Theory in the U.S.," Latin American Research Review, Vol. 12, No. 3 (1977).
23Examples in this tradition are František Charvát, Jaroslav Kučera, and Miroslav Soukup, "Toward the System Theory of Dependence: Further General Theoretical Remarks," in Mathematical Approaches to Politics, Hayward R. Alker, Jr., Karl Deutsch, and Antoine H. Stoetzel, eds. (San Francisco: Jossey-Bass, 1973): 263-86; Hayward R. Alker, Jr., "On Political Capabilities in a Schedule Sense: Measuring Power, Integration and Development," m Mathematical Approaches to Politics, Hayward R. Alker, Jr., Karl Deutsch, and Antoine H. Stoetzel, eds. (San Francisco: Jossey-Bass, 1973), pp. 307-73; Richard M. Emerson, "Power-Dependence Relations," American Sociological Review, Vol. 27, No. 1 (February 1962), pp. 31-41; Thomas Baumgartner, Walter Buckley, and Tom R. Burns, "Relational Control: The Human Structuring of Cooperation and Conflict," Journal of Conflict Resolution, Vol. 19 (1975): 417^0.
24František Charvát, Jaroslav Kučera, and Miroslav Soukup, p. 270.
Dependency and dependencia theory
61
to place more restrictive bounds on the general context to which dependence refers. Thus, for example, Dos Santos distinguishes as major historical phases, colonial dependence, financial-industrial dependence, and the new dependence based on multinational corporate investment.25 Second, for rigorous empiricists committed to precise measurement of concepts, dependence refers to a property of countries. Again adjectives are used but now to make more particular the measurable property. Thus, trade dependence, which is a favorite object of study, refers to the extent of (potential) external control of a country through its trade. This clearly is a different concept of trade dependence than it is for a dependencia theorist for whom the term refers to the particular form that the dependence relation (which exists independently of trade domination) takes in some historical context. Third, for analytical theorists who are steeped in the notions of purposeful systems, choice, and goal-directed behavior, dependence is an aspect or property of the relationships between such systems pursuing their separate goals. As such it is a concept akin to power in the Dahl and Harsanyi tradition26 and is specific to a particular context of social relationships. It is not a.stable property of any of the countries in a relationship as it typically is for rigorous empiricists. Neither is it a persisting and comprehensive context of relationships and the structural consequences of those relationships as it is for the dependencia theorists. Rather, it is a variable property of those relationships, per se.
Conceptual bases of the term dependence
The problem of multiple languages and different kinds of referential frames is compounded by a problem of different conceptual bases for the meaning of the term. Dependence not only refers to different kinds of concepts but for those different kinds of concepts there are also a set of different meanings. In particular, one can clearly distinguish two basic meanings, each with importantly different implications.
The word dependence derives from the conjunction of two sources: the Latin dependere (and/or the Old French dependre) which meant to hang or be suspended from; and the Latin suffix entia which denoted action, process, state, or quality. Etymologically, then, dependence means a process or state of hanging from. But historically two distinct senses of "the state of hanging from" have developed in conventional usage. At least as early as the Sixteenth Century it was used to refer to a state of contingent or conditional relationship. In this sense, something hangs or is suspended from another if some of its features are conditioned or determined by features or actions of that other. By the Seventeenth Century dependence was used
25Dos Santos, p. 232.
26Alker makes this conceptual linkage explicit. He focuses his attention on the Dahl-Harsanyi conceptions of power, and in doing so introduces the concept of dependence as the direct opposite of power, equivalent to powerlessness. Alker, pp. 308, 311.
62
International Organization
in a second more particular sense to refer to a state of subordinate or subject relationship. Here, to hang from something means to derive support and/or the basis of existence from that something.
One can see the residues of these two basic meanings in the various definitional statements offered for dependence in the contemporary literature of international relations. In the previously cited passage from Dos Santos, emphasis is placed on the conditioning of one economy by another and on the situation where the former is a reflection of the latter.27 Similarly, the definition offered by Cardoso that is mentioned above has as its primary operative term the word expression—the dependence situation is one in which features of the peripheral country are expressions of (i.e., reflections of, conditioned by) the operation of the international capitalist system.28 Senghaas points to decisive determination and reflex reaction.29 In each of these cases (and, indeed, throughout the dependencia literature), dependence is being used in the first of the two senses distinguished above, namely to refer to a contingent or conditional relationship. Alternatively, the definition provided in the previous section to exemplify the language tradition of formal theory places emphasis on the satisfaction of system needs and by implication is referring to dependence in the second sense as a supportive (or conversely subordinating) relationship.30
This two-fold distinction of meanings corresponds directly to the distinction made earlier by Keohane and Nye between "sensitivity interdependence" and "vulnerability interdependence."31 The first of these refers to the extent to which changes in the structure of relations have significant, effects on the parties to those relations. The second refers to the costs that are realized (or would be realized) by
27Dos Santos, p. 231.
28Cardoso, " 'Teória de la Dependencia' o Analisis de Situaciones Concretas de Dependencia?", p. 406.
29Senghaas, "Introduction," pp. 249-50.
30František Charvát, Jaroslav Kučera, and Miroslav Soukup, p. 270. These authors go on to make the linkage to supportive relationships explicit—"The dependence relation is not determined by what has happened, but by what might have happened if the behavior of another system would not, in certain respects, have corresponded to the selective demands of the first system.. .. The compatibility of two systems depends on their capability of mutually satisfying the system needs, on their complementarity and on their similarity as to the extent of required exchange with other systems in all the areas of need," p. 270.
31Robert O. Keohane and Joseph S. Nye, "World Politics and the International Economic System," in The Future of the International Economic Order: An Agenda for Research, C. Fred Bergsten, ed. (Lexington, Mass.: Lexington Books, 1973), p. 124. Others have offered enumerations oi types of interdependence, but these typically are based more on forms and manifestations than on fundamental conceptual distinctions. Because my concern is with the implications of the latter, I do not pursue the former. That is, I do not pretend in this paper to be offering the only important types of dependence, and hence, do not develop comprehensive typologies. Interesting and useful typologies are provided in Edward L. Morse, "Interdependence in World Affairs," in World Politics: An Introduction, James N. Rosenau, Kenneth W. Thompson, and Gavin Boyd, eds. (New York: The Free Press, 1976), pp. 660-81; Hay ward R. Alker, Jr. and Nazli Choucri, "Methodological Conclusions and Recommendations," Chapter IV oi Methodological Perspectives and Research Implications, Vol. Ill oi Analyzing Global Interdependence (Cambridge, Mass.: Center for International Studies, MIT, 1974), pp. 57-72. The types that these and other scholars offer are easily analyzable as forms of the two basic conceptual meanings developed here.
Dependency and dependencia theory
63
the parties as a result of such changes. Sensitivity is conditional relationship. Vulnerability is subordinate and/or support relationship. And just as vulnerability entails sensitivity but goes beyond it (something cannot be vulnerable if it is not sensitive), so dependence as subordinate relationship entails dependence as conditional relationship but goes beyond it (something does not derive its support or existence from another if that other has no conditioning effect on it).
Why is the distinction made here if, in essence, it has already been made? The answer is that those aware of the distinction have not made explicit some of its important implications. I attempt to do so here in terms of three issues: 1) implications for the nature of a theory of dependence; 2) implications for the nature of the parties to a relationship or set of relationships characterized as a situation of dependence, that is, the nature of the units or entities the relations among which define the dependence situation; and 3) implications for the measurement of a substantive concept of dependence.
Statements about dependence in the first sense are statements about contingency, about the conditions under which phenomena occur or acquire certain properties. Thus, they are exactly equivalent to statements in scientific theory in general.32 Dependence in this sense refers simply to a logical aspect of theory and need not indicate anything about the substantive content of the theory. A theory of dependence in this first sense, then, could be of virtually unlimited substantive concern. Of course, when dependence in this sense is used more restrictively to designate particular kinds of conditional relations, as it is by dependencia theorists, it does serve to delimit a theoretical substance. But even here all that is implied is the general substance of the theory; nothing is implied about particular conceptual content. Thus, dependence in the first sense may be nothing more than a label for theory, designating its general substance, in much the same way that the term relativity is used to label relativity theory. As with relativity, it need not, itself, be a substantive component of the theory. Rather, the theory label may simply be the script for a general referential frame. This is the relativity of relativity theory and, according to the argument of the previous section, it is the primary dependence of dependencia theory.
This point is of some importance because an apparent failure to recognize it has contributed to the lack of a genuine dialogue between dependencia theorists and North American empiricists. The line of thought of the latter group to date seems to have been that dependencia theory is about dependence, that dependence must, then, be a central concept in the theory, and finally then, that a systematic test of the theory rests on an adequate measurement ofthat central concept. But the point here is that dependencia theory can be about dependence without dependence being a central concept in theory if dependencia theory is about dependence in the first sense. In what follows, I attempt to show that dependencia theory must be about dependence in this first sense, and hence, need be nothing more than a frame, a label for theory.
'Simply consider the use of the term dependent variable in the scientific tradition.
64
International Organization
Statements about dependence in the second sense are statements about support and/or subordination. They are substantive statements. A theory of dependence in the second sense, then, is a theory of a particular substance in that it must provide a basis for determining the provision of support or the provision of means for the continued existence of some referent entity. This is to say that the theory must entail, on the one hand, notions of needs, or requisites, or other criteria of systemic existence, and/or, on the other hand, notions of cost, or value, or other criteria of systemic support. In either case, these are the notions of purposeful systems, whether functionalist (i.e., nonsentient goal-seeking) or rationalist (i.e., sentient goal-seeking) in character. A theory of dependence in the second sense must be explicitly a theory of purposeful systems; a theory of dependence between and among nation-states must be a theory of nation-states as purposeful systems.
A second implication of the two basic conceptual meanings follows directly from the preceding discussion. It concerns the nature of the units or entities which are in relation to one another and, hence, which define a relational situation of dependence. If dependence in the second sense is to be used meaningfully, the referent situation must entail unitary actors as the entities which are in relation to one another. Purposeful systems are of necessity unitary—needs, requisites, costs are notions applicable only to such entities. For dependence as a conditional relation the implication is not so restrictive. That which is conditioned must be an entity in some sense, but it can be an arbitrarily defined system, a mere social aggregate.33 In talking of, or theorizing about, international dependence, then, one implies a treatment of nation-states as unitary actors on the one hand or as social aggregates with some minimal degree of entitivity on the other.
This difference is consequential because the latter option is entirely consistent with an explicit perspective on the nation-state as a unit of social conflict, which perspective stands quite in opposition to that of it as a unitary entity. Thus, if the term dependence is used in the sense of a conditional relation, it is possible to talk meaningfully about international dependence while viewing nation-states as social formations that embody or represent the pattern of conflict relations among those social classes that are characteristic of the dominant mode of production at any time. In a word, the notion of international dependence is consistent with a Marxist theory of the state if dependence refers to a conditional relation. This is not logically possible if the term dependence is used in the sense of a support and/or subordination relation, that is, a situation of relational control, because the latter implies a treatment of the nation-state as unitary. Because in much of dependencia theory the nation-state is treated explicitly as a unit of social conflict and because it is a theory about international dependence, it must refer to dependence as a conditional rela-
33This is the entitivity of systemic interdependence in which "events occurring in any given part or within any given component unit of a. . . system affect. . . events taking place in . .. the other parts or component units of the system." Oran R. Young, "Interdependences in World Politics," International Journal, Vol. 24 (Autumn 1969): 726. A crowd, an audience can be an entity in this sense although it is clearly not an entity in the sense of a unitary actor. See Donald T. Campbell, "Common Fate, Similarity, and Other Indices of the Status of Aggregates of Persons as Social Entities," Behavioral Science, Vol. 3, No. 1 (January 1958): 14-25.
Dependency and dependencia theory
65
tion, not as an international subordination relation. This is an implication for the notion of dependence as relational control since that notion is tied quite clearly and directly to dependence in the sense of a relation of support and/or subordination. In particular, dependencia theory and arguments about relational control move in two quite different conceptual and theoretical directions. The problem of different "languages" between formal analytical theory and dependencia theory is overlaid with a problem of basically different conceptual meanings and different theoretical implications.34
A third implication of the distinction between two basic meanings of dependence is for the measurement of the concept. Most basically, of course, it means that one does not simply measure dependence in the abstract if measurement is to be meaningful and fruitful; one measures dependence in one particular sense or another. This fact alone should make us suspicious of those measurement models that have been developed without clear reference to a particular meaning of dependence as if a "totalizing" common-sensical meaning were apparent and sufficient. But more than that can, be said if we consider each of the two basic meanings separately.
To this point, when discussing dependence as a conditional relation, I have placed emphasis on the possibility of its being a general label for theory, connoting a referential frame, rather than a particular concept in theory to be measured. However, dependence in this sense may, and sometimes does, become a particular concept in theory. This occurs when some property of a conditional relation is, itself, a phenomenon to be explained or an element in the explanation of some other phenomenon. Thus, a central concern in dependencia theory for processes of increased peripheral-state autonomy, the decline of statist imperialism, and the emergence and development of a "state bourgeoisie"35 with its own class interest, capable of forming and shifting coalitional ties with other classes, is a concern for the explanation of the degree of political dependence, i.e., the degree to which the character and policy of government are conditioned by or contingent upon features and activities of actors from metropole countries. Dependence as conditional relation, then, may sometimes be a concept in theory, in which case its measurement might be (but not necessarily is) relevant.
34I am not saying that a concern for relational control, for asymmetric bargaining, the structuring of choice alternatives, etc., is uninteresting. On the contrary, these are very important and interesting problems. They just should not be mistaken as the same concerns as those of dependencia theory at a higher level of abstraction. They are fundamentally different, not simply different in level of abstraction. Thus, I am not disparaging, only differentiating.
35The term comes from Cardoso. "By way of a hypothesis, I would ask if we should not speak of the formation of a new social category that could be designated as a 'state bourgeoisie' in some Latin American countries, especially Brazil and Mexico... .This expression, which is formally contradictory, becomes relevant when one sees that the expansion of the public sector (which constitutes almost a national response to the imperialist challenge) in those Latin American countries that have taken the path of internationalization of the market takes place in such a way that the form of ownership of state enterprises is public, but their control is through a group which I have called bureaucratic, but which begins to show characteristics which are not explained by the phenomenon of bureaucracy." Fernando Henrique Cardoso, "Some New Mistaken Theses on Latin American Development and Dependency," unpublished manuscript, n.d., p. 21.
66
International Organization
How does one measure the extent of contingency? The answer that is implied by those scholars who claim to have developed measurement models for dependence to test dependencia theory is that one measures dependence by assessing the quantitative density and concentration of relational ties, or exchanges, as physical flows. Virtually universally, measurement models have been based on the extent and concentration of flows. But these can be appropriate indicators only to the degree that the extent of contingency, the extent to which aspects of an entity are conditioned by another, corresponds directly to the extent and concentration of physical flows between them. It seems patently unsafe to make an assumption of such correspondence. Alternative measurement strategies are needed.36
A more reasonable place to begin is directly with the logical character of conditional relations, namely that some phenomenon, Y, occurs or takes on a particular value in response to some other phenomena, Xx to Xn (for simplicity, call the set X) having occurred or having taken on particular values. This can be represented simply as X —> Y. The extent of dependence is the extent to which that representation is valid. Most generally we assess the latter by shifting to the perspective of statistical theory which is based on the concepts of probability and variance. Thus, to assess the validity of the X —> Y representation, and hence, in turn to assess the extent of (conditional) dependence is to estimate the degree to which the variance of Y is due to the variance of X and not due to other unspecified phenomena including, perhaps, pure chance. Where the variance of Y due to X is relatively large the contingency or conditional relation between them is strong and the extent of dependence, ipso facto, is large. To measure dependence of some form more directly and more adequately than through existing measurement models, then, is to estimate the extent to which the variance of (i.e., changes in) the "dependent" phenomenon (e.g., the product diversity of the export sector) is due to (i.e., "reflects") the occurrence of or changes in those phenomena on which it is allegedly dependent (e.g., the forms and extent of metropole capitalist penetration).
Three things are worthy of note at this point. First, the measurement of some form of dependence as a conditional relation requires time-series or change data. It is impossible to estimate variance for an entity at a single point in time. To say how dependent a particular country is in terms of some specific aspect or form of dependence it is necessary to observe change or variance over time in that aspect of the country (and in the phenomena on which it is dependent). The assessment, the measurement, of dependence is fundamentally based in the grasping of processes over time. This fact corresponds directly to the claim by dependencia theorists that dependence can be known only through historical analyses.
36It should be noted that I am not alone in being dissatisfied with these measurement strategies. "The measurement of transactions is a particularly inappropriate means of measuring sensitivity. . . . Few of the transactional flow models specify any relationship at all between relative level of flow and national economic or political sensitivity. This is not to say that the measurement of sensitivity is impossible, or even difficult." Morse, p. 676. See also Robert D. Tollison and Thomas D. Willett, "International Integration and the Interdependence of Economic Variables," International Organization, Vol. 27 (Spring 1973): 255-73.
Dependency and dependencia theory
67
Second, each X—Y set defines a different form of dependence. By generating estimates for several different X—Y sets over long periods of time, one is able to evaluate statements about the transformations in important forms of dependence as well as changes in the extent of each for a particular country. Statements about changes in the particular forms of dependence are a central part of dependencia theory—consider, for example, the major historical phases of dependence distinguished by Dos Santos.37 The basic measurement model offered here permits a direct assessment of the general validity of such statements. That possibility is lost if measurement is oriented toward some "totalizing" composite property of a country such as its economic dependence, even its trade dependence.
Third, to use the concept of dependence as a specific substantive concept in theory becomes very difficult. In particular, whatever phenomenon is said in substantive theory to cause or to be caused by some form of dependence must be something other than that which is itself involved in the contingency relationship called dependence; otherwise the alleged explanation is tautological. That is, because dependence is the X—Y set, a non-tautological explanation of dependence necessitates the stipulation of a relation involving some other phenomenon, Z. Then, the extent of the X—Y contingency relation is due to Z. Because non-tautological statements require a clear and precise indication of what is entailed—a clear specification of the elements of the X—Y set—and because such referential clarity is generally lacking in verbal theory, non-tautological explanations of dependence are not commonplace. But tautology is not the same thing as non-falsifiability or scientific irrelevance. An "explanation" of dependence can be identical to the measurement of dependence outlined above if dependence as a concept in theory is an underidentified referential frame rather than a precise substantive concept. In this sense, dependence would be a label for some particular relationship that is part of the substance of the theory rather than a concept that is itself an element of such a relationship. Thus, even as a concept in theory, dependence as a conditional relation of a particular form is apt to be a referential frame, a label, rather than a precise analytical concept. When explanations of particular forms of dependence are offered in dependencia theory they are of this kind.38 To test these explanations systematically is simply to measure dependence of the form at issue.
When dependence is used in the sense of a relation of support or subordination, a different orientation to measurement is called for. In particular, to measure dependence is to assess an actor's costs or needs and to determine the extent to which these are affected by the actions of some other actor(s). I do not pursue measurement models along these lines here because this concept is not of much relevance to dependencia theory which is my primary concern, and because their basic features, which appear to be very difficult to effectuate, have already been developed.39
37Dos Santos, p. 232.
38This is evidenced by the several articles in Frank Bonilla and Robert Girling, eds., Structures of Dependency (Stanford, Calif.: Institute of Political Studies, 1973). These articles deal with various forms and aspects of dependence.
39Measurement models are offered in Alker, and in Charvát, et al.
68
International Organization
The measurement models of rigorous empiricists do not seem to capture the essence of the contingency model outlined above; nor do they explicitly include notions of costs or needs. The question, then, is what is being measured in assessing the extent and concentration of physical relational flows? The answer, I believe, is the media or mechanisms of contingency and/or the potential for subordination or support. In other words, existing measurement strategies are inadequate in failing to get to the core of the concept of dependence in any basic sense and in mixing aspects of very different meanings in a single measure.
A rigorous empiricism for dependencia theory
In the previous sections I have attempted to show that not all discussions of dependence in international relations are part of or even relevant to dependencia theory. Indeed most of the rigorous empirical research that is alleged to reflect on that body of theory does not do so because researchers apparently fail to recognize the special and distinguishing character of the theory. Three theoretical features should be emphasized and must be taken into account if meaningful dialogue is to take place between rigorous empiricists and dependencia theorists.40 First is the role of the term as a referential frame connoting the general script for a substantive theory of peripheral capitalism and sometimes denoting more particular conditional relations within that general substance, but generally not as a precise element of the relational statements that constitute the substance of theory. Second, as a referential frame denoting particular conditional relations which are to be assessed systematically, the concept of dependence leads directly to the logical necessity of incorporating "history," change, or processes over time into the assessment. Third, partly because of the "historical" basis of dependence relations, an awareness of the particular context and form of those relations is very important for an appropriate assessment of their validity. In this section I address each of these three issues in the hope of promoting a fruitful dialogue.
To repeat a previous point, if dependence is a referential frame and not a precise element of theoretical statements, dependencia theory is not tested by assessing the claim that dependence affects development because in that claim dependence is a conceptual element. What then are the conceptual elements of the theoretical statements that constitute dependencia theory? This question is not altogether easy to answer because, as is charged by rigorous empiricists, there is a pervasive ambiguity in the verbal claims that constitute the theory, and because dependencia theory is not a clearly distinguishable theory per se. Indeed, it is probably inappropriate to refer to the body of literature and its knowledge claims as a theory at all.41
^Meaningful dialogue will not occur if dependencia theorists are adamantly opposed to rigorous empiricism in principle, as there is indication that some of them may be. See, for example, Cardoso and Faletto. All that can be done is to provide the possibility of dialogue.
41This point is made nicely by Richard R. Fagen, "Studying Latin American Politics: Some Implications of a Dependencia Approach," Latin American Research Review, Vol. 12, No. 2 (Spring 1977):
Dependency and dependencia theory
69
To answer the question, then, one must engage in some creative interpretation and piecing together of seemingly diverse arguments.
As the first stage of a research project under the direction of Professor Bruce Russett we have attempted a synthetic interpretation, a full reporting and justification of which is given elsewhere42 and hence is not reconstructed here. We believe that sixteen substantive concepts—not including a particular concept of dependence—constitute the theoretical core of the dependencia literature. Those concepts can usefully be grouped into four general types to permit a simplified representation of the basic character of the body of theory as we interpret it. This simple representation is given in Figure 1. For each of the four groups represented in Figure 1 there are several particular substantive concepts, examples of which are: 1) the extent of financial and technological penetration of the peripheral country as a primary aspect of capitalist economic and state imperialism; 2) the product diversity of the export sector as an aspect of the export enclave syndrome; 3) the extent of structural integration among economic sectors as a feature of constrained and distorted economic development; and, 4) the degree of coercive authoritarianism of the peripheral state as an important concept within the group labeled socio-political distortions.
The basic arguments that are offered in the dependencia literature are arguments about the various causal or conditional relations among these sixteen core concepts, relations both within and between the four groups. Those within the groups are not represented in Figure 1, but those between the groups of any real importance in the literature are represented by arrows in the figure. Each has a particular role in the overall literature referred to as dependencia theory. Relations designated as of type A—those explaining or accounting for the several conceptual features of the export enclave syndrome—are largely taken or accepted as providing an historical background by theorists, and are not regarded as phenomena about which to theorize except to the extent that the features of the export enclave syndrome are weakened or reduced in a particular country.43 Type B relations were the primary concern of Prebisch and ECLA scholars, in partial reaction to whom the dependentistas developed their early arguments.44 These, then, can be viewed as
3-26, an article which corresponds with this present paper in several respects. The point is also made in Cardoso and Faletto, and in Cardoso, "Teória de ía Dependencia o Analisis de Situaciones Concretas de Dependencia?''
42Raymond D. Duvall, Steven Jackson, Bruce Russett, Duncan Snidal, and David Sylvan, "A Formal Model of 'Dependencia' Theory: Structure and Measurement," (New Haven, Conn., 1977).
43The historical development of the export enclave syndrome is revealed in Stanley Stein and Barbara Stein, The Colonial Heritage of Latin America: Essays on Economic Dependence in Perspective (New York: Oxford University Press, 1970); Celso Furtado, Economic Development of Latin America: A Survey from Colonial Times to the Cuban Revolution (Cambridge: Cambridge University Press, 1970). A concern for the "new dependence" of multinational corporation investment in industrial sectors is a concern for the shift away from the dominance of the export enclave syndrome. See Dos Santos, and Osvaldo Sunkel, "Big Business and 'Dependencia': A Latin American View," Foreign Affairs, Vol. 50, No. 3 (1972): 517-31, for enlightening summaries of the current concern for the "new dependence."
44Raul Prebisch, Towards a Dynamic Development Policy for Latin America. (New York: United Nations, 1963). Economic Commission for Latin America, Development Problems in Latin America (Austin: University of Texas Press, 1970).
70
International Organization
Figuře 1. Theory
A Simplified Representation of the Concerns of Dependencia
capitalist
economic
and
state
imperialism
3
export
enclave
syndrome
fc
-x
4
constrained
and
distorted
economic
development
\1/   \l/
socio-political distortions
Dependency and dependencia theory
71
basically pre-dependencia concerns; unfortunately, they tend overwhelmingly to be the focus of alleged efforts systematically to test dependencia theory. Relations of type C were an important part of the reaction offered by dependentistas to ECLA arguments—constraints on and structural distortions of economic development, according to the dependencia literature, cannot be fundamentally and appreciably affected by alteration of the export enclave syndrome alone; rather the extent and forms of capitalist imperialist relations are determinative. Although an early part of dependencia theory, type C relations continue today to be a central feature of that body of literature.45 Type D relations are most generally associated with that variety of the literature attributable to A.G. Frank and his colleagues. As such they are not irrelevant to the dependencia tradition, but neither are they central to it.46 The reason, I believe, that neither type B nor type D relations are emphasized in dependencia theory is that, consistent with what was said above about type A relations, features of the export enclave syndrome are largely accepted as historical givens, and hence, are variable contextual constraints more than theoretical "operators." Type E relations are the primary concern of contemporary dependencia theory. They are an outgrowth and direct extension of the original concern with type C relations into the sphere of social-political configurations and practices.47 Finally, in completing the cycle ».relations of type F have not yet been much discussed, but they are now emerging as important foci of attention in the dependencia literature as concern grows for the ways in which political and governmental processes in
45Fernando Henrique Cardoso, "Imperialism and Dependency in Latin America," in Structures of Dependency, Frank Bonilla and Robert Girling, eds. (Stanford, Calif.: Institute of Political Studies, 1973): pp. 7-16; Anibal Quijano Obregon, "Imperialism and International Relations in Latin America," in Latin America and the United States: The Changing Political Realities, Julio Cotler and Richard R. Fagen, eds. (Stanford, Calif.: Stanford University Press, 1974) pp. 67-91; Celso Furtado, "The Concept of External Dependence in the Study of Underdevelopment," in The Political Economy of Development and Underdevelopment, Charles K. Wilbur, ed. (New York: Random House, 1973): 118-23; Osvaldo Sunkel, "Transnational Capitalism and National Disintegration in Latin America," Social and Economic Studies, Vol. 20, No. 1 (March 1973): 132-76; Susanne Bodenheimer, "Dependency and Imperialism: The Roots of Latin American Underdevelopment," Politics and Society, Vol. 1, No. 3 (1971): 327-59; Thomas E. Weisskopf, "Capitalism, Underdevelopment and the Future of the Poor Countries," Review of Radical Political Economy, Vol. 4, No. 1 (1972): 1-35; Peter Evans, "Continuities and Contradictions in the Evolution of Brazilian Dependence," Latin American Perspectives, Vol. 3, No. 1 (Spring 1976): 30-54, are English-language examples of this tradition.
46Andre Gunder Frank, Latin America: Underdevelopment or Revolution (New York: Monthly Review Press, 1969). Andre Gunder Frank, Capitalism and Underdevelopment in Latin America: Historical Studies of Chile and Brazil (New York: Monthly Review Press, 1967). Andre Gunder Frank, Lumpenbourgeoisie/Lumpendevelopment: Dependence, Class and Politics in Latin America (New York: Monthly Review Press, 1972). James D. Cockcroft, Andre Gunder Frank, and Dale L. Johnson, Dependence and Underdevelopment: Latin America's Political Economy (New York: Doubleday, 1972).
47Fernando Henrique Cardoso, "Associated-Dependent Development: Theoretical and Practical Implications," in Authoritarian Brazil, Alfred Stepán, ed. (New Haven, Conn.: Yale University Press, 1973), pp. 142-76; Anibal Quijano Obregan, Nationalism and Capitalism in Peru: A Study in Neo-Imperialism (New York: Monthly Review Press, 1971); Osvaldo Sunkel, "The Crisis of the Nation-State in Latin America: Challenge and Response," paper presented at the Latin American Studies Association Meeting, Austin, Texas, December 1971; Elizabeth Dore and John Weeks, "The Intensification of the Assault against the Working Class in 'Revolutionary' Peru," Latin American Perspectives, Vol. 3, No. 2 (Spring 1976): 55-83; Theotonio Dos Santos, "The Crisis of Contemporary Capitalism," Latin American Perspectives, Vol. 3, No. 2 (Spring 1976): 84-99, are English-language examples.
72
International Organization
peripheral countries can affect the structure of imperialist relations that condition and constrain them.48
According to this interpretation, if fruitful dialogue is to occur between depen-dentistas and rigorous empiricists, the latter must develop their test models in terms primarily of type C and type E relations, using type B and type D relations to define relevant contextual constraints and type A relations to account for changes in those contextual features. Thus, our test models should concentrate on the explanation of the structure and performance of the peripheral economy in terms of the extent and forms of capitalist penetration of it in the context of its changing export enclave syndrome, and especially the explanation of the structure and performance of the peripheral state in terms of capitalist penetration and economic structure and processes. But before such models can be developed appropriately two additional features of dependencia theory must be addressed, the issues of "history" and context, which clearly are related to one another.
It is important to keep clearly in mind that dependencia theory is fundamentally dynamic—it is concerned with "becoming," with structural "transformation," with "process." To be of any real intellectual importance, a test model must capture that dynamic character. But a short-term change formulation, such as one that focuses on first differences or momentary rates of change in theoretical concepts, is not the most appropriate to capture the relevant dynamics. Rather, what is needed is a formulation that reflects slow, historically-extended processes, because the theory is "historical" in the sense that it entails arguments that long-term processes of conditioning and determination have worked (and are working) over time to set the contemporary scene. Thus, contemporary features of the economy and polity of a peripheral country are affected, for example, by the extent and form of capitalist penetration of the country over the past 25, 50, even 100 years.
If we shift to the language of statistical models of the structural equation form, we can represent this kind of argument by a model of the following form:
[1] Yt = «oXt + aiXt_! + a2Xt_2 + . . . + axXt-x + [xt
where
Yt is the current value of some conditioned feature of the peripheral country, Xt,... , Xt_oo are the current and past values of the conditioning phenomenon, a0,... , «oo are the parameter values which represent the determining effect on
Y of each of the X's, and, fjit is an error term representing the imperfect character of the conditioning
relationship.
An obvious feature of model [1] is that the longer the continuous period of time
48Theodore Moran, Multinational Corporations and the Politics of Dependence: Copper in Chile (Princeton, N.J.: Princeton University Press, 1974); Cardoso, "Some New Mistaken Theses on Latin American Development and Dependency." The concern might be referred to as one for the possibilities of dependency reversal.
Dependency and dependencia theory
73
during which the peripheral country has been exposed to X (e.g., some form of capitalist penetration) and the greater the degree to which the periphery country is exposed to X at each point in time during that period, the greater (or lesser) will be the contemporary value of the peripheral feature (e.g., the product concentration of the export sector) that is conditioned by X. This same feature can be alternatively represented by:
[2] Yt = y0 Xt + 7l (Xt + Xt_2) + y2 (Xt + Xt_2 + Xt_2) + . . . + y«, (Xt + Xt_! +....+ Xt.«) + fit
which, when multiplied out, and with terms combined, becomes:
[3] Yt = (y0 + 7l + y2 + .. . + y«) Xt + (y2 + y2 + ... + y«) Xt_2 +
... + y» Xt_oo + /xť.
Model [3] is exactly equivalent to model [1] if
«o = 7o + yi + 72 + • • • + y»
«i =      7i + 72 + • • • + y»
a2 =                y2 + • • • + yoo
o^oo                                                               y oo
This means that if the y's are all positive (or all negative), which seems to be implied by the verbal theory, then a0>a1>a2>-- >«oo, which means that the current value of y is a function of current and past values of X with the past having a progressively diminishing impact. Alternatively and equivalently, the total impact on Y of any change in X is parcelled out over time with the strongest impact occurring relatively contemporaneously and progressively diminishing impacts occurring thereafter.
In general, one cannot test model [1] directly because precise data for X far back into history are frequently not available; nor could one generate efficient estimates of the structural parameters even if one did have the data. However, we can transform model [1] into an equivalent model for which efficient estimation by way of available data is possible if we make the assumption that the progressive decline of the a's over time is governed by, or at least can be reasonably approxi-
74
International Organization
mated by, a regular declining functional form. The simplest assumption, attributable to Koyck,49 is one of a simple geometric decline. That is,
[4] Wi = (1 - X) X1, and                 -      50
i
The parameter, X, then, describes the shape of the time-function, or historical process, by which X affects Y. It indicates how quickly the peripheral country "forgets" or overcomes its past.
Given an assumption of type [4] model [1] becomes, after lagging, multiplication and rearrangement of the terms,
[5] Yt = X Yt_! + a* (1 - X) Xt + /it - X^.
The estimation of model [5], in which the error term is now generated by a first-order moving average process, provides precise information about the nature of the historical process through which X conditions Y but requires data for only two points in time. Because ofthat, a test model such as [5] is far preferable to static and simple short-term change models for guiding á rigorous empirical test of dependencia theory.
The theory is also "historical" in a second sense which is the historical uniqueness argument of the contextual specificity of causal relations. An acceptance of this kind of historicist epistemology has led some dependencia theorists to reject explicitly the notion of theory in favor of (descriptive.) "analysis of concrete situations of dependence."51 If dialogue with rigorous empiricists is to occur, the extreme form of this epistemology must be rejected. But because the logic of an historicist epistemology is basic to the arguments that constitute the dependencia literature, it is important that the rigorous empiricist incorporate in tests the notion that context affects causal, or conditional, relations. To effectuate this requirement, it is necessary to interpret the verbal historicist arguments in terms of the important context-defining variables that are implied in the contextually-specific analysis. These context variables, Z, can then be incorporated in the test model as variables which affect or condition the causal or conditional relation between X and Y.
49Leendert M. Koyck, Distributed Lags and Investment Analysis (Amsterdam: North Holland, 1954). What I present here is a standard formulation for what are called distributed-lag models. These are nicely summarized in Zvi Griliches, "Distributed Lags: A Survey," Econometrica, Vol. 35, No. 1 (January 1967): 16-49.
50Model [1] can be written as
Y = <x*[W0Xt + WA-, + W2Xt_2 + ... + WooXt-oo] + /xt
where XfWj = 1. The a* term appears in subsequent formulations in this paper; it is used in the fashion consistent with this rewriting of Model [1]. For subsequent formulations, I assume that W0 = 0. That is, I assume that there are no instantaneous impacts of X on Y. 51Cardoso, " 'Teória de la Dependencia' o Analisis de Situaciones Concretas de Dependencia?"
Dependency and dependencia theory
75
The relevant context variables can be incorporated in the test either by using them as typological criteria—referent countries could be grouped into sets that are homogeneous on the context variables—or by specifying them in the general structural equation model as "state" variables that interact with the causal variables in determining Y. In general, the latter option is preferable because of the difficulty of achieving homogeneous sets of sufficiently large size for systematic analysis. But the preferred option leads to complex functional forms for representing the contextual interaction effects; nevertheless the representation must capture the notion that in such-and-such a context the effect of X on Y is different in a particular way from its effect in another context. Adequate test models must include complex interaction terms; the simple linear models that have been used to date allegedly to test dependencia theory cannot provide adequate tests because they necessarily ignore the importance of context in that body of theory.
We have attempted to put these various considerations into effect in developing a comprehensive and, we hope, adequate test model of dependencia theory. The full model is reported elsewhere.52 Here I present only one of the twelve equations that constitute the full model in order to give exemplary substance to the foregoing conceptual justification of the model. The twelve model equations are for the twelve concepts in the last three of the four groups represented in Figure 1; there are no equations for the four concepts in the group labeled capitalist economic and state imperialism—that is, we ignore relations of type F and assume that capitalist imperialism is explained by phenomena outside of the dependence model.
The example that I use is for one of the concepts in the fourth group in Figure 1, a concept that we call the degree of coercive authoritarianism of the peripheral state. The concern of theorists here is with the apparent growth in a number of peripheral countries of a statist class interest that largely develops out of a bourgeois control of the state but expands beyond that to a point where new class interests come to the fore. This is the emergence of a "state bourgeoisie" out of, and now apart from, the "national bourgeoisie." These statist class interests remain closely tied to the national bourgeoisie and to international capital but they come to have their own basis in the apparatus of the state—in the control of the means of allocation and coercion. As the military emerges as a politically dominant force and loses its class ties to the aristocracy and the national bourgeoisie, it responds in especially repressive fashion to manifest conflict in the society, generally acting to suppress labor and peasant class interests and to involve the state more deeply in the management and regulation of the peripheral economy. The issue, then, is an understanding and explanation of the emergence, and in some societies the dominance, of these statist class interests which are centered primarily in the military.53
52Duvall, et al.
53The need for such explanations is stated well by Cardoso in "Some New Mistaken Theses on Latin American Development and Dependency."
76
International Organization
No single source in the dependencia literature provides a complete and explicit explanation of the extent of coercive authoritarianism of the peripheral state.54 Thus the development of a test model equation must be based on a creative interpretation and synthesis of diverse arguments. Our understanding of the various theoretical arguments is that the emergence and development of coercive authoritarianism is due to two distinct historical processes. The first of these is the process of conflict, especially class conflict, in the society. Where conflict is intense, pervasive, and historically long-lived, the coercive apparatus of the state is strengthened and its governmental importance is increased. These political adjustments to conflict can and do occur fairly quickly. However, attention cannot be directed solely to short-term changes in conflict because the development of coercive authoritarianism is not in general a reversible process. This means that short-term declines in conflict are irrelevant, neither stimulating nor weakening the development of the coercive state. Hence, only the positive levels of conflict are relevant, not the direction of change in them. But recent values of conflict are clearly most relevant, so the time process should be one of quite rapid "forgetting," and recent sharp increases in conflict should, then, have the greatest impact. However, this impact is not argued to be uniform for all societies. Rather, the nature of the relationship is, itself, argued to be affected by the context provided by the particular peripheral society. In particular, the contemporary growth of coercive authoritarianism in response to past and present conflict is greatest where the state is currently actively and deeply involved in economic affairs, such as in Brazil, Argentina, and Mexico. In such situations, statist interests are particularly concerned with the control of destabilizing economic conflict. This means that there is an interaction between current state involvement in the economy and the historical process of conflict in determining coercive authoritarianism. Thus, in formal terms,
[6] Yt = f{Z,     (2 «Ä   )}
t i   .               t i
where
Yt is current coercive authoritarianism of the peripheral state,
Zx     is the immediately past value of the context variable, the extent of state
involvement in the economy, Xj     is the level of conflict in the society at time t-i back into history, «i is the impact parameter of conflict at t-i on current coercive authoritarianism, f represents some unspecified functional form.
54A number of works provide important arguments to be modeled, however. In particular, see Guil-lermo A. O'Donnell, Modernization and Bureaucratic-Authoritarianism (Berkeley: Institute of International Studies, 1973); Julio Cotler, "The New Mode of Political Domination in Peru," in The Peruvian Experiment, Abraham Lowenthal, ed. (Princeton, N.J.: Princeton University Press, 1975); Quijano Obregon, "Imperialism and International Relations in Latin America"; Cardoso, "Associated-Dependent Development: Theoretical and Practical Implications"; and, Abraham Lowenthal, "Armies and Politics in Latin America," World Politics, Vol. 27, No. 1 (October 1974): 107-30.
Dependency and dependencia theory
77
The second process that affects coercive authoritarianism is the political-military aid process, the extent to which the peripheral regime is and has been externally supported. If this process of political and military support is extensive, has been carried on for a long period of time, and is concentrated on a particular external source, the coercive apparatus of the recipient peripheral state is appreciably strengthened, its traditional domestic class ties and bases of support are weakened, and its political interests are made more homogeneous and similar to those of its dominant external source of support. All of this adds up to a probable increase in coercive authoritarianism. Thus, in formal terms,
[7] Yt = g {SfrXo    Z2   }
i               t  1        t  i
where
Yt is the current coercive authoritarianism of the peripheral state,
X2t_. is the extent of external political-military support for the regime at time
t—i back into history, Z2t_., the context variable, is the extent to which political-military aid at time
t—i was concentrated on a single external source, ßi is the impact parameter of external political-military support at time t—i on
current coercive authoritarianism, g represents some unspecified functional form.
If we combine [6] and [7], make some simplifying assumptions about f and g, and adopt the logic of model [4], we can generate a precise test model for the arguments of dependencia theory about the determinants of coercive authoritarianism of the peripheral state. For f and g we assume independent direct linear processes for which errors are random because this is the simplest formulation, and nothing more precise is consistently implied in the literature. Thus, we model the verbal theory as
[8] Yt = 0ZVi ácCiXít_) + IjSjZ^X^ + /xt
where elements are defined for [6] and [7], 0 is the context parameter for the effect that governmental involvement in the economy has on the causal relationship between conflict and coercive authoritarianism, /xt is the current extent of random error.
After lagging, multiplication, and rearrangement of terms, [8] becomes our precise test model:
[9] Yt = 0Ä   Xi    + 62Z1   Xi    + 03Z2   X2    +
04Z2t_2X2        +  05Yt-l  +  06Yt-2  +  €
78
International Organization
where elements are defined above, and
61 = a*0 - a*0X,
02 = a*0Xr - a*0T,    [62 < 0]
04  =   /3*tX  -  ß*T,       [04  <  0]
05  =   X  +  T,
06  = -Xr,              [06 < 0]
e = fjLt — 05Mt-i ~~ 06/^t-2> a second order moving average process, and X is the time-process parameter for the impact of past and present conflict on
coercive authoritarianism (equivalent to X in [4]) r is the time-process parameter for the impact of past and present political-military support on coercive authoritarianism (also equivalent to X in [4] but probably different than X in [9] because the time paths of the two causal processes are apt to be different). By estimating 61 through 06 in [9] and by using the above equations that define 0X through 06, we estimate theoretically relevant parameters, a*0,jS*,X, and r, which indicate the extent and temporal-historical form of the impact of the two causal processes on coercive authoritarianism.55
From this example it should be apparent that a systematic empiricism of de-pendencia theory is possible if a test model is carefully developed with an eye for the importance of history, relational context, and the particular substance of that theory as a concern for the structural transformations of peripheral societies in global capitalism. Such has not been characteristic of systematic empiricism to date. We hope that it will become so, in order to enable fruitful dialogue and possibly to prevent a premature and inappropriate rejection of the verbal theory. Its concerns are too important and of too lasting value to be treated lightly, simplistically, and faddishly.
55We cannot extimate a* and 0 separately, and, thus, we cannot speak of the precise form by which governmental involvement acts as a relevant context for the impact of conflict on coercive authoritarianism.