NATIONAL AND INTERNATIONAL SYSTEMS: A COMPUTER SIMULA'HON THESES FOR THE DEGREE 0F PH. D. MICHEGAN STATE UNWERSITY STUART ALLAN Barman; ‘ 1970 thesis entitled “ NATIONAL AND INTERNATIONAL SYSTEMS: A COMPUTER SIMULATION presented by Stuart Allan Bremer has been accepted towards fulfillment of the requirements for _Eh...D_.__degree in mm; Science I MW Major professor mam M70. 0-7639 ABSTRACT NATIONAL AND INTERNATIONAL SYSTEMS: A COMPUTER SIMULATION by Stuart Allan Bremer This research entails the development and evaluation of a computer simulation model of some aspects of inter— national relations. The model, the Simulated International ProcessER (SIPER) is an extension of the Inter—Nation Simulation (INS) model, a man-machine simulation. Decision— making participants have been replaced with sets of decision— making and information processing rules. These rules generate a variety of national behavior including revolutions, resource allocation, trade, aid granting, and diplomatic conflict. Arms races, economic growth and stagnation, and conflict spirals are just a few of the emergent properties of the model. Twenty—four five—nation international systems were created to evaluate the performance of the model. A series of hypotheses about the relationships between three attribute variables and fourteen behavior variables for real and simulated nations were tested. It was found that SIPER corresponds to the real world in about two—thirds of the relationships examined, while INS corresponds to the real world in just under one—half of the relationships. It appears that the behavior of SIPER—generated nations better approximates the behavior of referent nations than does the behavior of INS—generated nations. A comparison of some static, structural characteristics of SIPER, INS, and referent international systems suggests that the SIPER and INS systems correspond quite closely to the early nineteenth century European state system. Among contemporary referent systems, the correspondence for SIPER and INS systems is greatest with regard to developing systems. A comparison of some dynamic characteristics of SIPER and referent international systems was done using different time scales. Applying a time scale where one period of simulated time equals one year of real time indicates that the magnitude of change in the SIPER systems is much greater than that which has characterized the Western community in recent years. Further study of situations where rapid social change is found in referent systems suggested that the simulated international systems have more in common with referent systems preparing for war than referent systems suffering from economic depression. A time scale of one period of simulated time equal to one decade of real time yields better correspondence for the simulated systems and suggests that the model is better suited for generating long trends in behavior than short term variations. \ -. . .. .-.,.— - i' b lding shififdi-BHt . ' “qt." - Stuart Allan Bremer A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science © Copyright by STUART ALLAN BREMER 1971 ii ACKNOWLEDGEMENTS In a research project which borrows so heavily from the work of others, as this one does, those to whom acknowledgement is due grow so numerous as render the expression of gratitude an impossible task. Harold Guetzkow, Paul Smoker, and Rufus Browning have made special contribu— tions, and I wish to express my appreciation to them. Finally I wish to thank my wife Marsha who, through her careful editing, is chiefly responsible for any expository merit this work may contain. TABLE OF CONTENTS Page ACKNOWLEDGEMENTS. . iii LIST OF TABLES. . . . . . . . . . . . . . . . . . . . . Vi LIST OF FIGURES . . . . . . . . . . . . . . . . . . . .viii Qhag?erINTRODUCTION . . . . . . . . . . . . . . . . . . l II. THE SIMULATION MODEL . . . . . . . . . . . . . . 10 Section 1. Basic Economic Concepts and Relationships 2 Basic Political Concepts and Relationships 3. Basic Information Processing A. National Goals 5. Goal Attainment and International Trade 6. Goal Attainment, International Aid and Diplomatic Conflict III. THE EXPERIMENTAL INPUTS. . . . . . . . . . . . . 62 Section 1 Variable Initialization: Cross—System Constant 2. Parameter Settings: Cross—System Constant 3. Variable Initialization: Cross—System Variants 4. Parameter Settings: Cross—System Variants 5. The Experimental Design iv Chapter Page IV. SIMULATED NATIONAL SYSTEMS. . . . . . . . . . . 85 Section 1. The Behavior of National Systems The Method of Analysis The Attribute and Stability Variables The Attribute and Growth Variables The Attribute and Security Variables The Attribute and Cooperation and Conflict Variables National Attributes and National Allocations National Attributes and Behavior: A Summary CD \1 O\U-IJ:'LA)I'\) V. SIMULATED INTERNATIONAL SYSTEMS . . . . . . . . 155 Section 1 Comparing International Systems: Constraints and Limitations 2. The Correspondence Criteria: Measures of Systemic Behavior 3. System Structure and Static Correspondence 4. System Transformation and Dynamic Correspondence 5. The Effects of Parameter and Variable Settings 6. Conclusion IV. CONCLUSION. . . . . . . . . . . . . . . . . . . 234 Section 1. The SIPER Model in Perspective 2. International Autarky 3. Economic Stagnation 4. Military Escalation APPENDIX . . . . . . . . . . . . . . . . . . . . . . . 2U8 GLOSSARY . . . . . . . . . . . . . . . . . . . . . . . 282 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . 288 a a an. 1» .. f. 0 fl Table 10. 11. 12. 13. 1A. 15. 16. 17. LIST OF TABLES Common Initial Values for Basic Variables in Period 1 Variable Set I Variable Set II. Variable Set III Variable Set IV. Variable Set V . Variable Set VI. Alliance Change Schedule The Computer Run Schedule. Real World Nations Included in Sample. Correlations Between Attribute and Stability Variables. . . . . . . . Comparisons of Correlations for Attribute and Stability Variables. Correlations Between Attribute and Growth Variables. Comparisons of Correlations for Attribute and Growth Variables Correlations Between Attribute and Security Variables. . . . . Comparisons of Correlations for Attribute and Security Variables Correlations Between Attribute and Cooperation— Conflict Variables . vi Page 63 75 76 77 78 79 8O 81 8A 104 107 109 118 119 125 126 132 Table 18. 19. 20. 21. 22. 23. 2A. 25. 26. 27. 28. 29. 30. 31. 32. 33- Comparisons of Correlations for Attribute and Cooperation—Conflict Variables Correlations of Attributes and Allocation Behavior Comparative Propositional Summary of National Behavior . . . . . . . . . . . . . . . . System State Variables Rotated Factor Matrix of SIPER Systems . Contemporary Referent Systems. Rotated Factor Matrix of Changing Systems. Basic Simulated System Data. Analysis of Variance for Level of Consumption. Analysis of Variance for Level of Investment Analysis of Variance for Level of Defense. Analysis of Variance for Level of Trade. Analysis of Variance for Change in Consumption Analysis of Variance for Change in Investment. Analysis of Variance for Change in Defense Analysis of Variance for Change in Trade Page 1110 148 167 170 173 19A 218 220 221 222 223 224 225 226 227 Figure 1 2 CO\10\U‘IJ:‘UO 10. 11. 12. 13. 14. 15. 16. 17. l8. 19. LIST OF FIGURES Dimensions of Information Processing . . . . Simple Correlations Between Consumption, Investment and Defense . . . System Maintenance Activity. . . . . . . . System Expansion Activity. . . . . . . . . System Defense Activity. . . . . . . System Integration Activity. Past System Maintenance Activity Past System Expansion Activity Past System Defense Activity Number of Battles Involving Major European Powers, 1700— 1899. . . . . . . Past System Integration Activity Changes in System Maintenance Activity: Contemporary Trends. Changes in System Expansion Activity: Contemporary Trends. . . . . . Changes in System Defense Activity: Contemporary Trends. . . . . Changes in System Integration Activity: Contemporary Trends. . . . . Changes in System Activity: Severe Depression Changes in System Activity: Major War Changes in System Activity: Centenary Trends. Program Execution Sequence viii Page 27 142 174 177 178 180 183 185 187 189 190 201 202 206 210 214 249 CHAPTER I INTRODUCTION The research to be reported here entails the develop— ment and evaluation of a computer simulation model of international relations. The model has been named SIPER (Simulated International ProcessER) to acknowledge the generous support received from the Simulated International Processes project of Northwestern University and its director, Harold Guetzkow. We think it profitable to review some of the factors that led us to undertake this research, for these will help to establish the context within which the research is to be viewed. The Inter—Nation Simulation model, developed by Harold Guetzkow and his associates at Northwestern University,1 played a major role in the development of the SIPER model. It is perhaps inappropriate to refer to the INS model in the singular. The work that has been done with the Inter— Nation Simulation has produced a family of models. 1For a discussion of their early efforts see Harold Guetzkow, Chadwick F. Alger, Richard A. Brody, Robert C. Noel, and Richard C. Snyder, Simulation in International Relations (Englewood Cliffs, N.J.: Prentice—Hall, Inc., 19635. If we may be permitted to continue the analogy, we see three main blood lines emanating from the original work by Guetzkow. These blood lines correspond to the three fairly distinct purposes that the INS model has served: teaching, laboratory research, and testing and extension of the model itself. In the first lineage we find the Guetzkow and Cherry- 3 holmes version,2 the Skinner and Wells version, and William Coplin's World Politics Simulation“ to name just a few. Each of these efforts is directed at improving the initial INS model as a replicator of decision—makers' environments. Coplin, for example, has elaborated the internal aspects of the nation in an effort to more fully replicate the kinds of domestic pressures that are exerted on decision—makers. The primary objective in his efforts is to enhance for partici— pants the realism of the simulation. The second line of descent uses the INS to test in a laboratory setting certain types of experimental effects. These effects, such as nuclear proliferation, do not lend 2Harold Guetzkow and Cleo H. Cherryholmes, Inter— Nation Simulation Kit (Chicago: Science Research Associates, Inc., 1966). 3Donald D. Skinner and Robert D. Wells, Jr., Michigan Inter-Nation Simulation (Ann Arbor: The Department of Political Science and The Center for Research on Learning and Teaching, The University of Michigan, 1965). ”William D. Coplin, World Politics Simulation, II (Detroit: Department of Political Science, Wayne State University, 1967). :32 mm; .-_- .—'—-_——~— —. themselves to more traditional methods. Variations in the 5 INS model have been developed by Brody and Driver, Hermann 7 Meier and Stickgold,8 and and Hermann,6 Raser and Crow, Burgess and Robinson.9 The experiments using the INS model or one of its descendants are too numerous to recount, and their number continues to grow. The third lineage, and one that this work is heir to, is concerned with the evaluation and extension of the INS model as a theory of international politics. Most notable 5Richard A. Brody, "Some Systemic Effects of the Spread of Nuclear Weapons Technology: A Study through Simulation of a Multi—Nuclear Future," Journal of Conflict Resolution, Vol. VII, No. 4 (December, 1963), pp. 665—753. See also Michael J. Driver, "A Cognitive Structure Analysis of Agression, Stress, and Personality in an Inter—Nation Simulation" (Lafayette, Indiana: Purdue University, August, 1965). 6Charles F. Hermann and Margaret G. Hermann, "An Attempt to Simulate the Outbreak of World War I," American Political Science Review, Vol. LXI, No. 2 (June, 19 7 , pp. 400—416. 7John Raser and Wayman Crow, WINSAFE II: An Inter— Nation Simulation Embodying Capacity to Delay Response (La Jolla, Calif.: Western Behavioral Sciences Institute, July, 1964). 8Dorothy L. Meier and Arthur Stickgold, "Progress Report: Event Simulation Project—INS 16" (Evanston, Ill.: Simulated International Processes, Northwestern University, 1965). 9Philip Burgess and James Robinson, "Alliances and the Theory of Collective Action: A Simulation of Coalition Processes," in James N. Rosenau, ed., International Politics and Foreign Policy (New York: The Free Press, 1969), pp. 640—653. in this regard is the work of Richard Chadwick,lO Charles Elder and Robert Pendley,ll and, of course, Paul Smoker. The International Processes Simulation developed by Paul Smoker represents a quantum jump in the evolution of the Inter-Nation Simulation. It would be misleading to suggest that there has not been interaction between these three lines of descent. The teachers, experimentalists and modelers have borrowed from one another, and in some cases it would be difficult to identify their primary roles. In the case of this research, we have used the experience gathered by participating in and running the Inter—Nation Simulation in a classroom context, 10Richard W. Chadwick, "Developments in a Partial Theory of International Behavior: A Test and Extension of Inter—Nation Simulation Theory" (Evanston, Ill.: unpublished Ph.D. thesis, Department of Political Science, Northwestern University, 1966). 11Charles D. Elder and Robert E. Pendley, "Simulation as Theory Building in the Study of International Relations" (Evanston, Ill.: Simulated International Processes project, Northwestern University, July, 1966). , ”An Analysis of Consumption Standards and Validation Satisfaction in the Inter-Nation Simulation in Terms of Contemporary Economic Theory and Data (Evanston, Ill.: Simulated International Processes project, Northwestern University, November, 1966). ll Robert E. Pendley and Charles D. Elder, "An Analysis of Office Holding in the Inter—Nation Simulation in Terms of Contemporary Political Theory and Data of the Stability of Regimes and Governments" (Evanston, Ill.: Simulated International Processes project, Northwestern University, November, 1966). 12Paul L. Smoker, "An International Processes Simula— tiorm Theory and Description" (Evanston, Ill.: Simulated Irggernational Processes project, Northwestern University, 19 8). as well as the insight generated by the experimentalists and the extensive validation studies of the model itself.13 We think that this research, like Smoker's, represents a quantum jump in the evolution of the INS model. The two represent, however, in one fundamental sense, movements in different directions. Smoker, by extending the programmed aspects of the model, has greatly elaborated the national and international context within which participants are placed. This work seeks to program the previously unpro— grammed aspects of the basic Inter—Nation Simulation and render the INS model a ”complete" theory of international politics. We do not wish to debate the virtues of man—machine versus all—machine simulation models.lu We do not have sufficient information to make cost—benefit comparisons at this time. Our own experience suggests that computer models, compared to man—machine simulations, may be more costly, in time and money resources, in the development stage, but less 13For an excellent summary discussion and bibliography of the validation of the INS model, see Harold Guetzkow, "Some Correspondences between Simulations and ”Realities" in International Relations" in Morton A. Kaplan, ed., New Approaches to International Relations (New York, St. Martin's Press, 1968), pp. 202—269. lL‘For a comparison of three simulations, the Political— Military Exercise, the Inter—Nation Simulation, and the TEMPER computer simulation, see Hayward R. Alker, Jr. and Ronald D. Brunner, "Simulating International Conflict: A Comparison of Three Approaches," International Studies Quarterly, Vol. 13, No. 1 (March, 1969), pp. 70—110. trus- aim .19.le - 78h: «H1:- mum...»- .:.-.v «._' I ‘ ..... ~-‘ 5 dnumc -. om costly once they reach maturity. The matter of benefits raises the larger question of the purpose of this research. John Raser has described the motivation underlying the creation of ”skeletal” simulations in the following way. The researcher does not try to narrow his focus to one small segment or aspect of human social behavior; instead he tries to simulate a large and complex system, such as 'international relations.’ But he knows that he can not identify all the units or the relationships among them. So he selects those units and those relationships about which his information is greatest and, using them as a framework——as the 'bones'-—he builds a skeleton of international rela— tions. He hopes that by continually gathering more data in the field, by operating the simulation over and over and thus learning what is pertinent, he slowly will be able to flesh out the bones of his skeleton until someday he has a more complete simula— tion of the system in which he is interested. In the meantime, he must be aware that he has abstracted and cruelly abbreviated, that his simulation is to real life what a skeleton is to a living man. We feel that the all—computer mode of operation lends itself more readily to the incrementalist research strategy that Raser is talking about than does the man—machine mode of operation. The differences generated by a change in the model can be quickly observed and evaluated once the computer model has reached a certain level of maturity. Moreover, Raser's comments are directed at the heart of the underlying rationale for the present research. Our ultimate objective is to create a computer simulation model of international relations with which we can study the dynamics of international systems. This objective can only 15John R. Raser, Simulation and Society (Boston: Allyn and Bacon, Inc., 1969), pp. 27-28. be attained in a slow, evolutionary fashion, and the method of computer simulation seems well suited to this. The selection of an incrementalist research strategy explains, for the most part, why we chose to use an existing model, the Inter-Nation Simulation, as our point of departure rather than creating a totally new model like Technological, Economic, Military, and Political Evaluation Routine (TEMPER)16 or Benson's "Simple Diplomatic Game."l7 Our resources would not permit the creation of a model to rival TEMPER, and we feel that such an effort would be premature given our limited knowledge. According to Raser [TEMPER] has been defined as a completed simulation-— it is in the hands of those who must justify its existence by immediate use in policy development rather than as a vehicle for its own improvement. It seems likely that TEMPER will remain in its present state and that its weaknesses will be permanent draw— backs rather than takeoff points for improvement as is the case with the INS, which remains in the hands of its builders as a dynamic research technique.l8 Raser continues, "[AJny social simulation effort is not going to achieve high structural isomorphism at its early stages; the pertinent question is whether its builders are l6TEMPER: Technological, Economic, Military, and Political Evaluation Routine (Bedford, Mass.: Raytheon Company, Vol. 1—7, 1965—1966). 17 Oliver Benson, "Simulation of International Relations and Diplomacy" in Harold Borko, ed., Computer Applications in the Behavioral Sciences (Englewood Cliffs, N.J.: Prentice— Hall, Inc., 1962), pp. 574—595. l8Raser, op. cit., pp. 149—150. placing their highest premium on a 'product' or on a process."19 The computer model which serves as the basis for this research is not complete in the sense that TEMPER is. At best it is a partial theory of international relations embodied in a form that is intelligible to a computer. On the other hand, it is a step up from Benson's model, and, all things considered, we see the SIPER model as one of the most complex and consistent theories of international rela— tions in existence today. As we see it, there is a strong need for the develop— ment of computer models which have predictive power with respect to international phenomena. Jay Forrester argues that complex systems are counter—intuitive.2O By this he means that there is a strong tendency for choices which are based on experience gained from less complex systems to have the opposite effects from those intended. If this principle holds for the international system as well, it may be the case that complex computer simulation models offer us the only effective way of evaluating the long—term consequences of policy choices. 19Ibid., p. 150. 20Jay Forrester, Urban Dynamics (Cambridge, Mass.: The MIT Press, 1969). Chapter 7, "Notes on the Nature of Complex Systems," is particularly informative. We have tried to keep this introduction short, for the work itself is lengthy and complex. The next chapter presents the theory that the model embodies, and the following chapter indicates the parameter and variable settings that were used in the set of computer runs reported on in Chapters IV and V. Chapter VI is devoted to a discussion of the overall strengths and weaknesses of the model and the direction in which future research is to proceed. This is followed by an appendix con— taining the computer program and a glossary of terms frequently used. Laban 21:: 35013 mi 1:. ' 'vr'J I, I l i CHAPTER II THE SIMULATION MODEL The description of a complex model is always a difficult task, and to facilitate the understanding of this model, the description has been broken down into several sections. Sectionsl and 2 deal with some basic economic and political concepts and relationships. The substance pre— sented in these sections is derived largely from the Inter— Nation Simulation model,1 and the reader should thoroughly familiarize himself with these concepts and relationships before the later sections are attempted. Section 3 is concerned with basic information process— ing rules in the model, and their centrality is such that a discussion of this subject seems warranted before the decision—making processes in sections 4, 5 and 6 are dis— cussed. The decision processes in these latter sections are described in the order that they are executed by a simulated nation in the course of one period of simulated time. With lSee Harold Guetzkow, et a1., Simulation in Inter- national Relations (Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1963), Chapter 5, "Structured Programs and their Relation to Free Activity within the Inter—Nation Simulation," pp. 103—149. This is particularly illuminating regarding the structure and process of the Inter—Nation Simulation model. 10 11 reference to the total time cycle the reader should bear in mind that the concepts and relationships discussed in sections 1 and 2 define both the state of the national system in the pre—decisional stage and the consequences for the national system in the post—decisional stage. The consequences of decisions made in time period T constitute the pre—decisional situation in period T+l, as is character— istic of iterative models. Before beginning our description we should indicate some of the notational conventions that will be used in what follows. An effort has been made to allow the reader to refer to appropriate parts of the computer program contained in the appendix, and to facilitate this the relevant computer instruction numbers are contained in brackets where reference is appropriate. In the equations that follow T refers to the present period of time, I to the nation which is making the decisions, and J to a specific other nation which is the object of nation I's decision—making. In the case of multiply— subscripted variables, the first subscript is the source of action and the second is often the target of action. We will follow two conventions with regard to parameters. Those which are found in equations derived from the INS model will be found in the footnotes, while those in sections 3 through 6 will be discussed further in Chapter III. For ease of reference, a glossary of terms used in the model follows the appendix. 12 1. Basic Economic Concepts and Relationships The economic system is Keynesian in nature in the sense that there are three sectors of economic activity: consumption, investment, and government. Consumption refers to that sector of economic activity which is concerned with the production of "final goods" with which "the population replenishes or increases its energies and ministers to its wants and needs...."2 The value produced by this sector will be referred to as consumption satisfaction, or CS. Investment refers to that sector of economic activity which is concerned with the production of value which has the characteristic of being able to produce more value.3 We shall refer to this value as basic capability, or BC. For our purposes the government sector will be equated with that aspect of economic activity which is concerned with the maintenance of the internal and external security of the system.u Other governmental economic activities are considered to be either consumption, such as government transfer payments, or investment, such as subsidies to industries. The value produced by activity in this sector has the characteristic of being able to destroy 2Robert L. Heilbroner, Understanding Macro—Economics (Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1968), p. 13. 3 Ibid., p. 1A. “This assumption does not seem unreasonable since governments are typically defined as social institutions having a legal monopoly over the use of force. 13 other value. This will be called force capability, or FC. Value production occurs when resources are allocated to an economic sector. These resources, the factors of production, are represented in the model by a unidimensional measure of resource capability, total basic capability, or TBC. Allocation to the investment sector increases the resource capability of the national system in the future; hence, total basic capability may be thought of as the accumulated past basic capability (BC) production. The value produced by an allocation of TBC to an economic sector depends upon the size of that allocation and the efficiency of the economic sector. Each sector has a generation rate associated with it that states the output of the sector given a unit input of TBC. For example, the consumption satisfaction sector may have a generation rate of 1.4 for a particular nation, in which case an allocation of 100 TBC units will produce 140 units of CS value. Each nation has a set of three generation rates (CSGR, BCGR, FCGR), which may be considered analogous to what the economists call opportunity costs, and these rates will differ from nation to nation in response to their level of development and degree of specialization. There is a second kind of value accumulation which occurs in the economic system. Force capability value accumulates in such a way that at least part of the value produced in the present time period will be available for use in a future time period. The storage area for force 14 capability is called total force capability (TFC), and the level of this variable determines the amount of force capability which can be used in defense of the system at any time. These two value reservoirs, TBC and TFC, are assumed to depreciate. In other words, there is a flow from the reservoirs to entropy. Hence, to maintain constant levels, allocations to the investment and defense sectors are necessary. The rate of depreciation for TBC is either 2 per cent, 5 per cent, or 10 per cent, depending upon a stochastic determination in which each rate is given an equal probability of being used. The rates of depreciation for TFC are 20 per cent, 30 per cent, or 40 per cent, depend— ing upon a stochastic determination as discussed above. We can now establish some basic relationships in equation form. With regard to nation I at time T the amount of CS value produced is expressed as CS(I,T) = CSP(I,T) * TBC(I,T) * CSGR(I), (1) where CSP is the proportion of national resources allocated to consumption. Similarly the BC value produced is expressed as BC(I,T) = BCP(I,T) * TBC(I,T) * BCGR(I), (2) and the FC value produced is expressed as FC(I,T) = FCP(I,T) * TBC(I,T) * FCGR(I). (3) 15 CS value is completely consumed, but BC and FC value accumulate in the following ways: TBC(I,T+1) = TBC(I,T) + BC(I,T) — DBC(I,T) * TBC(I,T) and (4) TFC(I,T+1) = TFC(I,T) + FC(I,T) — DFC(I,T) * TFC(I,T), (5) where DBC and DFC are the selected depreciation rates discussed earlier. The setting of values for CSP, BCP and FCP constitute major decisions which have far ranging consequences for the simulated national systems, and it is to the nature of these consequences that we now turn our attention. 2. Basic Political Concepts and Relations In the previous section we discussed a set of decisions concerned with the allocation of resources to the production of value. This set of decisions involves the authoritative allocation of value, which, according to Easton, is the domain of the political system.5 The making of decisions necessarily entails the existence of a set of decision—makers, and in this context the term decision—makers may be thought of as parallel to the concept of elite. Whether we consider them the ”influential” 5David Easton, A Framework for Political Analysis (Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1965). 16 as Lasswell does,6 or the "active population" as Rashevsky 7 does, ”by an 'elite' we mean a very small (usually less than .5 per cent) minority of people who have very much more of at least one of the basic values than have the rest of 8 . . . ” In an Eastonian sense our "deCiSion— the population.... makers" are the authoritative allocators of the system. In the previous section we specified that economic activity in the consumption sector produced value that was consumed by the "population." These consumers, which, in conformity with the INS model, we shall call validators, may be thought of as the masses or non—elite. We need not be concerned at this level of abstraction with the question of who is and who is not a member of the elite and therefore a "decision—maker." We need only postulate that the popula— tion of a nation can be divided up for analytical purposes into those who have more and those who have less control 9 over the behavior of the nation. 6Harold D. Lasswell, "Introduction: The Study of Political Elites," in Harold D. Lasswell and Daniel Lerner, eds., World Revolutionary Elites (Cambridge, Mass.: The M.I.T. Press, 1965), pp. 4—6 7Nicholas Rashevsky, Mathematical Theory of Human Relations: An Approach to a Mathematical Biology of Social Phenomena (Bloomington, Ind.: Principia Press, 1947), pp. 148—49. 8Karl W. Deutsch, The Analysis of International Rela— tions (Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1968), p. 63. 9 Among the more recent works incorporating the distinc— tion between elite and mass is Ted R. Gurr, Why Men Rebel (Princeton, N. J.: Princeton University Press, 19 9 . l7 We begin our discussion of the programmed relationships between the decision—makers and validators with a considera— tion of the demands which the validators make of the decision— makers. [H2l—H38] The demands fall into two areas. 1) The validators expect a certain flow of CS value into their hands. 2) The validators expect a certain level of national security. The specification of these demand functions follows the formulations used in the Inter—Nation Simulation.10 With regard to the first demand, let us assume the existance of a minimum level of CS value flow below which the nation cannot go without ceasing to exist. This may be thought of as the subsistance level or simply the maximum deprivation that the validators will endure. We will call this variable CSmin, and it is a function of the CS value production potential of the nation (CSmax). CSmax is in turn a function of the value productive resources of the nation, TBC, and the productivity of the consumption sector, CSGR.ll CSmax = TBC * CSGR (6) The minimum CS value flow will be12 lOGuetzkow, op. cit., pp. 122-127. llIbid., p. 123. 12Ibid., p. 124. The parameter K is conventionally set at 380,000. l8 CSmin = (l—CSmax/k) * CSmax (7) ‘ I c CSmax and CSmin represent the maximum and minimum demands of the validators with respect to CS value flow. The validators give support to the decision—makers in response to the level of CS value flow at any point in time in relation to this minimum and maximum. This support is manifested in the variable of validator satisfaction with respect to consumption satisfaction, or VScs. As specified in the Inter—Nation Simulation, VScs is dependent on three factors. 1) 2) 3) For consumption near minimum consumption standards, validator satisfaction depends on the relation of consumption satisfaction to minimum consumption levels. Once minimum consumption standards have been met, larger and larger increases in consumption are necessary to produce corresponding changes in validator satisfaction. This saturation effect is more pronounced for 13 wealthier nations. 14 The formulation of this is as follows: VScs = l + r * (CS/CSmin—l) — v * CSmax/CSmin * (CS/CSmin—l)2 (8) l3 lL‘Ibid” p. 125. The r and v parameters are conven— Ibid. tionally set at 55.0 and 41.0, respectively. 19 In an aggregate sense, then, the support the validators give to the decision—makers is partially a function of the level of the CS value flow, the value productive resources of the nation, and the efficiency of the mechanisms that produce the CS value. The second area of validator demands is national security. [H53—H70] Here we postulate that the validators expect a distribution of world force capability favorable to their national security, as well as a favorable distribu— tion of potential force capability. The support the valida— tors give to the decision—makers in response to the satisfac— tion of this demand is called validator satisfaction with respect to national security, or simply VSns. However, in determining the distribution of world force capability, the validators do not perceive internal coercive forces as factors in their decision. Since total force capability includes forces for the control of external and internal systemic threat, we want to remove the force capability devoted to internal control (FCic) from the support equation. That equation i515 allies Z (TFC—FCic + a'*TBC) = * l vsns W 2 (TFC—FCic + a'*TBC) + b (9) non—allies The minimum value of VSns is 1.0, and the maximum is 10.0. A VSns of less than 1.0 indicates that the nation should be 15Ibid., p. 126. The suggested values for w, a‘ and b' are 3.0, 0.5, and 1.3, respectively. 20 considered "disengaged from the armaments race,"16 and a favorable balance of forces ceases to be a demand for the validators. In this case support is solely dependent on consumption flow. The aggregate support for the decision—makers, called VSm, is a weighted average of the two support factors dis— 17 [H71—H73] cussed above. VSm = e * VScs + g * VSns (10) It is clear that political systems differ in the degree to which decision—makers are dependent upon validator support for their continuation as decision—makers. The power to disregard the wishes of the validators is called decision latitude (DL).l8 Political systems with low decision lati— tude may be considered open,19 flexible,2O non—directive,21 or accessible.22 In any event, this may be considered a léIbid., p. 127. 17Ibid., p. 114. The suggested values for weights e and g are both 0.5. 18Ibid., pp. 115—117. 19James N. Rosenau, ”Pre—theories and Theories of Foreign Policy," in R. Barry Farrell, ed., Approaches to Comparative and International Politics (Evanston, Ill.: Northwestern University Press, 1966), pp. 27—92. 2OQuincy Wright, The Study of International Relations (New York: Appleton-Century—Crofts, Inc., 1955), pp. 543—553. 21Morton A. Kaplan, System and Process in International Politics (New York: John Wiley & Sons, 1957), pp. 54—56. 22Phillip M. Gregg and Arthur S. Banks, "Dimensions of Political Systems: Factor Analysis of a Cross—Polity Survey," American Political Science Review, LIX (1965), pp. 602—614. 21 structural variable which mediates the relationship between the decision—makers and validators. [H74—H91] Decision latitude is not necessarily constant. It is assumed that the validators will periodically seek to change the political system by making it more responsive to their wishes or demanding more leadership from the deci— sion—makers. In the model, a unit increment in DL, a unit decrement in DL, and no change in DL are equally likely outcomes of a stochastic decision process in any given period of time. The variable DDL introduces random shocks into the relationship between the decision—makers and validators, to which the system must adapt.23 Returning now to the question of the relationship between the degree to which the validators are satisfied and the stability of the political system, we assume, as INS does, that24 POH = a * (b—DL) * VSm + c * (DL—d) (11) 23In the original INS model, provision was made for the decision—makers to initiate increases in decision latitude. It was not included in this extension because an inspection of INS data indicated the option was seldom used by partici— pants and it was thought desirable to simplify the model somewhat by its exclusion. 2i¥Guetzkow, op. cit., p. 111. The suggested values for a, b, c, and d are 0.01, 11.0, 0.1, and 1.0 respectively. 22 POH, as it is used here, is a measure of the stability of the system, as suggested by Elder and Pendley.25 It will be recalled that in the VSns formulation, equation 9, there was a term FCic, or force capability devoted to internal control. The role of coercive forces in the control of internal threats to the political system is well established,26 and it is assumed that the decision— makers will alot some proportion of their total force capability to the performance of this function. The impor~ tance of this force will become clear when we consider another way in which the validators may manifest their support or lack of support for the decision—makers. Revolutions may occur in the simulated nations, and their occurrence is dependent on four factors. If the overall validator satisfaction, VSm, is above a revolution threshold, m, revolution is not considered possible.27 If this threshold value is not reached, then the probability of revolution is dependent upon the nature of the political system and the level of coercive forces, in the following 25Robert E. Pendley and Charles D. Elder, ”An Analysis of Office—Holding in the Inter—Nation Simulation in Terms of Contemporary Political Theory and Data on the Stability of Regimes and Governments" (Evanston, Ill.: Simulated Inter— national Processes Project, Northwestern University, November, 1966). 26Gurr, op. cit. 27VSm varies from 1 to 10, and the revolution threshold was set at 23 28 manner. * DL — k' * (FCic/TFC) + 1 h! ' PR = g The final decision as to whether a revolution occurs or not depends upon a stochastic decision process. Should a revolution occur, however, there are substantial costs to the national system. All force capability devoted to internal control (FCic) is considered lost in defense of the system. Furthermore, there are substantial losses in the productive capacity of the system; 20 per cent of the nation's total basic capability is assumed lost in the event of a revolution. On the other hand, there are benefits to be gained from a revolution in the form of momentary increases in the overall validator support. In the period following the revolution an increase in VSm of two units is credited and a one unit bonus is given in the period after that.29 It should be clear by now that we have described a set of conceptual variables that we may use to define the pre- decisional and post—decisional states of a simulated nation and a set of relationships which determine the transformation of the system given the outcome of the decisional stage. It is to this stage that we now turn our attention. 28Guetzkow, op. cit., pp. 130—131. The parameters g', k', and h' were set—at 0.1, 3.3, and 2.0 respectively. With these parameters the maximum ratio of FCic to TFC is 0.3. 29 Ibid., pp. 131—132. 24 3. Basic Information Processing In the following sections we will refer occasionally to the development of expectations by one nation as to the future behavior of another nation. In this section we will discuss how these estimates of future behavior are formulated.30 The central thesis of this section is that nations use information processing rules to forecast the behavior of other nations. Since much national behavior is, in part, antici- patory in nature, it is a matter of no small importance how future behavior is estimated. An underlying assumption of all the information processing rules to be discussed here is that the best estimate of future behavior is to be found in the analysis of past and present behavior. [Bl—B60] One of the simplest kinds of information processing rules involves the extension of the present into the future. For activity X, nation J's level of activity in the next time period, T+1, is X(J,T+1) = X(J,T) (13) This simple rule states that what is happening now is the best estimate of what will happen in the future. We have labeled this Rule 4. 30This and the following sections have benefited from the work of Richard Cyert and James March, A Behavioral Theory of the Firm (Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1963) and Charles Bonini, Information Processing and Decision-Making in the Firm (Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1967). 25 Rule 3 is a little more sophisticated in nature as it admits to the possibility of change, but change is viewed in a disjointed, incrementalist way.31 X(J,T+1) = X(J,T) + [X(J,T) — X(J,T—l] (14) In this rule the future level of behavior X for nation J is taken as the present level plus the last change in behavior. This rule is similar in many respects to the kinds of equations found in the Vietnam simulation model of Milstein and Mitchel.32 Information Rule 2, on the other hand, utilizes more past behavior than Rules 3 and 4, but, with regard to the expectation of change, it lies somewhere between the two. Rule 2 is T 2 x(J,K) =T— X(J,T+1) = EIITURIIIIII (15) where m+l indicates the number of time periods in the memory span. Rule 1 is considerably more sophisticated than the previous rules. It is based on the conception that behavior 31For a discussion of the disjointed incrementalist view see David Braybrooke and Charles E. Lindbloom, A Strategy of Decision (New York: The Free Press, 1963). 32See Jeffrey S. Milstein and William C. Mitchell, "A Quantitative Analysis and Predictive Computer Simulation," Peace Research Society (International) Papers, X (1968), pp. 161-213. They find that changes in behavior are often more significant than the absolute level of behavior. 26 over time manifests itself in the form of trends. The detection of these trends will enable the nation to esti— mate a future state. The rule is operationalized by the application of a linear regression of the past behavior on time. The rule states that X(J,T+l) = a + b * (T+l) (16) where a and b are the constant and slope of the regression line estimated with the standard linear estimation procedures on values of X prior to T+l. These four information processing rules are by no means exhaustive of the types of rules one could formulate, but they do represent the relevant aspects of two dimensions of information processing. The first dimension is concerned with the expectation of change. If one presupposes his environment to be relatively stable, holding few surprises, then one would be led to formulate his expectations as to the future state of that environment in a different manner than if the presupposition is the opposite. The second dimension we think important here is concerned with the presupposition of a signal—to—noise ratio. If one has great faith in the information upon which estimates of future behavior are to be based, then, again, we would expect predictions to be rendered in a different fashion than if that information were felt to be unreliable. Figure 1 illustrates these two orthogonal dimensions. Environment Presupposition Stable Rule 4: Rule 2: T X(J,T+l) = X(J,T) 2 X(J,K) _ K=T—m X(J,T+1) — t-m+l Accurate Approximate Information Presupposition Rule 3: Rule 1: X(J,T+1) = X(J,T) X(J,T+1) = a + b * (T+l) + [X(J’T) " X(J,T—l)] Unstable 28 We have placed the information rules in the quadrants they are representative of. The use of Rule 4 presupposes the existence of a stable environment and accurate informa— tion since it entails the dual assumptions that future behavior will be the same as present behavior and that pre- sent behavior is known accurately. Rule 3 similarly assumes that past and present behavior are known with accuracy, but the constancy of behavior is not assumed. Rule 2, on the other hand, assumes just the opposite of Rule 3, as does Rule 1 make the opposite assumptions from Rule 4. We will admit to one more rule which makes no binding presupposition about the stability of the environment or the quality of information it has to work with. This rule, Rule 0, may be considered the pragmatist's choice as it applies the criteria of what works best in the rendering of a forecast. Put in its simplest form, Rule 0 entails the use of that rule, among the four previously discussed, which hindsight suggests would have been the best one to use the last time behavior X was analyzed. Accordingly, the rule which best predicts the present level of behavior without the benefit of information about the present, is the one that is used to predict the future. Hence, we allow for the case where a nation may not constantly use the same rule when it becomes evident that there is a better one, and to this extent the nation is capable of learning and adapting with regard to forecasting. 29 The set of information processing rules is now complete. At the outset of a simulation run we can prescribe that all nations shall use the same rule or that each nation shall use a different rule. As we shall see, the set of simulation runs that will be reported on later involves only a limited exploration of the possible combinations of information rules. 4. National Goals Nations are open, complex, adaptive systems and, as such, their behavior is purposive. Their behavior is intended to reduce the perceived discrepancy between the present state of the nation and some desired future state of the nation. We are concerned here with the national defini— tion of that desirable future state. To define completely the state of a complex system, be it present or future, we would need a very large number of dimensions. Theory and prudence inform us, however, that it is essential that we carefully select a subset of these dimensions for scrutiny. The first goal area to be so isolated is political stability. Decision—making elites have as a major goal of their behavior the retention of their decision—making posi— tions. The elites will endeavor to use the resources of the political and economic systems they command to make their positions of command secure. This can, of course, have far ranging consequences. As Robert North pointed out, 30 During the summer of 19l4...the Austro—Hungarian leadership, feeling threatened by the spectre of Pan- Slavism, put forward the preservation of the dual Monarchy at all costs as their major policy goal.33 The second goal area that will guide the behavior of the decision—making elite of a nation is economic growth. The expansion of national productive capability has, parti— cularly in this century, been a major objective. Organski has stated, "wealth is [a] goal that is sought to some extent by every nation."3u The third end toward which national behavior is directed is national security. By this we mean that nations act to further the continuation of their existance in the face of real or imagined external threats. As Raymond Aron has noted, Each political unit aspires to survive. Leaders and led are integrated in and eager to maintain the collectivity they constitute together by virtue of history, race or fortune. Political stability, economic growth, and national security by no means constitute an exclusive set of national objectives. They are, however, quite universal among nations and clearly prominent in the literature of international relations. 33Robert C. North, "Decision—making in Crisis: An Introduction,” Journal of Conflict Resolution, VI, No. 3 (September, 1962), p. 198. 314A. F. K. Organski, World Politics (New York: Alfred A. Knopf, 1958), p. 57. 35Raymond Aron, Peace and War (New York: Praeger, 1967), p- 72. 31 Our task of specifying the goals that guide nations is far from complete, however. Singer has stated; ...goals and motivations are both dependent and independent variables, and if we intend to explain a nation's foreign policy, we cannot settle for the mere postulation of these goals; we are compelled to go back a step and inquire into their genesis and the process by which they become the crucial variables that they seem to be in the behavior of nations.3 In specifying the process by which goals are set and reset we have relied heavily on the work of Richard M. Cyert and James G. March.37 Their formulation of goal determination in the firm suggests a pattern for such behavior in all complex organizations, including nation—states. Organizations set levels of aspiration in areas of meaningful achievement, and in the short run seek to attain these levels. In the long run, however, these aspiration levels themselves are subject to change. The result of this process is a dynamic homeostatic equilibrium of aspiration and achievement. In what follows we will show how this formulation is applied in the political stability and economic growth goal areas. The Goal of Political Stability We have posited that decision—makers act to make their positions secure. The degree of security they seek at any 36J. David Singer, "The Level—of—Analysis Problem in International Relations” in Klaus Knorr and Sidney Verba, eds., The International System: Theoretical Essays (Princeton, N. J.: Princeton University Press, 1961), p.486. 37Richard Cyert and James March, op. cit., pp. 26—43. 32 given time we will call the nation's aspiration level for political stability (ALPOH). The current value of the aspiration level for political stability is dependent upon three factors: [020-030] 1) The past aspiration level for political stability. 2) The degree to which the past aspiration was achieved. 3) The achievement of a significant other with regard to political stability relative to one's own achievement. The relevant equation is, ALPOH(I,T) = ALPI * ALPOH(I,T—l) + ALPA * [POH(I,T)—ALPOH(I,T—1)] + ALPE * [POH(K,T)—POH(I,T)] (12) The first term embodies the assumption that goals change slowly and incrementally. Culture and tradition are inertia—generating forces, and the coefficient ALPI, aspira— tion level for POH inertia, indicates the influence that such societal factors have over the modification of goals. The second term is an adaptive or learning component in the formulation. It is a simple feedback loop with ALPA being the rate of adaptation. The coefficient is positive, hence over-achievement leads to a higher aspiration level and under-achievement leads to a lower one. The relationship between under—achievement and over—achievement is not 33 symmetrical, however. We shall assume that nations will more readily raise their aspiration levels given encouragement and more reluctantly lower them when failure is encountered. This asymmetrical adaptation is expressed in the following way. If POH is less than ALPOH, ALPA = ALPA/ALPAAS, otherwise ALPA = ALPA. (18) The coefficient of asymmetrical adaptation, ALPAAS, is assumed to be greater than one. The third component of the above equation is a demon— stration effect. That is, it is assumed that the achieve— ment of significant others with regard to political stability in relation to one's own achievement will condition the aspiration level. The coefficient, ALPE, may be considered the propensity to emulate. Since its value is assumed to be greater than zero, the aspiration level of a nation will be increased by the attainment of a higher level of political stability by another nation deemed significant. Nation K, the significant other, is chosen on the basis of similarity of resource capability. The nation which is most like the self nation with regard to resource capability will be selected for comparison. The Goal of Economic Growth The aspiration level for economic growth (ALGRO) is assumed to operate in the same manner as the aspiration level 34 for political stability. [031—034] The relevant equation is; ALGRO(I,T) = ALGI * ALGRO(I,T—1) + ALGA * [PDTBC(I,T)—ALGRO(I,T—l)] + ALGE [PDTBC(K,T)—PDTBC(I,T)] (19) where PDTBC is the rate of growth in resource capability, TBC. The interpretation of the components and coefficients is exactly the same as above. There is also an asymmetrical adaptation coefficient, ALGAAS, and the significant other nation, K, is selected using the criteria outlined above. The Goal of National Security Before we discuss the process by which nations set their aspiration levels for national security, it is essen— tial that we examine the way in which the international system is structured and stratified. Nations are assumed to be grouped into alliances. The model as it presently stands does not allow for the position of non—alignment. Furthermore, the international system is bipolar in nature, with a major power functioning as the leader or dominant member in each alliance. The perspectives of alliance leaders and alliance members are sufficiently different that their behavior with regard to national security questions deserves separate treatment. However, there are some common elements in their decision processes. National security is identified with 35 the ability to successfully counter the use of coercion by other nations. Hence when we speak of the aspiration level for national security, for both leaders and members, we will be referring to the level of defense that is considered adequate to counter external threat. We have modified Singer's now classic threat equation since national goal setting behavior is anticipatory in nature, i.e., the national desire is to be able to counter not simply present threat, but also future threat. Accord— ingly, our formulation equates expected threat to the pro— 38 In duct of expected intent and expected capability. section 3 we discussed specifically how these expectations are arrived at in the model. When a nation scans it‘s environment for possible threats to its security, it must be selective in its search pattern. The limitations of time and resources prevent the nation from treating all nations as potentially equally threatening. In the present model a simulated nation assumes that those nations in the international system with whom they have not entered into mutual security agreements are potentially threatening. The alliance structure serves as a guide to simplify the search for enemies. 38J. David Singer, "Threat—perception and the Armament- Tension Dilemma," Journal of Conflict Resolution, II,1 (1958), pp. 90—105. 36 The Alliance Leader When the leader of an alliance ponders the question of national security, [035-066] the nation perceives the ques— tion in terms of bloc security. As the leader of a bloc the nation takes upon itself the duty of evaluating the security position of its alliance vis a vis an opposing alliance. Leadership confers larger responsibilities than membership, and the security interests of the leader become intertwined with those of the group. Accordingly, the goal of national security merges with the goal of bloc security. The gap that the alliance leader watches closely, then, is the difference between the amount of threat expected from the opposing bloc and the amount of threat—countering 39 If the ability which his own bloc will have in the future. bloc's counter—threat capability will be adequate, then the alliance leader will be content with its current defense commitments and those of its allies. If, on the other hand, the counter—threat capability is not judged adequate, then a revision of alliance security policy will be sought. We will elaborate this more fully. Above we noted that nations were sensitive to expected threat, and we mentioned that expected threat was equal to 39Our formulation here is essentially like the Lagerstrom— North anticipated—gap model. See Richard P. Lagerstrom and Robert C. North, "An Anticipated—Gap, Mathematical Model of International Dynamics" (Stanford, Cal.: Institute of Political Studies, Stanford University, April, 1969). 37 the product of expected intent and expected capability. One of the differences between alliance leaders and alliance members is their estimate of expected intentions. Intent in the threat calculation may be considered the probability that a given nation's force capability will be used against one's own nation, or it may be considered the proportion of a given nation's force capability that will be used against one's own nation. Another formulation might specify that expected intent is equal to the product of the expected probability of attack and the expected size of the attack, expressed as a proportion of the attacking nation's total force capability. In any case, it seems worthwhile to set the upper bound of intent at 1.0 and the lower bound at 0.0. The assumption in the case of alliance leaders is that expected intent is always at the maximum value of 1.0. The reasons for this are several. It is assumed that the special responsibilities of leadership make a nation more cautious in its security calculations, and therefore it is likely to want to be able to counter the worst of all possible situations. It can do so, in part, because of its larger resource base and the associated consequence of being able to work with such a pessimistic view without being over— whelmed. And, of course, there is some realism contained in the special paranoia of alliance leaders. Their prominence and centrality in the international system make them primary targets for other nations. 38 The calculations that an alliance leader makes are as follows. The amount of threat the opposing alliance is likely to present in the next time period (OPOW) is OPOW = Z TFC(J,T+1) (20) j=non—allies The expected value of TFC(J,T+1) is given by the application of one of the information processing rules described in section 3 to data concerning nation J's past behavior with regard to total force capability levels. The counter—threat capability (APOW) of the leader's own alliance is APOW = TFC(I,T+1) + Z TFC(J,T+1) (21) j=allies If APOW is greater than OPOW, then the leader's aspiration level for national security (ALSEC) will remain unchanged. If this relationship does not hold, then a series of steps are undertaken to formulate a defense policy for the alliance that will close the gap. The leader first considers the amount by which the alliance, as a whole, must increase its military strength to counter the expected threat. The leader then computes the share of the increase that each ally should contribute based on its resource capability. The leader then modifies its aspiration level for national security in accordance with what it considers its fair share of the additional defense burden. In addition, it transmits cues (FCCUE) to 39 its allies suggesting to them what would be an appropriate level of defense allocation. This completes the determina— tion of the aspiration level for an alliance leader. The Alliance Member The alliance member, like the alliance leader, reacts to the expectation of threat. [C67—Cll3] For the most part, however, the member cannot afford to assume the worst, and it is not so much concerned about the security of the bloc as its own security. These and other factors compel the alliance member to be more discriminating in its assessment of threat. To do so, the alliance member examines the verbal conflict behavior (HOST) of each non—allied nation to make estimates as to the future intentions of these nations. The expected threat (AD) is computed in the follow— ing way. AD = Z ALSID * HOST(J,T,T+1) * TFC(J,T+1) (22) j=non—a11ies where TFC has the same meaning as above, and HOST(J,I,T+1) is the amount of verbal hostility that nation I expects to receive from nation J in the next time period. ALSID is a parameter which indicates the propensity to discount verbal statements when estimating intentions. The value AD is our estimate of what level of force capability the nation would need in order to give itself reasonable unilateral protection. This value is then con— verted into units indicating what proportion of the national 40 resources would have to be allocated to defense in order to match this threat. The alliance member now has two estimates as to what it should allocate to defense. Previously it has received a cue (FCCUE) from its alliance leader suggesting a certain level of armaments which would be appropriate according to the leader's assessment of the world situation. The member also has its own estimate based on his own observations of the world. Reconciling these views and setting a national security aspiration level can occur in one of two ways. If the leader's estimate is less than or equal to the member's own estimate, then the member acquiesces and accepts the leader's policy. If, on the other hand, the alliance leader's esti— mate is more pessimistic than the member's and the leader's estimate is greater than the member's, a negotiation process is begun, the outcome of which is determined in the following way. The outcome of the negotiation process is dependent upon the amount of power that the leader exercises over the member at the time of the negotiation. Etzioni identified three basic types of power in his discussion of political integration.”0 These are1nfilitariwior economic power, identive or ideological power, and coercive or military power. qumitai Etzioni, Political Unification (New York: Holt, Rinehart and Winston, 1965). 41 Working with this power typology Denis Forcese found that only the first two of these were effective in coordinating the behavior of alliance leaders and members.b'l Forcese's findings were based largely on data generated by the Inter— Nation Simulation model. We have made the outcome of the bargaining process dependent upon the amount of utilitarian and identive power that the leader exercises over the member. Utilitarian power (UPOW) is defined here as the degree to which the member is economically dependent upon the leader. UPOW varies from 0 to 1.0 and is computed in the following way. ”Ml—:1 TRADE(IL,I,K) + AID(IL,I,K) K 1 T N (23) Z Z TRADE(J,I,K) + AID(J,I,K) K=1 J=1 UPOW where T is the current time period, N is the number of nations in the international system, IL is the leader of the member's alliance, and TRADE(i,j,k) and AID(i,j,k) are the amounts of trade and aid sent to nation j by nation 1 in time period k. Identive power reflects a kind of moral suasion that an alliance leader can exert by the manipulation of symbolic rewards. We postulate first that the greater the ideological difference between leader and member, the less identive power 41 Denis Forcese, "Power and Military Alliance Cohesion" (St. Louis, Mo.: unpublished Ph.D. thesis, Department of Sociology, Washington University, 1968). 42 the leader will be able to exercise over the member. Consequently our preliminary formulation of identive power is: IPOW = l. _ DL(I,T)—DL(IL,T) (24) W IPOW varies from 0 to 1.0, and IL is member I's alliance leader. DL is, as we stated earlier, decision latitude. It is felt, however, that identive power will be maximally effective when the world is ideologically polarized and less effective as perfect polarization is departed from. Consequently IPOW is modified by a term which takes into account the degree of polarization. Perfect polarization is defined as one—half of the nations on each end of the deci— sion latitude continuum. This distribution gives maximum variance, and it is the standard deviation of this distribu— tion which we use to quantify perfect polarization. The actual polarization is defined as the standard deviation of the distribution of decision latitude values among the nations in the international system. Hence our effectiveness measure is N 2 Z [DL(I,T) — DL(T)] I=l IEFCT = N (25) [N—l + .51 * 5.4 + [H + .5] * 3.6 [2 l N £2 1 N where N is the number of nations. This produces a value between 0 and 1.0 which indicates how polarized the world is. 43 Finally, IPOW is modified in the following fashion. IPOW = IPOW * IEFCT (26) IPOW still ranges from 0 to 1.0. The compromise that is forged between the alliance leader and member, if only tacitly, is based upon the average of IPOW and UPOW in the following manner. ALSEC(I»T) = ALSEC(I,T) + [FCCUE(IL,I)—ALSEC(I,T)1*[939HEEEQN] (27) Consequently if leader IL had absolute power over member I (UPOW and IPOW equal to 1.0), member I would raise its own estimate of security needs, ALSEC(I,T), to the level, FCCUE (IL,I), suggested by its leader. Proportionately less power means proportionately less increase. At this point we have completed the setting of aspira— tion levels for the simulated nations. We now turn to the consideration of how these aspiration levels are to be attained. 5. Goal Attainment and International Trade At this stage each nation has a set of aspiration levels it wishes to attain. The next thing the simulated nation does is to operationalize these goals in terms of laying out a tentative resource allocation budget. [0115—0131] It makes preliminary estimates as to what proportion of its resources will have to go to consumer goods (CSP) to achieve uu its aspiration level for political stability (ALPOH) in the following way. CSP(I,T) = CSP(I,T—1) + ALPOR * (ALPOH(I,T)—POH(I,T)) * CSP(I,T—l) + ECAF * PR(I,T) (28) The first term embodies the idea that such decisions are incremental in nature, and the second states that the amount of revision is related to the degree that goal achievement has failed in the past. The third term causes the system to react when political stability is threatened by crisis. Parameter ALPOR is the goal operationalization rate and ECAF represents the propensity of the regime to react to a crisis of support by acceding to the validator's wishes with an emergency CS allocation. The proportion of resources needed to achieve the growth aspiration level, ALGRO, is called BCP and is given by a similar equation. BCP(I,T) = BCP(I,T-l) + ALGOR * (ALGRO(I,T)—PDTBC(I,T)) * BCP(I,T—l) (29) PDTBC(I,T) is the proportionate change in total basic capability from time T—l to the present time T. Because of the way in which the aspiration level for national security was computed, the estimated proportion of resources needed for defense (FCP) is already known. Hence, FCP(I,T) = ALSEC(I,T) (30) 45 The remaining allocation decision involves the establishment of the proportion of current total force capability (TFC) that will be used for internal control (FICP). FICP(I,T) = FICP(I,T—l) + .3 * PR(I,T) (31) The maximum vale of FICP is .30, as it is in the Inter— Nation Simulation model. These four calculations complete the preliminary goal attainment decisions of the simulated nations. The decisions represent the systems' efforts to unilaterally fulfill their goal requirements, however they do not constitute the systems‘ final efforts. Another major alternative open to the national system is that of exchanging goods with other systems. Accordingly, the nation next explores the possibility of profitably engaging in international trade. International trade entails certain non—economic costs which we will call sovereignty costs. The kind of costs referred to here have been alluded to by Keynes. Let goods be home—spun wherever it is reasonably and conveniently possible....We do not wish...to be at the mercy of world forces....We wish to be our own masters, and to be as free as we can make ou selves from the interferences of the outside world. Jan Pen concluded, "...nationa1ism leads to protection, the deliberate choking—off of imports with the intention of 142John Maynard Keynes as cited in Jan Pen, A Primer of International Trade (New York: Vintage Books, 1966), p. 93. as .1 1..lpu,.. -. -. :g. 1e35the state trading-monopo The simulated nations set import limits for each of the three kinds of goods. [Cl32-Cl38] These import limits (IMLIM) are a function of the size of the national economy and the particular priority that a given good has in the tentative allocation mix. The import limit on all imports (TOTIM) is TOTIM = ITAF * TBC(I,T) (32) where ITAF is the international trade autarky factor, or propensity to import. The import limit for a specific good, such as BC's, a type 2 commodity, would be B P(I T) C , IMLIM(I,2) CSP(I,T)_C§MF(I)+BCP(I,T)+FCP(I,T) * TOTIM (33) where CSMF(I) is the proportion of national resources that must be allocated to the CS sector to satisfy the CSmin requirement. In addition to deciding in the preliminary trading stage how many foreign goods will be allowed to enter the nation, the nation must also decide the prices at which it is ”31bid., p. 94. 47 willing to sell its goods to other nations. The national set of export prices (EXPRC) is a function of a series of factors. [0139—0158] The basic unbiased price, or what might be considered the equal profit price, is given by the following formula. GR(J,L)*GR(I,L)*[GR(I,K)+GR(J,K)1 (3”) EXPRC(I,J,K,L) = GR(I’K)*GR(J,K)*[GR(I,L)+GR(J,L)1 This formula states the amount of good L that nation I would want from nation J in exchange for one unit of good K. GR(I,1), GR(I,2), and GR(I,3) are nation I's generation rates for CS, BC, and FC goods respectively. The terms of trade given by this formula are such that nation I and J would derive an equal amount of profit by concluding a trade on these terms. The fifty—fifty profit split appears to be a powerful norm in human interaction. Simmel, Durkheim, Homans, and Schelling are just a few of the authors who have noted is prominence.Ml In referring to the INS trade negotiations, Sherman commented, "these findings demonstrate the pervasive— ness and importance of the fifty—fifty profit splitting norm "145 for the prediction of the negotiation outcome. 44 Georg Simmel, The Sociology of Georg Simmel, trans. and ed. Kurt H. Wolf (Glencoe, 111.: Free Press, 1950). Emile Durkheim, Professional Ethics and Civic Morals (London: Routledge and Paul, 1957). George C. Homans, Social Behavior: Its Elementary Forms (New York: Harcourt, Brace and World, 1961). Thomas C. Schelling, The Strategy of Conflict (Cambridge, Mass.: Harvard University Press, 1960). uSAllen W. Sherman, "The Social Psychology of Bilateral Negotiations" (unpublished Masters dissertation, Department of Sociology, Northwestern University, 1963), p. 47. 48 On the other hand, we have reason to suspect that there are factors which produce a departure from the equal profit price. The first we shall consider is the preference among allies for trading with one another. If we may define alliances as formal agreements to be responsive to one another, then we may postulate that this general state of responsiveness will lead allies to reduce their export prices to one another. Furthermore, it is reasonable to suggest that allies frequently interact and negotiate on a wide range of matters, and consequently the costs of trading are reduced.”6 By the same reasoning the costs of trading with nations that are not frequently inter- acted with, and to which one is not responsive, are higher. Consequently, EXPRC is modified in the following way. EXPRC(I,J,K,L) = EXPRC(I,J,K,L) — APPF * (2*ALLY(I,J)—l) *EXPRC(I,J,K,L) (35) where ALLY(I,J) equals 1 if I and J are allies and 0 other— wise. APPF is the alliance preference pricing factor, or, alternately, the price increase, expressed as a proportion, that a non—ally receives. Yet another factor in setting export prices is the relative economic strengths of the two nations involved. 46See Dean G. Pruitt, ”An Analysis of Responsiveness between Nations," JOurnal of Conflict Resolution, VI,l (March, 1962), pp. 5—18, for a discussion of the effects of frequent interaction between nations. 49 here are, of course, two ways that this can manifest itself. be UNCTAD version is that richer nations charge poorer ations more than they do other richer nations and are able 3 exploit poorer countries in this way. The opposite, or ATT, version, suggests that what Sherman observed in the VS trading patterns occurs in the real world. That is, a paternalistic attitude toward the smaller under—developed 47 Juntries" on the part of larger countries. The conse— 1ence of this would be richer nations charging poorer itions less than other richer nations. Both of these fac— >rs are embodied in the following modification of EXPRC. EXPRC(I,J,K,L) = EXPRC(I,J,K,L) + TBC(J,T) _ TBC(J,T)+TBC(I,T) .5] * EXPRC(I,J,K,L) (36) ESPF * f ' ESPF, the economic strength pricing factor, is greater ,an zero, then export prices are adjusted to the ability pay in such a way as to benefit poorer nations. If, on e other hand, ESPF is less than zero, trade prices are ased in favor of the wealthier nations. The final factor which may enter into pricing decisions that of risk. One of the fundamental principles upon which ade is based is trust. International trade involves two nds of risks for the nation involved. The first is that it u7Sherman, pp. cit., p. 32. 50 Tulfill its part of the bargain and the other nation 10t. Trading is never a simultaneous exchange, and Lsks involved require compensation. The second, more act, form of risk is that associated with the uncertainty : knowing whether the goods you sell will not directly lirectly inable the buyer to act contrary to your asts at some future time. Trading—with—the—enemy .ation is merely one form that this type of trade dis— 1ation can take. The objective of this factor is to [$8 the profit derived from trading with a hostile or 1sted nation in order to compensate for the increased ’8 Accordingly, the EXPRC is further modified. I,J,K,L) = EXPRC(I,J,K,L) + IRPF * W * Z HOST(I,K,T) K=l N Z HOST(I,K,T) + l * EXPRC(I,J,K,L) (37) K=l ,J,T) represents the hostile feelings that nation I for nation J at time T. The first devision term s relative hostility, or the degree to which it is trated, and the log term relates this to the total of hostility. u8For a recent discussion of the effects of hostility de see Richard E. Gift, "Trading in a Threat System: S.—Soviet Case," Journal of Conflict Resolution, XIII,4 ber, 1969), pp. 418—437. 51 Ens completes the setting of export prices for the lated nations. The actual trading may now begin. [D1— ] The trading process itself begins with each nation uting the foreign and domestic prices for each commodity, preparing a list of profitable trades, which is rank— red by profitability. A trade round consists of each am making a bid to trade a specific commodity to another Lfic nation for another specific commodity. After each 3n has bid its most profitable trade a check is made to ;f any trade offers have been reciprocated. If this is ;he case, then a new trade round is begun and the nations ‘or their next most profitable trade. This again is Iwed by a reciprocity check, but this check includes bids in both the first and second rounds. Trade rounds nue until all nations except one have either reached import limit and/or have no profitable trades yet to When this state is reached trade ceases. The reciprocation of bids entails the agreement on the of trade, and it is only the actual quantities that a to be determined. This is done by taking the smaller 3 two import limits involved and basing the quantity Lon on this. The quantities thus exchanged are sub— xi off the import limits of the two nations involved, 1e trading continues. >al_.Attainment, International Aid, and Diplomatic Conflict At this stage of the decision—making process the nation plxored the two major avenues of goal achievement open 52 t. The avenues of self—fulfillment and exchange—fulfill— are both attractive since they require compromising the pity of the national system in only small ways. Another rnative, requesting aid, involves a greater compromise 1e national integrity and is chosen only when the nation 5 itself in a serious situation. It is at this stage that the nations begin to evaluate 7 overall position with reference to all goal areas. 1 to this point, activity in each goal area was carried on >endently from activity in the other goal areas. The .em of goal conflict is considered at this point, and a .ution of any such conflict is sought. Before this evaluation can take place, however, the In must adjust its allocation decisions in accordance any trade commitments that may have been made. [E21—E52] process involves the shifting of resources into sectors export commitments have been made, away from sectors import commitments have been received. Consequently alues for CSP, BCP, and FCP, the tentative proportional ation decisions, may require some adjustment. The nation, at this point, considers the possibility it may have over—extended itself. [E53—E58] If the f CSP, BCP, and FCP is greater than one, the nation is with a deficit and a budget crisis. In the event of situation, the following processes are activated to re the crisis. [E59—E76] 53 First, a new emergency budget is drafted based on two s: the size of the commitment that has been made in ector as indicated by the tentative proportional tion decisions, and the national priority for each This latter set of values, which may be considered -crisis—resolution weights, indicate in a critical ion the degree of importance that is ascribed to each Accordingly, the revised CS allocation would be PSPRI*CSP(I,T) PSPRI*CSP(I,T) + EGPRI*BCP(I,T) + NSPRI*FCP(I,T) (38) T) PSPRI is the relative priority assigned to the area of .cal stability. The new values for BCP and FCP are .ated in an analogous way with EGPRI and NSPRI indicating lative priorities of economic growth and national .ty. The nation now has a budget that is acceptable but ‘cessarily desirable in terms of its consequences. The nation may find, for example, that the level of Lption flow that will result, after adding CS to be ed and subtracting CS to be exported, is far short of .t feels it needs to achieve its aspiration level for (cal stability. It may, under these circumstances, make ,est for a grant of CS value from another nation in the .ational system. [El27—El38] There are, however, some basic rules constraining ,s in making aid requests. We may, at some future time, o relax some of these constraints, but for the moment 54 are considered necessary simplifications. The first of these provides that alliance leaders may make aid requests. Such requests would compromise their tion of authority and undermine their prominence in r’alliance. Alliance members, on the other hand, when require aid, direct their requests only to the leader 1eir alliance. This is a constraint that we reluctantly 3e and resolve to relax in the future. Within these constraints a nation requests aid of Lfic commodities to the degree that its prior considera— s revealed a discrepancy between the value level needed :hieve a goal and the value level that is expected as a aquence of fulfilling budget decisions. An alliance leader, then, may be confronted with a . number of aid requests, and the rendering of decisions rning these requests proceeds in the following manner. F117] There are four factors which exert control over ranting—of—aid decisions. First consideration is given e leader's economic ability to fulfill the aid requests 8 received. If the leader has found that it can meet spiration levels and have uncommited resources remaining, 11 consider aid requests. If not, any aid requests it eceived will be ignored. [F21—F36] If, on the other the leader's surplus is sufficient such that all aid sts may be granted without sacrificing the attainment of H1 goals, it will grant all requests. 55 hithe event of a situation where the leader has a lus,but this surplus is not adequate to satisfy all the tequests, the leader must decide how much and to whom fill be given. There are three criteria that a leader to make such decisions. The first consideration for the leader is the degree to 1the requesting nation's economy and its own are inter- xdent. [F38-F50] The greater the economic linkages en the two nations, the greater the share of available esources the aid requesting nation will receive. The (mic linkage with nation J, TF(J), is TRADE(J,I,K) IIMH K 1 N T 2 z TRADE(J,L,K) L=l K=l TF(J) (39) T is the present period, N is the number of nations in ystem, and TRADE(i,j,k) is the exports of nation i to 1 j in period k. TF(J), which varies from 0 to 1.0, ites the degree to which nation J, the aid requesting 1, has concentrated its trade with nation I, its alliance ’a a The second consideration of the alliance leader is the to which the aid—requesting ally has followed its tions concerning defense policy in the past. [F50—F58] ecision is produced by an algorithm which has as its ial. component a Pearson product moment correlation ciern3. [Gl—G22] The TFC levels of the alliance leader 56 aid requester are correlated over time and the resulting ficient is modified such that an r of —l.0 indicates an ance fidelity value (AF) of 0 and a correlation of +1.0 ices an alliance fidelity value of 1.0. Finally, consideration is given to the need of the aid— asting nation as embodied in its request relative to 7 nations' requests. The total value of aid that nation Ll receive from nation I, TAID(J), is AIDREQ(J) L: allies * SURPLUS(I) (40) IJ) : AIDREQ(J) is the total aid requested from the leader Ltion J. This total aid is then divided up among the >us commodities in relation to the degree that each 1dity was originally requested by a nation. [F59—F138] The sequence of allocation decisions is completed with ,pportionment of any resources remaining uncommited into us sectors in proportion to the size of existing sector tments. [Fl72—Fl83] Provision is made for nations to express hostility d oneanother during each time period. [Fl72—F183] This matic or verbal conflict behavior is generated by a set uations relating the level of conflict, HOST, to several rs. The first component in the determination of hostility s is a reactive factor. The action—reaction phenomena 57 sbemifrequently discussed, and the work of Zinnes is 49 wicmMrly relevant here. Zonnes examined both histori— Landsimulate data and found a positive relationship wem1"x's expression of hostility to y and y's perception mfiTiendliness," and that "there is a positive relation— p between the perception of unfriendliness and the "50 ression of hostility. This would lead us to believe t HOST(I,J,T+1) = REACT * HOST(J,I,T) (41) 1 reasonable formulation. However, we will add to this ;c formulation the proposal that nations react to expected ;ility by anticipating how hostile another nation will be he future with the aim of deterring that behavior. rdingly, the basic information processing rules discussed ier are used to yield a value for HOST(J,I,T+1), and the tion becomes HOST(I,J,T+1) = REACT * HOST(J,I,T+1) (42) HOST(J,I,T+1) is an estimated value. The reader will 1 tfliat when Information Processing Rule 4 is used the qtuations given above are identical. 49Dina Zinnes, ”A Comparison of Hostile Behavior of ioerMakers in Simulate and Historical Data," World 1138, XVIII, 3 (April, 1966), pp. 474—502. SOIbid” p.471 58 Ewre are factors which work to repress the expression homfllity. We have included three kinds of buffers which fify Hm expression of hostile feelings. In some cases we MMTers serve to store hostility indefinitely, and in mr cases they serve as a means of displacing hostility )m mm nation to another. The first of these buffers is that which comes into y when there are great power differences between actor target. Rummel reports a significant positive associa— n between the discrepancy in military power between a r of nations and the level of threats, accusations, and 51 tests that pass between the nations. Brody, Benham, Milstein found that if a weaker simulated nation :eived hostility emanating from a stronger one, it was ; likely to respond with verbal hostility than if it 52 Lated from a weaker one. This finding has been ially supported in analyses using real world data, as Erich Weede reports, "powerful states are more likely ngage in verbal conflict activities than relatively "53 rless states. 51Rudolf J. Rummel, "A Social Field Theory of Foreign tict Behavior," Peace Research Society (International) 31, IV (1965), p- 143- 52Richard A. Brody, Alexandra H. Benham, and Jeffrey fiQLstein, "Hostile International Communications, Arms .ctixon, and Perception of Threat: A Simulation Study" ;for%i, Cal.: Institute of Political Studies, Stanford Insity, July, 1966). 53 Erich Weede, "Conflict Behavior of Nation—States." defilivered at the Midwest meeting of the Peace Research ty' (International) on April 17, 1969, p. l. 59 Mm power buffer term we offer is; TFC(I,T+1) — TFC(J,T+1) TFC(I,T+1) + TFC(J,T+1) PBUFF * a1naficn I's expected military strength, TFC(I,T+1), is my nation I will suppress the expression of the amount hostility indicated by PBUFF. The second buffer operating in the hostility expression ation may be considered an alliance tolerance factor. The eral formulation of this factor is such that a nation will ore and/or not react to an ally's hostile behavior if it s not exceed a certain threshold value, ABUFF. Zinnes 1d that there was a tendency for a nation to perceive less :ility from an ally than a non—ally and to express less 54 :ility to an ally than a non-ally. Consequently we .ude a term ABUFF * ALLY(I,J) (AA) e ALLY is 1 when I and J are allies and 0 otherwise. a are two reasons why we feel the inclusion of this term ecessary. First, it would seem that allies should be ready to perceive hostility from one another and less gtive to any hostility that is perceived. Secondly, ;aiJiing the unity of the alliance requires a certain .t of‘ "turning the other cheek," and the parameter ABUFF atens how much hostility will be ignored. 5L1 Zinnes, op. cit., pp. 484—486. 60 Hm third buffer factor and final component in the stiflty equation concerns the effect of close economic es. The reasons for including this term are similar to asegflven above for the alliance buffer factor. The rongn°the economic dependence, the more effort will be 1e torepress the expression of hostility, up to a thres- _d value, for the sake of maintaining economic ties. arelevant formulation is EBUFF * NTRADE(J,I,T) (45) Z TRADE(K,I,T) K=l re TRADE(i,j,k) is the flow of goods from nation 1 to Lon j in period k, and the parameter EBUFF indicates the 1nt of hostility that would be ignored if all of nation imports came from nation J. The complete hostility equation for nation I with rence to nation J at time T is HOST(I,J,T+1) = REACT * HOST(J,I,T+1) TFC(I,T+1) — TFC(J,T+1) TFC(I,T+1) + TFC(J,T+1) - PBUFF * — ABUFF * ALLY(I,J) TRADE(J I T) _ *.____:__2___ EBUFF N z TRADE(K,I,T) K=l (46) calculate the conse— Sectien 1 and 2 are called upon to ences=of'the-decisions that have been made and define the :uational context for the next round of decision—making. Our description has not included the more technical ects of the computer model. We have chosen rather to us on its theoretical aspects. Those who are interested for example, the various input and output options that model allows are directed to the appendix. We do not 1 to imply that such matters are trivial, for as anyone has worked with computer models knows, they often seem lemand a disproportionate amount of time. The formulations given in this chapter constitute what onsider to be the best that could be assembled given the As the resources traints of time, money and competence. nd, so will the model. CHAPTER III THE EXPERIMENTAL INPUTS The results reported in the next two chapters are ed mltwenty—four runs of the simulation model, each run ng ten periods in duration. This chapter is devoted to licating the values that were used as variables and ameters to generate these data. Sections 1 and 2 discuss variable and parameter settings which the scheduled nty—four system had in common. Sections 3 and 4 discuss variables and parameters that were varied within the schedule. Variable Initialization: Cross-System Constant The recursive nature of the model demands values for basic variables which have been borrowed from the er—Nation Simulation model. These variables are: total c capability (TBC), total force capability (TFC), valida— satisfaction overall (VSm), probability of office— ing; (POH), decision latitude (DL), and probability of liition (PR). This set of variables determines the pre— sicnial state of each national system, and it is the 3 (3f these variables at time T=l that we are concerned rnere. Each of the twenty—four systems is composed of‘ ruations, and the values given in Table 1 for these five 62 63 timm were the same for all twenty—four systems. TABLE 1 COMMON INITIAL VALUES FOR BASIC VARIABLES IN PERIOD 1 NATION {IABLE 1 2 3 4 5 ?BC 7500. 17000. 9000. 34000. 37000. lFC 100. 1000. 800. 2850. 2713. ’Sm 4. 4. 5. 6.5 6. ’OH .8 .6 .7 .7 .9 DL 6. 4. 3. 5. 7. PR 0 0 0 0 0 These starting values are based on ones used in the SAFE II Inter-Nation Simulation runs conducted by John er and Wayman Crow.l The TFC variable is a weighted sum their conventional and nuclear capability categories. shall have occasion to refer often to these series of runs in the setting of variables and parameters. The WINSAFE II INS runs were designed to explore the Lfications of the capacity to delay response, that is, "invulnerability and deliverability of a retaliatory :e after accepting the most devastating blow or series lows the initiator of a nuclear attack can deliver."2 1John R. Raser and Wayman J. Crow, WINSAFE II: An rudflation Simulation Study of Deterrence Postures Embody— Capacity to Delay Response (La Jolla, Cal.: Western vixxral Sciences Institute, July, 1964). 2Ibid., p. I—l. 64 xperimental intervention involved giving to nations for periods of time the capacity to delay response. The es in national behavior were then noted and conclusions about the effects of this capacity. This set of INS runs will be compared to the SIPER in the following two chapters, and we must keep in mind the behavior of the INS model is in part due to these a1 experimental conditions as well as to the basic e of the model. However, this set of data is the best is available at this time. The intervention effects r to be less pronounced than in the Brody—Driver INS for example.3 Two other variables that require initialization, but nique to the SIPER model, are the aspiration levels olitical stability and economic growth. All five nations with the same aspiration levels, and again, all twenty— systems begin with the same national aspiration levels. The aspiration level for political stability (ALPOH) ssigned an initial value of .8 for all nations. An ction of the cabinet meeting minutes from the WINSAFE II suggested that subjects seemed to desire a POH value of .8 on a scale of 0 to 1.0. Accordingly, this value was in this series of computer runs. 3The Brody—Driver runs, known as INS—8, are described chard A. Brody, "Some systemic Effects of the Spread of ar Weapons Technology: A Study Through Simulation of tinuclear Future," Journal of Conflict Resolution, VII, cember, 1963), pp. 663-753. 65 The initial value for the aspiration level for economic )Wth (ALGRO) was similarly determined. The value used was ve per cent growth per period. The use of this as the itial value for ALGRO seems reasonable both in terms of at the participants in the WINSAFE II runs appear to have nted and what economists today consider a "good" growth te for a national economy.Ll Assigning an initial value to the aspiration level for .tional security (ALSEC) is not necessary since the equa— .ons pertaining to ALSEC are not recursive with respect to JSEC. The value of ALSEC in any period is not directly apendent on any previous value of ALSEC, as was the case ith ALPOH and ALGRO. The alliance and influence structure variables require nitialization as well. All systems were initially set up ith the following bipolar configuration. ALLIES NATION 1 2 3 4 5 1 1 0 1 0 1 2 0 1 0 1 0 3 1 0 1 0 1 *4 0 1 0 1 0 *5 1 0 1 0 1 11 indicates that the row nation and column nation are ”Jagdish Bhagwati, The Economics of Underdeveloped Smnmries (New York: World University Library, 1966). 66 lied, and the asterisks identify the two leaders of the )posing alliances. Parameter Settings: Cross—System Constant As we noted earlier, each simulated national system s a generation rate or productivity coefficient associated th each of its economic sectors. The same set of national neration rates, listed below, was used for all twenty—four mulated international systems. GENERATION RATE NATION CSGR BCGR FCGR 1 1.2 1.0 1.5 2 1.0 1.2 0.7 3 1.1 1.0 0.7 4 1.4 1.3 2.0 5 1.4 1.2 2.1 ese rates are based on those used in the WINSAFE II simula— on runs. They are generally in conformity with those used other INS runs. The SIPER model requires the setting of five types of rameters: information processing, goal determination, goal erationalization, international exchange, and international stility. The information processing parameters were varied ross systems and will be discussed in section 4. 11 Determination Parameters The goal determination parameters were introduced in iee previously discussed equations: . '5 :‘léi. _.fi _. ALGI * ALGRO(I,T-l) + ALGA ALGRO(I,T) = * [PDTAC(I,T) — ALGRO(I,T—l)] + ALGE * [PDTBC(K,T) - PDTBC(I,T)] (19) AD = X ALSID * HOST(J,I,T+1) * TFC(J,T+1) non-allies (22) The parameters for the first and second of the above equa— tions are essentially the same, and we can discuss them con— currently. The ALPI and ALGI parameters are measures of the inertia of goals. These parameters indicate how the aspiration levels change when there is no pressure for goal change being exerted. This would be the case when POH(I,T) = ALPOH(I,T-l) and POH(K,T) = POH(I,T) in the first equation. One might ~ postulate that in the absence of stimulation, either through success or failure, goals tend to rise or fall, but here we have assumed that in the absence of such stimulation goals remain constant. Hence, the parameters ALPI and ALGI were assigned values of 1.0. 68 ALPA and ALGA are parameters that govern the rate of adaptation or the speed at which any gap between POH and ALPOH is closed. The determination of reasonable values for these two parameters is not a simple matter, and we have sought only tentative values. Since the variables ALPOH and ALGRO are conceptual constructs, there is little hope at this time of deriving rigorous measures of the dependent and independent variables and determining para— meter values by the use of standard estimation techniques. In view of this, we have relied on a less rigorous estima- tion procedure for the parameters ALPA, ALGA, ALPE, and ALGE. We assumed at the outset that one's own experience is twice as salient as the experience of others. Hence, 2*ALPE = ALPA and 2*ALGE = ALGA. Returning to the cabinet meeting minutes of the WINSAFE II runs, we sought to esti— mate the rates of change of goals among the participants. Through a series of trial and error curve fittings, the parameters ALPA and ALGA were finally assigned a value of 0.1 and ALPE and ALGE a value of 0.05. Although these methods seem primitive, the alternative of assigning these parameters zero values and thus prohibiting any change in goals seemed less attractive. The ALSID parameter was set at 0.5 for the following reason. An inspection of the WINSAFE II cabinet meeting minutes suggested that nations become alarmed about the hostile intentions of another nation when they receive two 69 or more hostile messages from it in one period of time. Statements of hostile intention are discounted by one—half when national security matters are under consideration. Goal Operationalization Parameters The goal operationalization parameters were introduced in the following equations. CSP(I,T) = CSP(I,T—l) + ALPOR * [ALPOH(I,T) — POH(I,T)] * CSP(I,T—l) + ECAF * PR(I,T) (28) BCP(I,T) = BCP(I,T—l) + ALGOR * [ALGRO(I,T) — POTBC(I,T)] * BCP(I,T—l) (29) ALPOR, the aspiration level for POH operationalization rate, is a scale parameter indicating the proportional increment in CS production that is necessary to achieve the desired POH level. We can determine a range of effective values for ALPOR with knowledge of the relationship that the level of CS production has to the determination of POH, but not a specific value due to other factors which determine POH. From this range of acceptable values the figure 0.25 was selected. This figure indicates that under the worst of all possible conditions, that is, ALPOH equal to one and POH equal to zero, the maximum increase in CSP would be 25 per cent. An inspection of referent consumption data suggested that this figure is a reasonable estimate of the upper limit of a nation's ability to shift resources from one rate, .gyfifiasnsemeequal that used to set the ALPOR parameter. International Exchange Parameters The international exchange parameters are used in the following equations. TOTIM = ITAF * TBC(I,T) (32) EXPRC(J,I,K,L) = EXPRC(I,J,K,L) - APPF * [2 * ALLY(I,J)—l] (35) * EXPRC(I,J,K,L) + ESPF * TBC(J,T) ' 5 TBC(J,T) + TBC(I,T) ‘ ' (36) * EXPRC(I,J,K,L) + JRPF * HOST(I,J,T) N z HOST(J,K,T) K=1 N * Loglo Z HOST(I,K,T) + 1 1 * EXPRC(I,J,K,L) (37) The international trade autarky factor, ITAF, was assigned the value of 0.10. Referent data suggests that on the average a nation's imports represent about twenty per cent 71 of the value of its_gross national product,5 while the appropriate level for INS nations is a good deal lower. The figure used was selected as a compromise between the two data sources. The APPF parameter, the alliance preference pricing factor, and the ESPF parameter, the economic strength pricing factor were varied across the twenty—four runs, and they will, therefore, be discussed specifically in section 4. The international risk pricing factor, IRPF, was assigned a value of zero for this set of runs for the reason that it was considered desirable at this stage to keep the loop between dyadic conflict, in the form of diplomatic conflict, and dyadic cooperation, in the form of trade, open. In the future, when we close this loop, we can fully understand the results of linking these phenomena. International Hostility Parameters The equation given below contains the parameters relevant to the transmission of hostility. 5Table 46, "Foreign Trade (Exports and Imports) as a Percentage of G.N.P." in Bruce M. Russett, et a1., World Handbook of Political and Social Indicators (New Haven: Yale University Press, 1964, pp. 162—165, indicates that the mean trade level for 81 nations is 38 per cent. 72 HOST(I,J,T+1) = REACT * HOST(J,I,T+1) — PBUFF * TFC(I,T+1) — TFC(J,T+1) TFC(I,T+1) + TFC(J,T+1) — ABUFF * ALLY(I,J) - EBUFF * TRADE(J,I,T) N z TRADE(K,I,T) =1 K (46) The response parameter REACT was assigned a value of 1.0 indicating neither a systematic propensity to escalate tensions by over—reaction nor to deescalate them by under— reaction, but rather to respond directly to the expected level of hostility. The power buffer parameter, PBUFF, was tentatively given the value 0.25 when an extrapolation of INS data suggested such a value as reasonable. ABUFF, the alliance buffer, was set at 1.0 indicating that a nation would ignore 1 unit of hostility if the nation of its origin was an ally. Finally, the economic buffer, EBUFF, was given a value of zero for the same reason that we assigned the international risk pricing factor, IRPF, a zero value. 3. Variable Initialization: Cross—System Variants There are several other variables that require initial values in addition to those discussed in section 1. They are: the proportion of resources allocated to consumption (CSP), investment (BCP), and defense (FCP), as well as the proportion of total force capability assigned the role of internal security (FICP). The dyadic trade, aid and hostility variables also require initial values. We must also specify 73 what changes we wish to make in the alliance structure out— lined in section 1. These variables constitute the set of decisions taken by the various nations in time period 1. The second period values of the variables total basic capability (TBC), total force capability (TFC), overall validator satisfaction (VSm), probability of office—holding (POH), decision latitude (DL), and probability of revolution (PR), are the consequence of these decisions, and they constitute the predecisional setting when the computer model is activated. Six different sets of initial values were used in the twenty—four runs. Tables 2 through 7 indicate the values in each set. These values are the results of the first decision—making period of six of the WINSAFE II runs. Since there was no trade or aid during the first period of any of these runs, the variables TRADE and AID were assigned zero values. Table 2 indicates, for example, that in variable set I nation 1 allocated its resources in the following way: 94.87 per cent went to consumption, 5.00 per cent went to investment, and 0.13 per cent went to defense. Thirty per cent of its total force capability was devoted to internal security, and no hostile messages were transmitted. Table 2 also indicates what the consequences of the decisions taken in period one were for variable set I. These values set the stage for action by the computer model. 74 Table 8 indicates the scheduled alliance changes for each of the six variable sets. These alliance changes mirror the changes that occurred in the six WINSAFE II INS runs that serve as our data base. The variable sets I through VI create systems moving in different directions from a common origin. In Chapter V we will consider the effects of setting the computer model down these diverse paths. 4. Parameter Settings: Cross—System Variants In this series of computer runs we have decided to vary two types of parameters. The first type regulates the manner in which expectations of future behavior are developed by a nation, and the second type governs the degree to which non—economic factors are introduced into international trade. In section 3 of Chapter II we discussed the various optional information processing rules that we could specify. Two of those rules have been selected for systematic considera— tion. Twelve simulated systems were generated using the null rule, Rule 4. This rule states that future behavior will be the same as present behavior. Rule 0, the pragmatic rule, was used in the remaining twelve computer runs. This rule specifies how an information processing rule is selected, rather than how information is to be processed. That is, the choice of Rule 0 activates a process by which each of the four basic rules are in turn used to estimate the present level of behavior utilizing information from periods past. 75 TABLE 2 VARIABLE SET I NATION VARIABLE 1 2 3 4 5 CSP .9487 .9588 .9300 .8824 .8919 BCP .0500 .0294 .0589 .1029 .0811 E FCP .0013 .0118 .0111 .0147 .0270 g FICP .2000 .2000 .1500 .0437 .1000 I HOST(I,1) 0 0 0 0 0 N HOST(I,2) 0 0 0 0 0 S HOST(I,3) 0 0 0 0 0 HOST(I,4) 0 1 0 0 1 HOST(I,5) 0 0 0 0 0 C TBC 7725. 13128. 9350. 33270. 35168. N TFC 105. 790. 810. 2865. 4057. E VSm 5.5 3.0 3. 4. 6. 8 POH . 8 6 E N DL 6 5. 3 5. 7. 8 PR 0 .4 0 s .— 76 TABLE 3 VARIABLE SET II NATION VARIABLE 1 2 3 4 5 CSP .9333 .9500 .9053 .8529 .9162 BCP .0400 .0329 .0778 .0824 .0865 E FCP .0267 .0171 .0189 .0647 .0297 I FICP .1500 .0000 .3010 .2000 .2000 I HOST(I,1) 0 1 0 1 1 N HOST(I,2) 0 0 0 2 0 S HOST(I,3) 0 1 0 0 0 HOST(I,4) 7 3 0 0 0 HOST(I,5) 3 1 2 0 0 c TBC 7050. 17332. 9320. 32360. 35700. N TFC 390. 1103. 873. 6340. 3240. E VSm 4. 5.5 2. 5. 5. 8 POH .8 .4 E N DL 6. 4. 3. 5. 7. E PR .55 O .55 O 0 77 TABLE 4 VARIABLE SET III NATION VARIABLE 1 2 3 4 5 CSP .9667 .9559 .9222 .8529 .8919 BCP 0 .0353 .0556 .0794 .0676 E FCP .0333 .0088 .0222 .0676 .0405 I FICP .150 .20 0 .0 .100 I HOST(I,1) 0 0 0 0 0 N HOST(I,2) 0 0 1 0 0 S HOST(I,3) 0 1 0 0 1 HOST(I,4) 2 2 1 0 1 HOST(I,5) 0 0 0 0 0 0 TBC 7350. 16530. 9120. 32620. 37060. N TFC 465. 850. 900. 6130. 5340. E VSm 5.5 5.0 3.0 4.5 5.5 U POH .8 .7 .5 .7 .9 8 DL 6.0 5.0 3.0 5.0 7.0 E PR 0 0 .65 0 0 78 TABLE 5 VARIABLE SET IV NATION VARIABLE 1 2 3 4 5 CSP .9000 .9559 .9444 .8824 .9189 BCP .1000 .0441 .0556 .0588 .0622 E FCP .0000 .0000 .0000 .0588 .0189 I FICP .10 .30 .10 .10 .10 I HOST(I,1) 0 1 0 1 0 N HOST(I,2) 0 0 1 0 0 S HOST(I,3) 0 0 0 2 1 HOST(I,4) 0 1 0 0 3 HOST(I,5) 0 0 0 2 0 0 TBC 6800. 17560. 9320. 33620. 34620. N TFC 90. 900. 720. 6190. 3660. E VSm 3. 5. 3.5 5. 6. 8 POH .7 7 .5 .8 8 DL 6.0 4.0 3.0 5.0 7. E PR .65 0 .65 0 0 79 TABLE 6 VARIABLE SET V NATION VARIABLE 1 2 3 4 5 CSP .9333 .9118 .9500 .9000 .8541 BCP .0667 .0706 .0500 .0588 .0689 E FCP 0 .0176 0 .0412 .0770 I FICP .02 .10 .10 .10 .10 I HOST(I,1) 0 0 0 0 0 , N HOST(I,2) 0 0 0 0 1 S HOST(I,3) 0 0 0 1 0 HOST(I,4) 1 2 0 0 1 HOST(I,5) 1 2 4 0 0 c TBC 7850. 17800. 9100. 33620. 39320. N TFC 90. 1110. 720. 5115. 7784. E VSm 6.5 2.0 6.0 4.0 6.5 8 POH .9 .5 .7 .7 .9 8 DL 6.0 4.0 3.0 5.0 7.0 g PR 0 .55 0 .65 0 80 TABLE 7 VARIABLE SET VI NATION VARIABLE 1 2 3 4 5 CSP .9200 .9118 .9111 .8824 .9027 BCP .0267 .0529 .0889 .0735 .0595 E FCP .0533 .0353 0 .0441 .0378 I FICP .10 0 0 .0441 .0378 I HOST(I,1) 0 0 0 0 0 N HOST(I,2) 4 0 0 0 0 S HOST(I,3) 0 0 0 1 0 HOST(I,4) 2 2 3 0 0 HOST(I,5) 0 0 4 2 0 C TBC 7150. 12984. 7970. 25576. 36660. N TFC 690. 1270. 680. 5235. 5013. E VSm 4.5 2.0 2.0 2.5 7.0 8 POH .8 .6 .4 .6 .9 8 DL 6.0 5.0 4.0 5.0 7.0 E PR 0 .7 .65 .7 0 81 TABLE 8 ALLIANCE CHANGE SCHEDULE VARIABLE FORMER NEw I SET PERIOD NATION ALLIES ALLIES . I 7 3 1,5 2,4 II 4 3 1,5 2,4 II 7 1 5 2,3,4 III 4 2 4 1, 3,5 III 6 3 1,2,5 4 IV 3 3 1,5 2,4 IV 6 1 5 3 ,4 2 3 , 4 5 VI 5 1,3 5 2.4 82 The most accurate of the four rules is then used to render a prediction concerning future behavior. In section 5 of Chapter II we discussed the way in which noneconomic factors may influence the prices at which nations are willing to sell goods to one another. Two of those factors, the alliance preference pricing factor (APPF) and the economic strength pricing factor (ESPF), were varied in this set of computer runs. Twelve runs were made with pricing parameters set equal to zero. Trade in these systems is based strictly on comparative advantage; i.e., the relative differences between national generation rates. A departure from pure comparative advantage was introduced in the other twelve simulation runs. The alliance preference and economic strength pricing factors were set equal to 0.20 and 1.0 respectively. The first of these settings indicates that allies will be charged twenty per cent less for goods and non—allies will be charged twenty per cent more. The second indicates that larger nations will be asked to pay more for goods by smaller nations and the converse. For this specific parameter setting the price increase or decrease is proportional to the relative sizes of the nations' resource bases. That is, the proportional change in the prices that nation I is willing to sell goods to nation J is TBC(I,T) TBC(J,TT + TBC(I,T) ‘ -5 (47) Generally we will refer to the pure comparative advantage 83 pricing as unbiased pricing, and the departure from comparative advantage will be referred to as biased pricing. 5. The Experimental Design The changes in variable initializations and parameter settings made in this schedule of computer runs were arranged in a factorial design. The model was run with each Of the six variable settings in conjunction with each of the parameter combinations. Table 9 indicates the character Of each of the twenty-four systems generated. We will consider the effects of these variations specifically in Chapter V. 84 TABLE 9 THE COMPUTER RUN SCHEDULE INFORMATION EXPORT RULE PRICING VARIABLE.SET. SYSTEM Pragmatic Unbiased I l Pragmatic Unbiased II 2 Pragmatic Unbiased III 3 Pragmatic Unbiased IV 4 Pragmatic Unbiased V 5 Pragmatic Unbiased VI 6 Null Unbiased I 7 Null Unbiased II 8 Null Unbiased III 9 Null Unbiased IV 10 Null Unbiased V ll Null Unbiased VI 12 Pragmatic Biased I 13 Pragmatic Biased II 14 Pragmatic Biased III 15 Pragmatic Biased IV 16 Pragmatic Biased V 17 Pragmatic Biased VI 18 Null Biased I 19 Null Biased II 20 Null Biased III 21 Null Biased IV 22 Null Biased V 23 Null Biased VI 2A CHAPTER IV SIMULATED NATIONAL SYSTEMS It has been noted that "...the problem of verifying simulation models remains today perhaps the most elusive of all the unresolved problems associated with computer simula— tion techniques."l Given this state of affairs we must nevertheless attempt an evaluation in order to estimate the model's validity and diagnose errors to be rectified in the future. Generally it has been concluded that validating simula— tions is a special case of the general problem of verifying models of all sorts. However, simulations like SIPER and INS pose special problems when ”whole sets of variables in the complex of national and international life are represented by simplified, generic factors, supposedly the prototypes of more elaborate realities."2 Hence, as Hermann concludes, "comparisons of the simulation's variables and parameters with their assumed counterparts in the observable universe... can be particularly troublesome when the definitions must 1Thomas H. Naylor et al., Computer Simulation Techniques (New York: Wiley and Sons, 1966), p. 310. 2Harold Guetzkow, "Structured Programs and Their Rela— tion to Free Activity within the Inter—Nation Simulation," in Harold Guetzkow et al., Simulation in International Rela— tions (Englewood Cliffs, N.J.: Prentice—Hall, Inc., 1963), p 85 86 correspond to a simulation variable or parameter that is either an analogue or a prototype intended to combine numerous features of the reference system."3 Computer models like SIPER, on the other hand, have the advantage of perfect internal validity.Ll That is, given than no "errors of functioning" occur, as Thuring calls them,5 any simulation run can be perfectly replicated given the same input, including, of course, the same string of pseudo—random numbers if the model has stochastic processes. It is more difficult to assess the "face validity" of complex computer models than it is for other more visible types of games and models. We can make evaluations of the apparent realism of specific components or processes in the model, but such assessments of the overall model are without foundation when so much that is going on is invisible to the observer. Observing the computer while the simulation is being run does not enable the researcher to get the feeling of realism or unrealism that one gets when one observes an Inter-Nation Simulation world in action. 3Charles Hermann, "Validation Problems in Games and Simulations with Special Reference to Models of International Politics," Behavioral Science, XII (May, 1967), p. 222. ”For a further discussion of types of validity see Hermann, op. cit., pp. 220—224, and Charles F. Hermann and Margaret G. Hermann, Validation Studies of the Inter—Nation Simulation (China Lake, Calif., U.S. Naval Ordinance Test Station, December, 1963), pp. 26—3A. 5A. M. Turing, "Computing Machinery and Intelligence," Computers and Thought, ed. Edward A. Feigenbaum and Julian Feldman (New York: McGraw—Hill, 1963), pp. ll—35. 87 Another type of validation that has been proposed is "event validity." The question becomes whether simulation produces the same discrete actions or occurrences that are observed in the referent system. Charles Hermann's simula— tion of the outbreak Of World War I is clearly a case for which "event validity" is an appropriate validation criterion.6 However, it does not seem appropriate here, as the desire is to reproduce patterns of behavior rather than to replicate particular actions or occurrences. We are not, however, without some guidance in this question of validation. Guetzkow has commented, ...by using some systematic rigor in making comparisons between simulations and realities, by taking reference data largely from extant international systems rather than from laboratory or field research about noninter- national phenomena, and by finding in simulations internal processes and outputs which correspond to reference processes as well as reference outcomes, a convergence of evidence is gained which increases the credibility of the theoretical construction of simula- tions. It is worthwhile to consider the statement made by Guetzkow concerning the nature Of the referent data. The SIPER model may be compared either to its parent, the Inter— Nation Simulation, or their common referent, the "real" world. The predominant strategy that will be followed here, however, 6Charles F. Hermann and Margaret G. Hermann, "An Attempt to Simulate the Outbreak of World War I," American Political Science Review, LXI (June, 1967), pp. 400—416. 7Harold Guetzkow, "Some Correspondences Between Simula— tions and 'Realities' in International Relations," New Approaches to International Relations, ed. Morton AT_Naplan (New York: St. Martin's Press, I968), p. 208. 88 will be to compare both SIPER and INS to the real world to see if the offspring is an improvement on its ancestor. Guetzkow's distinction between process and outcome seems to deserve further examination, as well. From the viewpoint of the international system, the processes that determine the behavior and structure of that system are the patterns of behavior exhibited by the nations that make up the system. The outputs, on the other hand, of these processes are the descriptions of the state of an interna— tional system at a point in time. In other words, national behavior processes produce outputs which in turn define the status of the international system. Our focus will be at two levels, the national system and the international system.8 First we shall want to ask the question, do simulated nations behave like real world nations? This chapter is an attempt to answer that question. Secondly we want to know if the simulated international systems are like the observed international system. This question is addressed in the next chapter. I. The Behavior of National Systems Since we are generating behavior for prototypic rather than isomorphic nations, our methods for comparing the 8For a further elaboration of the advantages and dis- advantages Of each Of these levels of analysis see J. David Singer, ”The Level—Of—Analysis Problem in International Relations," The International System: Theoretical Essays, ed. Klaus Knorr and Sidney Verba (Princeton: Princeton University Press, l96l), pp. 77—92. 89 behavior of national systems in the simulated and referent worlds must be adjusted. If we ask whether any simulated nations behave like the United States, we are not likely to progress very far in reaching a conclusion about the validity of the model. If, on the other hand, we ask whether large simulate nations act like large referent nations, like the United States, we are likely to achieve better insight. If, for example, we were to find that larger nations engage in more diplomatic conflict than smaller nations in the real world, then we would want to find a similar relation— ship in the simulate data before we would grant the model some validity. Similarly, if we find no relationship in the real world between the size of a nation and the amount of diplomatic conflict it engages in, then in order for the model to receive a passing grade we would not want to see the relationship in the simulate data. We shall, therefore, test a series of hypotheses about the relationships between attributes and behavior Of real and simulated nations. The attributes we have chosen have been isolated both theoretically9 and empiricallyl0 as variables Of special importance. These are size, development, and accountability. 9James N. Rosenau, "Pre—Theories and Theories of Foreign Policy," Approaches to Comparative and International Politics, ed. R. Barry Farrell (Evanston: Northwestern University Press, 1966), pp. 27—92. lOJack Sawyer, ”Dimensions of Nations: Size, Wealth, and Politics," American Journal Of Sociology, LXXIII (September, 1967), pp. 145-172. 90 Size is a measure of magnitude. We commonly mean the bulk or mass of an entity when we refer to its size, but with reference to national systems there is also the connotation Of resource potential. Several of the measures which have been isolated in factor analyses of real world data seem to have the common element of resource potential.11 We will use as our measure of size in the real world data national population, which has the advantage of being a rather complete and reliable statistical series.12 In the SIPER and INS data we will use total basic capability as our size measure, which seems to correspond conceptually with population more than any other simulation variable we have. Certainly in both the referent and simulate worlds we are measuring the resource potential of nations with these variables. Measuring development commonly means some assessment Of the efficiency of a national economy. When we call a nation developed we are asserting that its economic input— Output ratio is relatively lower than other nations, which are considered undeveloped. As our aggregate measure of the llRudolf J. Rummel, "Indicators of Cross—National and International Patterns," American Political Science Review, LXIII (March, 1969), pp. l27—IA7. l2Population data is taken from Bruce Russett, gt al., World Handbook of Political and Social Indicators (New Haven: Yale University Press, 196A), pp. 15—21. The choice of this indicator for the size variable is supported by Rummel, op. cit., p. 134. The indicator is log10 transformed to reduce the skewness of the distribution. 91 productivity (output/input) of real national economies we will use per capita gross national product. With all its shortcomings, this measure is still considered the best single indicator of development.13 We will use an analogous measure for the simulated nations. Let us set the variable gross simulated product equal to the sum of consumers' goods, investment goods, and security goods produced. If we divide this by total basic capability, we will have an aggregate measure of the productivity of the simulated nation's economy.lLl The third attribute variable we have chosen, account— ability, is more difficult to define than the previous two. It is unlikely that the differences between the political systems of real nations may be reduced to one dimension without losing a good deal of information about those systems. But it would be folly to try to deal with the full variety of political systems at this stage of our work. Fortunately, we are not without some guidelines in examining national 13For a discussion of the advantages and disadvantages of this indicator see Irma Adelman and Cynthia Taft Morris, Society, Politics, and Economic Development (Baltimore: John Hopkins Press, 1967), pp. 84—90. 1“It should be evident that development is a weighted average of the national generation rates. Development = a * CS generation rate + a * BC genera— tion rate + a * FC generation rate where the weights al are équal to the proportion of the total basic capability adf the nation which is allocated to the production of consumer goods (CS), investment goods (BC), and security goods (FC) respectively in a given period. 92 political systems, and the work of Gregg and Banks is 15 particularly noteworthy. The first factor which they extracted in their factor analysis of the cross—polity survey data was named access.16 It would seem that Gregg and Banks have captured the same phenomenon that Rosenau speaks of as accountability, but the nominal perspectives are different. If a citizen has access to the political system, then similarly the political leader is accountable to the citizen. We shall take our measure from another study of the cross—polity survey data in which a factor analysis Of the nations in the survey was performed.17 An inspection of the rotated factor matrix indicates that the nations that load heavily on the first factor are precisely those nations that we would describe as having open, accessible, or accountable political systems. We have therefore decided to use a nation's loading on this first factor as a measure of the accountability. The 15Phillip M. Gregg and Arthur S. Banks, "Dimensions of Political Systems: Factor Analysis of a Cross—Polity Survey," American Political Science Review, LIX (June, 1965), pp. 602—614. 16The variables which loaded positively on this factor were: electoral system, constitutional regime, group opposition, status of legislature, horizontal power distri— bution, representativeness of regime, press freedom, aggre— gation by legislature, military neutral, conventional ideological orientation, articulation by parties, articula— tiog by associational groups, and modern bureaucracy. Ibid., p. 08. 17Arthur S. Banks and Phillip M. Gregg, "Grouping Political Systems: Q—Factor Analysis of a Cross—Polity Survey," American Behavioral Scientist (November, 1965), pp. 3—6. 93 measure of accountability in the simulate worlds is ten minus the decision—latitude of the simulate nation. Obviously the higher the elite's decision—latitude, the less accountable they are to the citizenry. Hence, we want to take the difference between complete decision— latitude, ten, and the decision—latitude of the simulate nation as our measure of the accountability of its political system. Let us now consider what behavioral phenomena we wish to relate these attributes to. From the wide variety of behavior that we observe nations exhibiting we must, by necessity, select particular phenomena for study and neglect others. Our choices as to the exclusion or inclusion of a variable were guided by theoretical, empirical, and practical considerations. The variables chosen should have theoretical significance; that is, they should be essential to virtually any theoretical abstraction drawn from the relevant observable phenomena. Their theoretical import should be confirmed with empirical evidence. Lastly, the variables should be measurable. Four areas of national behavior are examined: behavior pertaining to political stability, behavior per— taining to economic growth, behavior pertaining to national security, and behavior pertaining to international coopera— tion and conflict. The political stability area contains five variables. The first three of these, turmoil, conspiracy, and internal 94 war, have been isolated theoretically and empirically by 18 several authors. We do not think we can improve on Gurr's definitions of these variables. Turmoil: relatively spontaneous, unstructured mass strife, including demonstrations, political strikes, riots, political clashes, and localized rebellions. Conspiracy: intensively organized, relatively small—scale terrorism, small—scale civil strife, including political assassinations, small—scale terrorism, small—scale guerrilla wars, coups, mutinies, and plots and purges, the last two on grounds that they are evidence of planned strife. Internal war: large—scale, organized, focused civil strife, almost always accompanied by extensive violence, including large—scale terrorism and guerrilla wars, ciygl wars, private wars, and large—scale revolts. Gurr's indicators of these variables for the referent nations are used in the analysis. INS and SIPER do not produce any direct measure Of turmoil behavior; however we can measure this behavior indirectly. Validation Satisfaction overall (VSm) is a general measure of the satisfaction of the citizenry with the political system, and if we transform this variable by 18R. J. Rummel, "Dimensions of Conflict Behavior Within and Between Nations," General Systems Yearbook, VIII (1963), pp. l—SO; Raymond Tanter, "Dimensions of Conflict Behavior Within and Between Nations, l957-l960," Journal of Conflict Resolution, X (March, l966), pp. Al—OA; Ted A. Gurr, ” Causal Model of Civil Strife: A Comparative Analysis Using New Indices," American Political Science Review, LXII (December, I968), pp. llOA—llZA. l9 Gurr, 9p. cit., p. 1107. 95 subtracting it from the level of perfect satisfaction, ten, we receive a measure of the citizenry's dissatisfaction. Although this measure is affective rather than behavioral, we will assume we are measuring potential turmoil. This approximation should be sufficient for our purposes. As to conspiracy and internal war, we find more directly analogous variables in the simulated worlds than with tur— moil. We shall call a simulated nation's probability of revolution (PR) its conspiracy score and the costs of revolution (cR) it incurs in a particular period its internal war score for that period. The fourth variable, stability, does not lend itself so easily to definition or measurement. The prevailing views in the literature seem to fall into two groups: those that see stability as the absence of destabilizing events20 and those who see stability as to some degree independent of, although not necessarily orthogonal to, instability.21 The former interpretation is followed here. 2OFeierabend and Feierabend computed political stability by counting events of aggressive behavior directed against the political system. See Ivo K. Feierabend and Rosalind L. Feierabend, "Aggressive Behavior Within Polities, 1948—1962: A Cross-National Study," Journal of Conflict Resolution, X (September, 1966), pp. 249-271. The emphasis in this point of View is on the degree to which the continued existence of the system is threatened by violence. 21Banks and Textor speak more in terms of major consti— tutional change when they discuss stability, and more emphasis is placed on whether major change has taken place in the political system, rather than the means by which this change was brought about. See Arthur S. Banks and Robert B. Textor, A Cross—Polity Survey (Cambridge, Mass.: M.I.T. Press, 1963). 96 The real world measure of stability is based on Gurr's Total Magnitude of Civil Strife index.22 Since we want to measure stability rather than instability, the index has been transformed by multiplying each nation's value by —1. In the SIPER and INS data we have a variable which evidence gathered by Charles Elder and Robert Pendley indicates is analogous to stability of the system, the probability of continued office holding (POH).23 The fifth and last variable in the political stability area is consumption. It is included here because of the theoretical and empirical relation between economic depriva- tion and political stability. Consumption is the proportion of national economic resources which are allocated to consumers' goods. As a real world measure we will use the gross private consumption expenditures as a per cent of gross national product from the World Handbook of Political and Social Indicators.24 In the simulate data sources we will use the level of consumption goods production expressed as a per cent of gross simulated product. 22See Gurr, gp. cit., pp. 1107—1109, for a discussion of the construction of this index. 23Robert E. Pendley and Charles D. Elder, "An Analysis Of Office—Holding in the Inter—Nation Simulation in Terms of Contemporary Political Theory and Data on the Stability of Regimes and Governments," (Evanston, Ill.: Simulated International Processes project, Northwestern University, November, 1966). 24Russett, gt gl., pp. l70—l7l. 97 The variables in the second behavior area, economic growth, require less elaboration. The first variable in this area, growth, is an absolute measure of the change in the referent and simulate national product. Consequently, for real nations growth equals the annual increment in 25 gross national product and for simulate nations it is the period increment in gross simulated product. The second variable in this group, growth rate, is simply the gross national or gross simulated product at time T+l divided by the gross national or gross simulated product at time T. Hence when there is no change, growth rate will equal unity.26 The third variable is investment, and here we want to examine the proportion of the national and simulate product which is located in the investment sector. For the real world we again turn to the World Handbook for the series gross domestic capital formation as a per cent of gross 27 national product. As with consumption, our ratio will be investment goods produced over gross simulated product. 25Absolute economic growth is the product of gross national product in 1957 in U.S. dollars (Ibid., pp. 152— 15A), annual growth of G.N.P. per capita (Ibid., pp. 160— lOl), and annual percentage of increase in population (Ibid., pp. A6—A8). The indicator is transformed using log (x+l) transformation to reduce the skewness of the dis ribution. 26Growth rate equals annual growth of G. N. P. per capita (Ibid. , pp. 160— 161) times annual percentage of increase in population (Ibid., pp. A6— A8). 27Ibid., pp. 168—169. 98 The third area, national security behavior, includes three variables: force capability, defense effort, and defense spending. The first of these, force capability, is an aggregate measure of the coercive power of a nation. In the real world we have taken the product of defense expenditures and size of their armed forces28 as our indicator, and in the simulate worlds we will use total force capability as our measure. Defense effort, like consumption and investment, is the proportion of the gross national or simulated product that is security goods. The real world data are taken from the World Handbook of Political and Social Indicators.29 This variable may be thought of as a measure of relative defense spending. Defense spending, the third and last in this group, in the real world is the amount a nation spends in its budget for defense, and in the simulate worlds we will use the amount of security goods produced in a period. The real world data again are taken from the World Handbook 28This data is taken from the World Handbook of Political and Social Indicators. Force capability is the product Of total population (Ibid., pp. l8—2l), military personnel as a percentage Of total population (Ibid., pp. 74—76), expenditure on defense as a percentage of G.N.P. (Ibid., pp. 79—80), and gross national product in US dollars (Ibid., pp. 152—154). The variable is log 0 (x+l) trans— formed to make the distribution less skewed. 29Ibid., pp. 79—80. 99 of Political and Social Indicators.30 The last area of interest, which we have called international cooperation and conflict behavior, contains three variables. The first, relative trade, measures the relative importance of the foreign sector to the national economy. As our real world measure we will take the ratio of total exports and imports to gross national product,31 and the ratio of total exports and imports to gross simulated product for the simulate nations. Related to this is the variable trade magnitude, which is an absolute measure of international exchange. Here we simply use the numerator in the above measure as our indicator.32 Finally, we shall look at diplomatic conflict. This variable was one of the three basic dimensions of foreign conflict behavior delineated through a factor analysis of 33 a variety of foreign conflict behavior. The variables 30Defense spending equals gross national product in US dollars (Ibid., pp. 152-15A) times expenditure on defense as a percentage of G.N.P. (Ibid., pp. 79—80). Defense spend- ing is loglO (x+l) transformed to make the distribution more normal. 31 32Trade magnitude is the product of foreign trade as a percentage of G.N.P. (Ibid., pp. 164—165) and gross national product in US dollars (Ibid., pp. 152—15“). A Ibid., pp. 164—165. log (x+l) transformation is applied to the variable to make the distribution less skewed. 33 R. J. Rummel, "Dimensions of Conflict Behavior Within and Between Nations," General Systems Yearbook, VIII (Washington: Society for General Systems Research, 1963), pp. l—50. 100 which loaded most heavily upon this dimension included the sending of threats, accusations, and protests. We shall use Rummel's factor scores as a real world measure of t.34 In the SIPER data we have a variable diplomatic conflic which is analogous to diplomatic conflict, hostile communi— cations sent. We shall take the total hostile communica— tions sent to all nations as our measure of diplomatic conflict. A content analysis of all international communi— cation in the INS runs was done, and we shall take the total number of messages sent in a period which were classed as a threat, accusation, or protest as a result of that analysis, as our INS diplomatic conflict indicator.35 Below is a summary list of the variables we will be examining. Attribute Variables Stability Variables Size Turmoil Development Conspiracy Accountability Internal War Stability Consumption 34 The factor scores are taken from the appendix of the above article. 35The analysis was carried out under the direction of Richard W. Chadwick. See Richard W. Chadwick, Definition of Simulation Threats, Accusations and Protests (Evanston, Ill.: Northwestern University Department of Political Science, April, 1965). The Inter—Nation Simulation data is from a set of runs conducted by John Raser and Wayman Crow at the Western Behavioral Sciences Institute. See their Winsafe II: An Inter—Nation Simulation Study of Deterrence Postures Embodying Capacity to Delay Response (La Jolla, Calif.: Western Behavioral Sciences Institute, July 31, 1964). 101 Growth Variables Security Variables Growth Force Capability Growth Rate Defense Effort Investment Defense Spending Cooperation and Conflict Variables Relative Trade Trade Magnitude Diplomatic Conflict It remains now for us to see how the attribute and behavior variables will be related in the referent, SIPER, and INS systems. 2. The Method of Analysis We shall rely upon correlational analysis to generate our basic measures of correspondence. However, all the correlations listed in the following sections will be second order partial correlations. For example, when we examine the relationship between size and turmoil, we want first to partial out the effects of development and account— ability. We can eliminate spurious correlations between attribute and behavior variables through spill—over effects by using partial correlation methods. In the tables that follow significance levels are given for the partial correlation coefficients when they are significant at .1 or less. We will examine the significance patterns of the coefficients to compare the direction of correlation in each data source for each attribute—behavior pair. This procedure is similar to that which Chadwick used in his 102 validation work on the INS model.36 We also use the more rigorous method of testing the hypothesis that there is no difference between the partial correlations from the three data sources for a given pair of variables. Since the test is seldom used, we will elaborate on it here.37 Both partial correlation coefficients are trans— formed in the following manner, Zr = 1/2 loge %%—;—%% The difference between the transformed coefficients has as its standard error where N1 and N2 equal the number Of observations that correlations rl and r2 are based on and M1 and M2 number of variables in the regression equation. In this are the case M1 and M2 are equal to 4, one dependent behavior variable and three attribute variables. Dividing the difference between the transformed coefficients by its standard error in this fashion 36Richard W. Chadwick, ”Developments in a Partial Theory of International Behavior: A Test and Extension of INS Theory" (Unpublished Ph.D. dissertation, Department of Political Science, Northwestern University, 1966). 37Helen M. Walker and Joseph Lev, Statistical Inference (New York: Holt, Rinehart, and Winston, 1953), pp. 255—256. Extension of the test for simple correlations to partial correlations was aided by John Stapleton, chairman of the Department Of Statistics at Michigan State University, and I wish to thank him for his assistance. produces the familiar Z statistic, which is an ordinate of the normal curve. In the following tables the differences between the partial correlations and the Z statistic of these differences are given, as is the significance level Of the Z values. A brief discussion of the nature of the samples used in this part of the evaluation is necessary. The real world data are drawn from several data sources, and the number of nations for which data are available ranges from 133 to 62, depending upon the variable. Reducing the sample down to nations about which we have complete infor— mation; i.e., a value for each of the behavior and attri— bute variables, leaves a sample of 41 nations. While this sample is not as large as one might wish, it should be sufficient for our purposes here. Table 10 lists those nations included in the sample. The sample nations tend to be a little larger, more developed, and more Open, but the bias does not appear too large. The sample drawn from the computer simulation data requires further elaboration. As was explained in the previous chapter, the simulation was used to produce twenty—four international systems. There were six different initial variable settings and four different parameter settings. The combinations thus produced were 104 TABLE 10 REAL WORLD NATIONS INCLUDED IN THE SAMPLE United States United Kingdom Canada Ireland West Germany France Belgium Netherlands Italy Switzerland Sweden Denmark Norway Finland Portugal Spain Greece Turkey Yugoslavia Soviet Union Mexico Guatemala Cuba Dominican Republic Venezuela Colombia Ecuador Brazil Peru Argentina Chile Israel Ceylon Burma Thailand Philippines Japan South Korea Australia New Zealand South Africa 105 twenty—four in number. In this part of the analysis we will ignore between-systems variance and treat the twenty— four systems as subsystems of one large international system. Later on in the next chapter_we will want to examine specifically the between—systems variance, but our focus now is on the variance between nations. Since there are five nations in each of the above systems-treated—as—subsystems, our n is 120 observations at any time point. We have chosen Period 7 as our time point for reasons to be discussed below. In the Inter—Nation Simulation data we have six international systems which we shall treat as subsystems Of a single international system. With five nations we then have an n of 30 to work with. Period 7 was chosen to give the systems time to develop but avoid ending effects. Ending effects do not, of course, occur in the computer simulation, but Period 7 was used to give the computer simulated worlds equal development time. A final comment is required concerning the elabora- tion of empirical findings that follow. For the most part, theoretical elaboration of an attribute—behavior relation— ship will be given when either 1) there is a lack of internal consistency, i.e., the three data sources are in disagreement concerning the nature of the relationship under study, or 2) there is a lack of external consistency, i.e., the relationship reported here, particularly in the real world nations, is different from that reported elsewhere. 106 3. The Attribute and Stability Variables The correlations between the attribute and stability variables are given in Table 11. If we decide that a correspondence exists between a simulate world and the real world when their respective correlation coefficients are both significantly greater than zero, or both not significantly different from zero, or both significantly less than zero, then the SIPER model produces corresponding relationships in eleven of the fifteen relationships under study and the INS model produces eight corresponding 38 relations. The correspondences and non—correspondences are distributed in the following ways. INS SIPER Correspondence Non—Correspondence Total Correspondence 7 4 11 Non—Correspondence l 3 4 Total 8 7 15 It is clear that there is a tendency for both SIPER and INS to produce correct relationships at the same time, but it is worthy Of note that when SIPER is right there is a stronger tendency for INS to be wrong than there is for SIPER to be wrong when INS is right. When SIPER is wrong in its predicted relationship, the error tends to be shared by INS. It would be premature to conclude that SIPER is 38A relationship is considered significant if it is significant at the .10 level or less. 107 .OHCH pan» mflmpgooosg Haze poo no owmo who Eon oH. oo Hasop HO QNEP mmmfi %H PQ®HOHflM®OO QQP MO HQ>®H ®0Q®0H%HQWHW H WHW .o>onw Uopwfla mopSOm dome can EH «mOHQMth> opsoflhppm hogpo can mo mpoommo who 90% msfiaaospcoo .O>opw ompwfla pappfipp> opsoflppop on» one ocpa who so ooomflfl pagpapp> om mzH omH ppaflm a: prm hoa>mgmp map Coozpon QOfipMHOppoo Hmflppmg pcpho Baooom u h ONHm OHQEmm mooo. aao. oo. :oo. mam ma. wH.- mo.- oH.| ma.u om.| om.- om. mH.- p soaposoweoo :Ho. mooo. moo. moo. mam oo.: oo.: mm. oo.- oo.- ma. mm. NH. oo. 6 soaaaomom wflm mm.u mo. om.- mo. mo. No.» mo.- mo.. ma.. 6 so: HosppoeH moo. moo. mam :m.: oo. mm.1 mm. oH.- HH.- mm.| mo.- EH.- 6 soppfiomeoo mooo. mam AH.I oo. mo.- om. mo. HH.| mo.- oo.- mo. 6 aflospse mzH pmgflm Hopm mzH ppgflm Hopm mzH sogflm prm mpfififlnmpssooo< onEQoao>mQ omflm mepmHmBB< moH> MBHQHmOH Ho. esp pm pQwOHMHCmHm ooQOHOMMHU u ** HO>OH mo. cap on nemOHEchflw ooQOHOQMNo u * I I .o>opw : mm oopmfla mltm MIFM\\\/ moopsom dump on» CH Ussom m>onm popmHH H H pHeere> poooHepoe one one oopH one so :HIH _HIH OHQwHHw> Hoa>mgon can mo wQOprHOHHoo =N+H mNOH N\H I .H+H mNOH m\H u N HeHosoo one awesome poempoHHHo who u =e-.p *oN.H- mw.ou mm.o **oo.m- :m.o- **NN.N **oH. oo.H *oH.mu N Nm.- Hm.- oH. mm.- om.- mm. mo. om. om.- :EI.N QOHpa55mcoo *emN.mu**mo.m **NN.N mm.ol oo.o oH.H Ho.H- Ho.H: Hm.o- N em.- :N. wo.H no.1 NH. mm. HN.: mm.o eo.| :E-_s NHoHHHoepm :o.H mm.o mm.H| Nm.o- mm.o- No.o- om.o wm.ou mm.H- N mm. oo. om.u oo.: mH.n oo.- mo. mH.- mm.- :pu_p He: HoepppeH HH.H mm.o| *JN.H: *oN.H| mm.Hu mo.o- **Nm.m *No.H oo.o- N am. mo.» mm.- mm.- om.- Ho.- om. mm. NH.: EH...H HoesHomeoo oH.H N:.o 2N.ou eo.ou mm.H: oo.H- **mo.m **:o.m 3N.o N om. NH. :H.- om.- om.- oH.- em. oo. :H. =EI.E HHoepoe mzH mzH epon mzH mzH NpeHm mzH mzH Epon = ngflm Hmmm Hmom moaflm Hmom Hwom HOQHm Hmom Hmom _ mpfififlnmpcsooo< pcmEQOHo>OQ ONHm maomHmeee mOH> MBHQHmoH oonommmstm u mHm .m>onw Umpwmfi OONSOm mpmp esp CH .moanm om mzH quee ppserooe Npgoo ego mo wooommp one pom omH ppon wQHHHOHono fl969w UmpmHH oHQmHHw> mpsnmappm H: Hmom map Usm pmoa can so Ucpmma OHQmHHm> Hom>mgwp mflp Emmzpmn COHD‘mecHQOo Hmflphmg chUcHO UCOomm H .H wNflm OHQEwm mNo. mooo. mHm mm. mo.| No. mo.- NH.| Hm. Hm. :3. mo.: 6 oeosome>eH mmo. moo. :Ho. mHm Hm. mo.- mo. mH. mo. om.- mH. om. mm. A poem coeopo mwo. mooo. mooo. mooo. mHm mm. mH.- mo. Hm. oo.: mm. MH. mm. Nm. p Epsopo mzH prHm Heem mzH NemHm Hopm mZH poon Hepm hpflfiflhdpssooo< psosdoao>oa mmmm meomHmBBH mOH> mBBOmU Qz< MEDmHmBB< ma mqmOH Ho. map pm psmonHcmHm mocmHoHNHo n ** Ho>OH mo. map pm HQQOHNHcmHm moswmomme n * MHEB + MHFR .o>onm : paw _ mm oopmHH mooHSQm H H \\\l mpwo who QH Uszom o>onm popmHH oHomHHm> opsoHprm osp Usd pHOH map so mHQmHHw> HOH>wgoo :HIH .HIH one mo wQOprHoHHoo HmeHmQ =H+H omOH m\H I _H+H 6m0H mxH u N one awesome poceppmmHo pep u =6u.p mm.H| HN.oI mm.o Hm.o| mo.H *mo.m mH.H Ho.H| **:N.ml N my wm.n mH.: OH. HH.| mm. mm. mm. om.| 0:.I ...HI?H psospmo>sH 1 mm.H| m:.ol mw.o w:.o| *NN.HI *mw.H| m:.o wH.H ww.o N mm.: mH.- NH. oH.- mH.] mm.- oo. Hm. mH. =p-.p poem goeoeo *mH.mI wo.HI mo.H mm.HI *Nm.H **wm.: mm.o **m©.: **om.m N m:.| mm.| om. om.| Hz. Hm. om. :N. :m. :.H|..H Epsomo mzH mzH ngHm mzH mzH HegHm mzH mzH Hpon : HOQHm Hmmm Hmom HogHm Hmom Hmom HOQHm Hmom Hwom . mpHHHQMpqsooo< psoEQOHm>OQ ONHm mBDmHmBB< mOH> mBBOmw Qz< MEDmHmBE< mom mZOHBmH coHOOHHHQmHm n me .m>opm OopmHH OoHSOm mpmv map CH nmOHanHw> OHSQHprm Hogpo map Ho muoommo 839 How mHHHHOHpqoo A959m UopmHH OHomHHm> OHSQHHHHO map Osm pHoH one so ompmHH pHepHHe> HOH>ee66 one om mzH omH Hoon H: prm Hoospon QOHHOHOHHOO HprHmQ HOOHO Usooom n H mme OHQEmm OOO. :OO. HNO. mOOO. mOOO. mOOO. mOOO. mOOO. me mcHOcOQm mm.- Nm. mm.- No.1 :N. :N. mm. mm. mN. H pmepHoo mHO. OHO. mOO. mOOO. HOO. mOOO. ONO. me pHomHm O:.| mm. 22.1 NH. 2». mm. OO. H:.| mm. H omsmmom Omo. NmO. mOOO. mOOO. mOOO. mOOO. mOOO. mHm NHpHHHdemO OH. HO. mm.| NH. :O. HO. mm. om. mm. H OoHom mzH HogHm Hepm mzH HemHm Hepm mzH Hoon Heom mpHHHnmpcsooo< p:oEQOHo>OQ ONHm mBDmHmBB< mOH> MBHmDOMm DZ< MBDmHmBB< zmszmm mZOHBOH HO. can pm unmonchHm ooQOHOHHHO H ** Hm>mH mo. can pm HQOOHHHQme ooQOHOHHHU n * MHHBI + MHFEI .O>opw : Ocm . mm UmpmHH wooHsom mumv H H \> map SH OCSQH o>onm OopwHH OHQOHHO> mpsoHHpoo one one HHpH who 90 6H66 :cHlH .cHIH IHcHHw> HOH>®S®D mflp ..HO mQOdeHmcfiHoo ...H+H mmOH N\H l ..H+,m OMOH m\.m H N Hwflphwg QQO. QOQSPOQ ®0C®.\H®.w..HHU OED H :cHI_.H **Hm.m mm.o **mo.mu **mm.m **mm.m oo.o **mN.:u Nm.H- **:N.m N weHoeoom mm. mo. mm.n Hm. Hm. oo. mm.: oH.- on. =H-.H omeoHpm **Nm.m oH.o **mm.mn **mm.m Hm.H *mm.H| **:m.m| :H.H **Nm.: N pHOHmm mm. mo. oo.- Nm. mm. Hm.: mH.: mm. NN. :H-.H pmcpmom mN.o- *mo.m- *mm.H| H:.H oo.H mm.ou mm.Hu Hm.o: mH.H N HoHHHoeeeo NH.| Hm.- Hm.- mm. mH. mo.- mm.- No.1 mH. =H:.H popom mzH mzH HomHm mzH mzH HeQHm mzH mzH Hpon = HOQHm Hmom mem HOQHm Hmom Hmom HOQHm Hmom Hmom . hpflHHprQSOOOHH PCOEQOH®>®Q ONHm MEDmHmBB< mOH> MBHmDomm Qz< MBDmHmBB< mom mZOHBOH ooQOOHHchHm u mHm .o>opm UmpmHH ooHSOm mpmw Ono CH nmOHQw om mzH IHHo> oosoHHooo Hoooo poo Ho moooooo moo poo omH Hoon mQHHHoHono a969m OoomHH OHQmHHm> ooonHopm H: Hmom map Ono HHOH onp Ho OopmHH oHQmHHm> HOH>m£oo oflp soospon HOHomHOHHoo HmeHmQ HoOHo Ocooom u H omHm OHQEmm moo. Hoo. Nmo. moo. mHo ooHHogoo mm.l OO. om.| OO.I mm. :m. mm.| OO. mH. H OHHOEOHQHQ mooo. mooo. mooo. mHm moooHcmoz om. oo. oo. No. mH.n 3N. oo. om. :N. H moooe Hoo. mooo. mHm moooe oH. oo.- Ho. mo.- om.» mH. oo.- Ho.- mo.- o o>HooHom mzH Hoon Hoom mzH HoeHm Hoom mzH HooHo Homo mpHHHnopssooo< oHoEQoHo>OQ ONHm mHomHmHH< moH> BOHHMZOOIZOHBwH HO. .Hw>wH mo. .w>0hm who om othHlowHw wothwHHHo n ** who ow ohOOHlomHm wothwHHHo n * mw OwowHH monsom mI=z + mI_z oooo who eH pejoo w>ooo owomHH H H I\> wHooHHo> ooSoHHooo who oco onH who ho thmHHw> HOH>mhwh =HIH who oo thHomeHHoo HmHono :H+H wwOH m\H I .H H meH m\H u N who hwwzowh wothwHHHO who u :HI_H *OO.H m:.O *HN.HI *NN.H *NO.H mm.O *mm.H mm.O mo. oH. mm.I mo. mo. Ho. oo. =HI_H oHoonHoHo oo.H: oo.oI mm.o NO.OI **om.m **mm.m *mH.m **mm.m *HH.m wOSlowoz om.I :m.I OO. OH.I so. ow. m:. :m. wowHB :m.HI ON.OI oo.o mm.HI mm.O *Om.m mm.O **mo.mn **:w.MI wowHB Nm.I oH.- oo. om.I mH. Ho. mo. mo.I w>Hoonm mzH mzH Honm ozH mzH Hoon mzH Honm HwQHm Hmwm Hmwm HwQHm Hmwm Hmwm HwQHm wam moHHHhmohsooo< ohwEQOHw>wQ mesmHmBE< mOH> BOHHMZOOIZOHBwH wthOHHHhmHm mHm .w>oho onsom mono who hH flmehMHHoNV woShHHoom om mzH Hwhoo who Ho moowoow who Hoo whHHHOHohoo omH Honm .w>ohw OwomHH thMHHm> woshHHoom who hon H: Howm oowH who ho owowHH thmHHw> HOH>mhwh who Ho othOHHHwoo hOHoMHwHHoo HwHoHoo HwOHo Ohoowm H wNHm wHQEmm mHO. wHO. mOO. mOOO. HOO. mOOO. ONO. me oHomom o:.I mm. ::.I NH. :N. mm. OO. H:.I om. H wwthwQ mNo. mooo. mHo mm. mO.I No. oo.: NH.I Hm. Hm. on. mo.I H ohwsoow>hH mo. mOOO. NHO. :O. :OO. mHm OH. OH.I mO.I OH.I mN.I wm.I mm.I om. mH.: H hoHoQESmhoo mzH HomHm Howm ozH HomHo Howm mzH Hoon Homo HoHHHhothooo< ohwEQOHw>wO wNHm mesmHmBB< mOH>¢mmm ZOHEwH mo. who ow oGOOHmHhMHm wHw hOHhB whOHoMHwHHoo wmhwwwml O.I ohwsomw>hH wmhwmwo OH. ohwEomw>hH HON Hod QOHoQESmhoo QOHoQESmhoo mzH Howm wmthwQ who .ohwaomw>hH .hoHoQESmhoo hwwzowm mQOHomeHHoo wHoEHm N wHSMHm 143 at variance with Russett's work.80' He found that over time there was a tendency in the United States for increases in defense to be partially paid for by decreases in invest— ment. The real world conclusions here are based on cross— sectional rather than longitudinal data, and Russett him— self notes that the U.S. pattern is by no means cross— culturally constant. He finds that Canada, for example, exhibits the same behavior that we find above; there is a tendency for increases in defense to be paid for by decreas— ing consumption. Real world nations and SIPER nations provide quite a constrast with regard to the relationship Of consumption to the other two sectors. As consumption declines real world nations seem to devote the released resources pre— dominantly to investment, while SIPER nations devote the surplus to defense. It would appear that SIPER nations give national security a higher priority than economic growth and that, to some extent, the opposite is true in the real world. The pattern of INS trade-Offs is clearly in between both SIPER and the real world. As consumption declines in INS nations, investment and defense increase, with the increases in defense being slightly better correlated to the decreases in consumption than increases in investment. The mixed nature of the INS pattern emerges clearly in a 0Bruce M. Russett, "Who Pays for Defense?", American Political Science Review, Vol. LXIII, NO. 2 (June, 1969), pp. 412—426. 8llbid., p. 416. 144 statistical test of the significance of the differences between the correlations presented in Figure 2. INS and SIPER do not differ in the degree of correla— tion Of consumption and investment, but both are significantly different from the real world at the .01 level. On the other hand, INS and the real world do not differ significantly with respect to the correlation between consumption and defense, while both differ significantly, again at the .01 level, from SIPER. Clearly, INS has factors in common with both SIPER and the real world, while the latter are mirror images of one another. A resetting of program parameters can remedy this non— correspondence. It will be recalled that a set of parameters discussed in section 6 of Chapter II, the budget crisis resolution weights PSPRI, EGPRI and NSPRI, establish the priorities Of the expenditure sectors when conflict among goals is encountered. For this set of runs, an equal weight was given each sector. This means that the same proportion will be cut in each sector. By way of example let us consider a hypothetical case where a nation wants to allocate its resources in the following way. Consumption 4/5ths of resources Investment l/5th of resources Defense l/5th Of resources This would, of course, be a crisis situation. Giving equal wieght to the sectors would reduce each sector by 145 one sixth.82 The new allocations would be Consumption 2/3rds of resources Investment l/6th of resources Defense l/6th of resrouces Proportionately consumption has been cut as much as defense or investment, but in absolute terms, it has been out four times as much as the other two sectors. Since the overall mean percent levels of allocation are 84, 5, and 11 for the consumption, investment, and defense sectors respectively for SIPER nations, it is likely that 1g tgg gtggt gt g budget crisis the type of reduction proposed above would take place. Such reductions would tend to produce the correlations found in the SIPER data. Increasing the budget crisis resolution weight, PSPRI, of the consumption sector may enable us to replicate the real world pattern without further modification of the computer model. Further study is needed, however, and such study may suggest that a reformulation is necessary. It is interesting to note, however, that SIPER nations in one sense act as Bruce Russett suggests real world nations should act. 82The desired level of expenditure is 4/5 + l/5 + 1/5 = 6/5, and if each sector is given equal weight, the new value will be 5/6ths of its former value,‘or cut by 1/6th. 146 It is bad to sacrifice future productivity and resources for current defense or war-fighting activities; insofar as possible such activities 8 "should" be financed out of current consumption. 3 Viewed from another perspective, however, the sacrificing of butter for guns may be an example of misplaced values. 8. National Attributes and Behavior: A Summary One way of analyzing the overall validity of the computer model is to follow a practice we have used in previous sections. Defining a correspondence as existing when there is mutual confirmation of a positive, negative, or zero relation by the real world and a simulate source, we find that SIPER corresponds to the real world in 27 of the 42 relationships we have considered, and INS corresponds to the real world in 20 of these relationships. The break— down is as follows: INS SIPER Correspondence Non—Correspondence Total Correspondence 14 13 27 Non—Correspondence 6 9 15 Total 20 22 42 SIPER is correct in just under two—thirds of the rela— tionships while INS is correct in just under one—half. The occasions when SIPER is correct and INS is incorrect are better than twice as frequent as when SIPER is incorrect 83Ibid., p. 416. 147 and INS is correct. Furthermore, there appears to be a slight tendency for both models to produce non-correspon- dence when SIPER does not correspond to the real world. We think this evidence supports the contention that the SIPER model, although far from perfect, is superior to the Inter—Nation Simulation. However, in order to facilitate independent overall evaluation of the simulations, we have prepared a table summarizing in a verbal and symbolic manner the relation- ships that we have found in the previous sections. Table 20 presents this information in the form of propositions that have found support in the analysis. A careful reading of this table should enable the reader to judge for himself the validity of the simulation with respect to its ability to replicate national behavior patterns. Following each paragraph is a brief symbolic restate— ment of the relationships. S stands for size, D for develop— ment, and A for accountability, and the sign below each indicates the nature of the relationship that was found with respect to the dependent variable on the left. 148 I NoMHHomhoo <<3 DO m .moHHthohsooow Ho wmewo Ho ohwEQOHw>wU Ho Hw>wH HHwho Ho wwwHOwawH .HwHHwEm me hOHhe wmoho hmho NowHHomhoo mwwH wothHwosw waHOH me hloz thHowz I HHoEHSB <<3 DO m .hoHHHhmohdooom Ho wmewU Ho ohwEQOHw>wO Ho Hw>wH HHwho Ho wmeOmewH .HwHHmEm wHw hOHhB wmoho hwho HHoEHso mme wothHwoxw meHmH me hOHhS thHowz momHHthoo <10 DO U30 .NomHHomhoo wwwH wothHwoxw hOth wmoho hwho hwoo mme Ho: hwoo wHoE flUwQOHw>wU mme Hoh OwQOHw>wO wHoE .HwHHwEm Hoh meHmH Hwhonh me momHHomhoo wHoE wothHwoxw hOth thHomz O o o < D m HHoEHSB .HHoEHSo mme wothHwoxw hOth meho hmho hwoo mme HOh hwoo wHoE nowQOHw>wHo mme Hoh OwQOHw>wO wHoE .HwHHmEm Hos waHwH Hwhonh me HHoEHSo wHoE wothHwoxw hOHhe thHomz homHHomhoo .ohwEQOHw>wO Ho Hw>wH Ho wNHm HHwho Ho mmeUHmwwH .hwoo wme me hOth wmoho hwho HomHHomhoo wwwH wothHwoxw hwoo wHoe wHo hOth thHomz O O O HHoEHSB < Q m .HHoEHSo mme wothHwoxw hOth meho hmho hwoo mme HOh hwoo wHoE nOwQOHw>wO mme Hos owQOHw>wO wHoE flHwHHmEm HOh waHwH Hwhonh me HHoEH5o wHoE wothHwoxw hOHh3 thHomz wQOHooz mzH mOH>¢mmm mQOHomz mmmHm whOHowz wam HHBwo Ho Hw>wH HHwho Ho mmeOHmwwH ahwoo mme Ohm meHwH me hOHhe wmoho hmho thmom mHHwOHoHHoo mme wHw Homo wHoE who HwHHmEm wHw hOHhs whoHowz Hm: HthwohH <<3 C10 vac .Hmz HthwohH mme wothHwoxw hOHhs meho hmho hwoo mme Ho: hwoo wHoE .uwoo IHw>wU wme Hos Owdo IHw>wU wHoE .HwHHwEm Ho: meHwH Hwhonh wHw Ho: HwhHwohH wHoE wohw IHHwon hOth whoHomz HoHHHooom .ohws IQOHw>wo Ho Hw>wH Ho wNHm HHwho Ho mmeOmewH nhwoo wwwH wHw hOth wmoho hmho thmom NHHOOHoHHoo mme me hwoo wHoe wHw hOth thHomz Hm: HwhHwohH <10 Do (00 .sz HthwohH wwwH wothHwoxw hOth meho hwho hwoo wwwH Hoh hwoo wHoE .UwQOHw>wo mme Hoh owQOHw>wo wHoE nHwHHwEm Hoh HwMHmH Hwhonh wHw Hm: HthwohH wHoE wothHwaw hOHhe thHowz NoHHHooom <-+ CIO UJO .ohwE IQOHw>wU Ho Hw>wH Ho wNHm HHwho Ho mmeOmewH nhwdo mme wHo hOth meho hdho thmom NHHOOHoHHoo wHoE wHw hwoo wHoE me hOth mQOHowz sz HthwohH wo mme Hos owQOHw>wU wHoE .HwHHmEm Hoh waHwH Hwhonh wHw sz HwhHwohH wHoE wothHwoxw hOth wQOHomz mQOHowz mZH wCOHowZ mmmHm UwzhHoQoollom MHmwo Ho Hw>wH Ho wNHm HHwho Ho mmeUHowwH «hwoo mme wHw hOHhe meho hmho mEHwo woSHomhw hH wHos SOHN hwoo wHos wHw hOHhe whoHowz I COHoQESwhoo <20 90 w .HoHHHoo Iohsooom Ho wwwao Ho ohwe IQOHw>wO Ho Hw>wH HHwho Ho mmeOHmwwH .waHwH wHw hOHhe meho hmho hoHoQESmhoo oo wHoe NHw>HowHwH woOOOHHw HwHHOEm wHo hOth thHomz hozooo <<3 Do (0+ .hoHHHhmohsooow Ho wmewO Ho ohwEQOHw>wo Ho Hw>wH HHwho Ho mmeOHmme .HwHHwEm me hOth meho hmho mEHwo wosHowhm hH wHoE 30Hm waHmH me hloz thHomz COHoQESmhoo .hwoo wHoE who .OwQOHw>w© wHoe qHwHHmEm me hOth wtho hmho QOHoQESmhoo oo wHoE NHw>HowHwH woOOOHHw hwoo mme Ohm nHowQOHw>wU mme .waHmH me hOth thHomz O + + ho30HO < Q m .moHHHhmthOoom Ho wwwao HHwho Ho wwwHOwawH aOwoo IHw>wO mme Ohm HwHHmEm me hOth wtho hdho mEHwo wosHomhm hH wHoE 30Hw UwQOHw>wU wHoE Ohm meHwH wHw hOth thHomz hOHoQESwhoo <<3 Q mc: .HoHHHhmohzooow Ho wwwaO Ho wNHm HHwho Ho mmeonwwH .Uwoo IHw>wO mme wHw hOth meho hmho QOHoQESm IhOo co mme NHw>HoOHwH woOOOHHm OwQOHw>wo wHoE wHw hOth thHomz mQOHomz mzH mQOHowz mmmHm UwSQHoQOOIIom mqmQH <10 00 01c .ohwEomw>hH oo wwwH hHw>HomeH wowOOHHm hOth meho hoho hwoo mme Hoh hwoo wHoE .UwQOHw>wU mme Hoh Owoo IHw>wU wHoE .HwHHmEm Ho: meHmH Hwhonh me ohwE Iomw>hH oo wHoE NHw>HomeH l 5 l wowooHHw hOHhe thHowz o woom hoeooo o (HO O < .mwoMH hoBOHm stoH w>wh hOHhe wwoho hwho hwoo mme Hoh homo wHoE .OwQOHw>wO mme HOh OwQOHwHwU wHos .HwHHOEm Hoh waHwH Hwhonh me mwowH hoonw Hwthh w>wh hOth thHomz ohwEomw>hH <10 D (0+ .HoHHHhmohsooow Ho wmewo HHwho Ho mme IOHmwOH nOwQOHw>w© wHoE who HwHHwEm wHo hOth meho hmho ohwEomw>hH oo wHoE HHw>Ho ImeH woOOOHHw OwQOHw>wO mme who waHmH wHw hOHhe whoHowz woem ooeooo <10 DO UJ+ .moHHthohsooow Ho wmewO Ho ohwEQOHw>wO Ho Hw>wH HHwho Ho mmeOHmme nHwHHwEm me hOth wmoho hmho mwowH hoEOHw HwhmHh w>mh waHmH me hOHh3 thHomz ohwEomw>hH <<3 DO 03C .ohwEomw>hH oo mme hHw>HomeH woOOOHHm hOHhe wmoho hmho hwoo mme HOh hwoo wHoE nHowQOHw>wHo mme Hoh owQOHw>wO wHoe nHwHHmEm Hos waHwH Hwhonh me ohwe Iomw>hH oo wHoE NHw>HowHwH womooHHo hOHhe mCOHomz woom hoseoo <10 Q UJ+ .NoHHthohsooow Ho wwwao HHwho Ho mme IOmewH flOwQOHw>wO wHoE Ohm HwHHwEm wHo hOHh3 wmoho hwho mwowH hoBOHm Hwthh w>mh OwQOHw>wo mme Ohm waHmH me hOth thHomz mQOHooz mzH mCOHowz mmmHm UOSQHoQOOIIom mqmwU Ho Hw>wH Ho wNHm HHwho Ho mmeOmewH .hwoo wHoE wHo hOth thHoms coho wmhwmwv oo wHoE HHw>HowHwH womooHHm hwoo wme wHw hOHhe thHomz o + + HoHHHooooo wooom < Q m .NoHHHhmohsooom Ho wwwaO HHwho Ho mmeUHmwwH aowQOHw>wU wwwH Ohm HwHHwEw wHw hloz meho hwho meoHHHhmowo onOH Hwowme mmwmmoo Owoo IHw>wU wHoE who waHmH me hOth whoHooz I oHOHHm wmthwQ wO mme .waHmH wHw hOHhe meho hwho wmthwO oo wHoE HHw>HomeH woOOOHHw hwoo wHoE Ohm .UwQOHw>wO wHoE .HwHHwEm wHw hOth thHooz HoHHthQmO wwHom wO mme who HwHHmEm me hOHh3 meho hmho meoHHHthmo onOH HwoowHw mmwmwoo owQOHw>wO wHoE Ohm waHmH wHw hOHhe thHooz <1: Q-+ (0+ oHOHHm wmthwQ .tho wHoE Ohm AOwQOHw>wU mme aHwHHmEm wHw hOth wmoho coho wmthwO oo wHoE HHw>HomeH womOOHHm tho mme Ohm .OwQOHw>wO wHoE .waHwH me hloe thHomz + + NoHHHooooo wooom o o m .hwoo wHoE Ohm flHowoon>wO wwwH AHwHHwEw me hOth meho hmho wwHoHHthQmo onom Hwomem mmwmmoo hwoo mme Ohm UwQOHw>wO wHoE nwaHmH wHw hOth thHomz mQOHooz ozH mQOHomz mmmHm GOSHHHCOOIION MHmHowme H Q m .wHos HHw>HowHwH waHo hOHhe wmoho hwho hwoo mme Hoh hwoo wHoE .OwQOHw>wO mme Hoh OwoonHwU wHoE AHwHHwEm Hoh meHmH Hwhonh me mme HHw>HowHwH wowHo hOth thHomz I O + whHwthm mewHwQ < Q m .ohwEQOHw>wO Ho Hw>wH HHwho Ho mmeOwawH «hwoo wHoE Ohm HwHHMEm wHw hOHhB meho hwho wwhwmwu oo mEHwo woSHomhw hH wHoE womooHHm hwoo mme Ohm waHwH wHw hOHhs thHomz womHB w>Howme .onHthohsooow Ho wwwaO Ho wNHm HHwho Ho mmeOHmwwH nHowQOHw>wHo wHoE me hOth meho hmho mme HHw>HomeH wOwHo owQOHw>wo wme me hOth thHooz + + + whHthQm wmhwowQ < Q m .tho wme Ohm .OwQOHw>w© mme nHwHHme wHw hOHhe meho hwho wmhwowo oo wEHwo wosHomhw hH wHoE woOOOHHm tho wHoE Ohm owQOHw>wo wHoE .waHmH me hOHhS thHomz I womHB w>Howme <<3 Do m .moHHHhmohsooom Ho wmewU Ho ohwEQOHw>wo Ho Hw>wH HHwho Ho wwwHOHmwwH qHwHHwEm wHw hloz meho hwho mme NHw>HomeH wOmHo meHmH me hOth thHomz I + + whHothm wmthwQ < Q m .tho wHoE Ohm .OwQOHw>wO mme nHwHHwEm wHw hloz wmoho hwho wmhwoww oo mEHwo wosHomhm hH wHoE woOOOHHm tho mme Ohm owQOHw>wo wHoE qHwMHmH wHw hOth thHomz meHooz mzH thHodz mmmHm UwSQHoQOOIIom mqmwO Ho Hw>wH HHwho Ho mmeOHmme flhwoo wHoE Ohm meHmH wHw hOth meho hwho ooHHHhoo oHowsoHoHo wHoE hH wmmwhw tho wme Ohm HwHHmem me hOth thHomz woslowmz wowHB <<3 Clo 030 .mEHwo woSHomhm hH mme womHo hOHhB meho hwho hwoo mme Hoh tho wHoE nHowQOHw>wO mme Hog OwQOHw>wO wHoE nHwHHwEw Hoh meHwH Hwhonh wHw mEHwo woSHomhm hH wHoE wvao hOHhB thHomz ooHHoeoo OHoonHoHo <30 Q+ 030 .HoHHthohsooom Ho wwwao Ho wNHm HHwho Ho mmeOmewH .UwQOHw>wO mme me hOHhS meho hwho oOHHHhoo oHowEOHQHHo wHoE hH wwmmhw owQOHw>wO wHoE wHw hOHhe thHomz wUSlommz wOmHB <30 QC) (0+ .NoHHthohoooom Ho wwwaU Ho ohwEQo IHw>wo Ho Hw>wH HHwho Ho mmeOHmwwH qHwHHwEm wHw hOHhS wmoho hwho mEHwo woSHomhm hH wHoE wOmHo waHwH me hOHh3 whOHomz oOHHoeoo eHooEOHoHo <10 Q+ 030 .hoHHHhmohsooom Ho wmewU Ho wNHw HHwho Ho mmeUHmme nUwQOHw>wo mwwH me hOth meho coho ooHHoeoo oHooEOHoHo wHoE hH wwwmhw OwQOHw>wO wHoE wHw hOth wCOHomz wUSlome wowHB <:0 Q + m-+ .NoHHHhmohsooom Ho wwwaO HHwho Ho mmeOHmwwH nHowoo IHw>wO mme Ohm HwHHwEm me hOth wmoho hmho mEHwo woSHomhm hH wHoE womHo owQOHw>w© wHoE who waHmH me hOth thHomz mQOHooz mzH mHOHoOZ mmmHm UwSQHoGOOIION MHm 15 — 0 21 —36 —33 9 —10 — ( 84) —14 2 15 18 —35 27 6 —30 ( 81) — 2 6 6 ( 59) —13 ( 59) — 2 — 1 18 13 48 10 (—68) -10 4 (-69) 0 2 4 —13 14 ( 67) —16 13 10 —64) 12 2 24 17 ( 52) (—53) 6 8 —55) 2 28 21 5 ( 53) -23 ll —32 -45 —13 ( 56) 6 11 15 (—69) — 6 (—54) 9 14 41 _ 8 3 l2 ( 70) 23 9 (—64) — ~ 8 9 —30 —10 10 {-52) 25 ( 73) ll 0 22 {—53) -43 22 —15 —32 35 —46 — 8 18 37 —19 ( 66) 22 -34 13 - 5 44 4 35 29 -41 (~75) — 5 8 2 Per Cent Total 29 16 11 11 15 10 4 3 Variance 171 dimensions that define these types account for 92 per cent of the total variance. We will select the highest loading system from each of the first six factors as our sample of systems. Hence, our sample of systems and the indentifiers to be used in the analysis will be: System Identifier 8 A 12 B 16 C 19 D 20 E 21 F The INS systems that will be used in the analysis are six twelve—period runs of the Inter—Nation Simulation model conducted by John Raser and Wayman Crow at the Western Behavioral Sciences Institute in 1962. These runs, which were part of the WINSAFE II exercise, have been discussed in Chapter III. The referent systems used here will generally be of two kinds. The world system encompasses, in theory, all Of the nations on the globe, but in practice this system must be more limited because Of data limitations. We have endeavored, however, to make the sample as large as possible. The other referent systems that will be used here are essentially subsystems of the world system. For the most part we will rely on subsystems isolated by Bruce Russett.lO When possible we will examine four systems: 10Bruce M. Russett, International Regions and the International System (Chicago: Rand—McNally, 1967). 172 the western community, Eastern Europe, Latin America, and the Afro-Asian subsystem. In certain cases we will find the need to substitute another subsystem, the Middle East, for one of these, and at other times we will examine only one Of the above subsystems. Our choices were guided by the availability of reliable data. Contemporary Systems This section of our analysis will be devoted to the consideration of the SIPER and INS international systems as they compare to referent systems which are drawn from the contemporary world. The referent systems to be studied here include the nations listed in Table 23. The system value for each variable was based on as large a sample of nations as could be assembled from existing data sources. The referent data for this part of the analysis are drawn, for the most part, from the World Handbook of Political ll and Social Indicators. Patterns of System Maintenance Activity Figure 3 indicates the system maintenance activity found in the simulate and referent data. On the average, 11Table 48, "Private Consumption as a Percentage of G.N.P.," Table 47, "Gross Domestic Capital Formation as a Percentage Of G.N.P.," Table 23, "Expenditure on Defense as a Percentage of G.N.P.,” and Table 46, "Foreign Trade (Exports and Imports) as a Percentage of G.N.P." in Bruce M. Russett, et al., World Handbook of Political and Social Indicators (New Haven: Yale University Press, 1964), were the source Of data for the consumption, investment and defense and trade variables. 173 TABLE 23 CONTEMPORARY REFERENT SYSTEMS Latin America (1a) Paraguay Haiti Argentina Brazil Honduras Columbia Costa Rica Jamaica Western Community (wc) Peru Chile Venezuala Bolivia Guatemala Uruguay Panama Dominican Republic Nicaragua Ecudor Cuba El Salvador Mexico Trinidad & Tobago Netherlands Ireland Finland Portugal West Germany Italy Turkey Afro—Asian (aa) Belgium Norway Austria Cyprus United Kingdom France United States Denmark Switzerland Sweden Canada Greece Spain Iceland Taiwan Laos South Vietnam Cambodia South Korea Burma Indonesia Pakistan Libya Morocco China Thailand Camerroon Malava Ethiopia India Afghanistan Somalia Tunisia Philippines Sudan Ceylon Rhodesia— South Africa Kenya Nyasaland Ghana Nepal Liberia Tanganyka Nigeria Eastern Europe (ee) East Germany Bulgaria Poland Soviet Union Hungary Middle East (me) Jordan Iraq Israel Iran Syria Egypt Lebanon World (w) - all of the above nations plus Yugoslavia Australia New Zealand 174 *mzmemwm Bzmmmmmm Om dd 03 dH 3 .Ewomzm homw hH thHowh othwHwH HOH mm thwB wwmx mSMBmMm mmmHm m m Q 0 m < MEH>HBO< MozH30¢ ZOHmzQH IIM. EGGS I. my 178 .Ewomzm homw hH mHOHomh othwHwH HOH mm thmB wwm* , *mzmbmwm Bzmmmmmm mzmemwm meHm mszmwm OZH we mm 03 OH 3 m Q Q 0 m < O O x m m H :52: 5;; ii; TH. I OHowm m ooSUOHm IwmhwowQ 1M. how: In. wBH>HBo< Mmzmmmg EmBmMm m mmDUHm 179 figure of about 3.9 per cent. In general, the correspondence Of the INS systems in this sphere of activity is quite good. The same cannot be said, however, with regard to the SIPER systems. Their aVerage level of system defense activity is more than twice as high as that observed in the INS or referent systems. The average defense—product ratio for the SIPER systems is about .09, which compares favorably only to the Middle East subsystems. It seems evident that SIPER systems are highly militarized in general, but there are notable exceptions. System E, with a defense—product ratio Of .04, is not substantially different from the referent world system. Nevertheless, it is clear that on the whole the correspondence of the SIPER systems to con— temporary referent systems with respect tO system defense activity is not good. Patterns Of System Integration Activity Figure 6 gives the relevant information concerning system integration patterns in the INS, SIPER, and referent systems. In terms of economic activity, it is clear that the INS systems are characterized by a very low level of interaction among the components of the system. The average trade—product ratio is .045, the lowest system's ratio is .014 and the highest system's ratio is .114. This does not compare at all favorably with any of the patterns found in the referent data. The SIPER systems are, on the average, about one and one—half times as active in system integration as the INS 180 .Ewowhm homw hH thHowh othwme Hom mm thwB wwm* *WSMBmMm Bzmmmmmm mSMBmwm mmmHm WEMEmwm mzH we mm 03 wH 3 m Q Q O Q < O m z m m H O _I.__I__I__I_ CC [CC TIH. MEH>H90< ZOHBHBO< MozhH how: MBH>H90< 20HmzH90< mwzmmmm SMEmMm BmHBO< ZOHBcH ICOflpQEzmcoo QOflpmhwoch omCoroQ QOflmchxm mocmcopcflmz .m.D mmmHm .m.D mmmHm .m.D mmmHm .m.D mmmHm om! _l :NI I.NHI I w I — _ _ _ 0 FL E 1.NH+ l.wfl+ I.:m+ om+ ZOHmmmmmma mmm>mm ma mmDUHm ”MBH>HBo< Ememwm 2H mmwzQH ICOHpQESmcoo QOprhmoch omcouoa COHmeme mocmcopcflwz .m.D mmmHm .m.D mmmHm .m.D mmmHm .m.D mmmHm m<3 mow<2 "MBH>HBO< Smemwm 2H mmwzcH oflpmm posoopm ICOfipdEszoo Coamcwdxm mBmmz mmmHm oocwcmpcflmz HBmmz mmmHm fill; _IIIJ OHI EFL fim + Ioa+ va+ vom+ szmmB wmHBO< EmBmwm 2H mmwz mo soooogm mo mogmswm coapwflpw> opmefipmm wmopmom e0 85m mo oopsom ZOHBmSDmZOO mo qm>mq mom moz mo mHm%Q mo Eovoopm Mo mohmsvm soapmHhm> opwfiHpmm moopmom mo Sam mo meadow BZMEBmM>ZH mo Hm>mq mom Moz mo mHqu¢z< hm mqm o.smH H o.smH moHsm COHmehomsH m oonHmw> mo Eocoomm go momdem COHpmHam> omeHpmm woohmmm mo 85m mo meadow mmzmmmm mo qm>mq mom Moz mo mewq mo Eovoopm no mohdsvm SOHpmem> opwEHpmm moonmom mo esm mo meadow mm¢mB mo Hm>mq mom moz mo mHmMH¢z< mm mqm go Eoooopm mo mohwsvm COprHmm> omeHpmm mooumoa mo Esm mo meadow ZOHBmzbmzoo 2H mwz mo mewq wo Eooooam go mmhwsvw QOprHsm> opwEHpmm mmogmom mo Esm mo oomsom BZMEEWM>ZH ZH mwz mo mHmMHHm H H.>Hm moHsm COHmemomsH m moswflaw> mo Eowoomm mo moamsvm QOHpMHhm> omeHpmm moopmom mo 85m ho mopsom mmzmmmm 2H mwz mo mHmMH opwEHpmm mommmom mo 85m e0 oohsom WQ mo mHmMH¢Z< mm mqm¢B 3) 4) 5) 6) 228 The rate of decrease in consumption spending for pragmatic rule systems is 2.45 per cent per period, significantly different from the decrease of 0.99 per cent per period for the null rule systems. The average level of consumption spending for pragmatic rule systems, 82.5 per cent, is signifi— cantly lower than that registered by null rule systems, 86.6 per cent. The average level of defense spending, 13.24 per cent of the gross simulated product, is signifi— cantly greater in the pragmatic rule systems than the 8.64 per cent observed in the null rule systems. Pragmatic rule systems allocate somewhat less to investment, 4.46 per cent, as opposed to 4.94 per cent for the null rule systems. The introduction of non—economic factors in the determina— tion of export prices has the following effects. 7) 8) The level of trade is significantly lower in biased pricing systems. The mean trade ratio is 3.60 per cent for biased price systems as opposed to 10.68 per cent for unbiased price systems. There is some evidence to suggest that the 3.23 per cent per period expansion of trade in the biased pricing systems is significantly different from the 1.43 per cent per period contraction of trade observed in the unbiased pricing systems. 229 No significant interaction effects were found between the two factors for any of the eight variables. In Chapter III we indicated that the international risk pricing factor (IRPF) and economic hostility buffer (EBUFF) parameters were set equal to zero to prevent a feedback loop from developing between cooperation and conflict in these runs, and the lack of significant interaction suggests that none developed. It remains for us to complete the loop in future runs and determine the effects. We find, generally, that those systems whose nations operate under the pragmatic information processing rule are significantly more militarized than those which use the null rule. The pragmatic rule entails the use of one of the four information rules, depending upon which one seems most applicable. Unfortunately we have no record of the information rules used in this series of runs, but we can be sure that Rule 4, the null rule, was not frequently used when the pragmatic option was specified. It would appear that estimates of future behavior based on past behavior, with regard to matters of national defense, lead to significantly faster arms races. In future work we will assess the influence of each of the five rules separately. Introducing non-economic factors into export prices significantly lowers the level of trade in the system, presumably due to the loss of profitable trades. It should be noted that if all nations traded up to the limit allowed by the international trade autarky factor (ITAF) for this 230 set of runs the average trade ratio would be approximately 20 per cent. In the unbiased price systems the average trade ratio is 11 per cent, indicating that approximately one—half of the permitted trade is not exchanged due to lack of opportunity. As the number of nations in the system is increased, trade opportunities should increase also, and the trade ratio is expected to rise. We have indicated previously that trade is generally too low in the simulated systems, as compared to the referent world. We are confident that the number of nations and the autarky factor parameters will produce an increase in trade if they are increased, while the trade price parameters produce a reduction in trade if they are increased. It remains a matter for future study what blend of these parameters is optimal for the simulation of a particular referent system. Let us turn briefly to the consideration of the effects which different initial variable settings have upon the model. In Chapter III we indicated the beginning values for consumption, investment, and defense for each of the six systems. Since the initial trade level was zero for all systems we will not be concerned with it here. Each system's level and rate of change in consumption, investment and defense was correlated with its initial consumption, investment, and defense levels. We find that the initial consumption level correlates quite highly, .70, with the mean level of consumption and somewhat less so, 231 .55, with the rate of change in consumption. Both of these coefficients are significant at the .01 level. The initial setting of the consumption variable accounts for approximately one—half of the variation in the levels of consumption and somewhat less than one-third of the variation in the rates of change in consumption acrossed the twenty—four systems. The initial investment variable settings account for a little less than one—half of the variation in subsequent system investment levels, with an r of .68, but the initial setting accounts for virtually none of the variance in rates of change in investment (r=.03). The variation in the initial settings of the defense variable explains 36 per cent of the variation in systems defense levels (r=.60), but very little of the variation, 10 per cent, in defense rates of change (r=.3l). Since much of the behavior generated by the computer model is based on past behavior, we are not surprised to find that the subsequent values of consumption, investment and defense are significantly related to their initial values. This only tends to underscore the need to carefully select these initial values in future computer runs. It is interesting to note, however, that the initial values account for very little of the differences in the dynamic nature of the systems. These differences appear to be due to parameter variation as indicated above. 232 Diagnostically speaking, we see three areas for future development. 1) Investigate the specific effects of each of the five information—processing rules. 2) Examine more extensively the effects of the trade parameters. 3) Initialize systems variables with information drawn from referent materials. 6. Conclusion What have we learned about the simulation model as a result of this analysis? Certainly one of the most important discoveries to emerge was that the model, in its present formulation, is best suited for generating long spans of time rather than short periods. The dynamic nature of the model suggests that if we wished to simulate a decade of referent time some significant revision of the model would be necessary. On the other hand, we recognize that in some respects the parameters of the model are too rigid to allow us to replicate accurately real world behavior over a long period of time. The generation rates, for example, are constant throughout the life of the system, while it seems more reasonable to conclude that in the real world they may change over the course of one hundred years. If we elect to revise the model in order to replicate long periods of time we shall have to reexamine the constancy of some parameters. 233 Moreover, this objective requires the addition of certain sub—models so that we may generate certain kinds of behavior which are important when a long—range perspec— tive is assumed. For example, over a period of 100 years the gradual or sudden movement from a bipolar to a multi— polar world may have great significance for the future evolution of the system. Consequently, it will be advan— tageous for us to seek formulations for military conflict and alliance changes for future inclusion in the model. In the future we should also endeavor to initiate computer runs with variable settings that correspond, as closely as possible, with known systems at some point in time. With the knowledge that we have gained here, this should be a manageable task. On the basis of these runs, we would conclude that the model produces behavior that might be considered con— servative. That is, political stability and national security seem to be the objectives that are pursued with the most vigor. The more adventurous objective of economic growth is not sought, apparently, with the same intensity. Moreover, nations in the simulated international systems appear to be rather loosely coupled economically. Many of these factors have been alluded to in the previous chapter, and it remains to be seen whether the suggested revisions will improve the model. CHAPTER VI CONCLUSION We feel that we have subjected the simulation model offered here to some rather exacting tests, given the preliminary nature of the research, and we think that some discussion of the model's strengths is warranted before we consider at some length its major weaknesses. In a research project of this complexity, the failures are often more apparent than the successes, and, perhaps on this point alone, we are justified in underscoring the latter at this time. 1. The SIPER Model in Perspective An important part of the success we feel we can claim has hitherto gone unstated; that is, the model is a Viable system. By this we mean that the model did not produce national or international systems that exploded or collapsed. In the case of an iterative model, this is no mean achievement, for complex, dynamic models have demon— strated again and again their capacity to devine paths to extinction that were completely unanticipated by their designers. Beyond this we think some credit is due the model for being able to accept a variety of input configurations with— out being overly sensitive to them. There is an implicit 234 235 trade—off between parameter—variable sensitivity and the viability of the model. The more sensitive a model is to parameter and variable settings, the greater the likelihood that the model will lack stability and tend to move toward extinction of one form or another. On the other hand, we must guard against the development of ultra—stable models, for their insensitivity to input variation renders them poor vehicles for experimentation and elaboration. We think that we have successfully avoided both of these pitfalls. Our work should be viewed in relation to other international relations simulations if our accomplishment is to be appreciated. Neither Benson's "Simple Diplomatic Game" nor Reinken's computer simulation of the balance of power system are as complex as the SIPER model, and neither has been subjected to empirical evaluation. While it was not our intention to construct a model which would rival TEMPER in complexity, we do feel that our model has been subjected to more careful scrutiny than the TEMPER model, and our work must be judged accordingly. The Inter—Nation Simulation model is the only other simulation model of international phenomena that has been compared to referent data in any systematic way, and our findings here suggest that SIPER is superior to INS. We found in Chapter IV that in about two—thirds of the relationships examined the SIPER model produced sub— stantially the same relationship between national attributes 236 and national behavior as was observed in the referent system. The results of Chapter V are not so easily sum- marized, but it appears that the SIPER—generated systems bear substantial resemblance to the nineteenth century European state system in both their static and their dynamic properties. Both of these factors suggest that we have met with some success in modelling referent world processes. We are pleased with the performance of the model in one final respect. The serious errors that we have uncovered appear to be rectifiable by means considerably less drastic than substantially rebuilding the model itself. These major errors seem traceable to a select set of factors rather than to the wider framework of the model. We think the model is sound enough to proceed with the incremental strategy of development that was discussed in Chapter I. This perhaps is the greatest contribution of this work. In keeping with this, let us proceed to a discussion of these major errors. In our analysis of both national and international simulated systems we observed a recurring set of phenomena which would suggest that the model is deficient in several reSpects. We are referring to the high level of inter- national economic autonomy, the low level of economic growth, and the pronounced tendency for military escalation to occur. We will discuss each one of these in turn, beginning with what we consider the least serious of the three problems. 2. International Autarky We observed in both Chapter IV and Chapter V that simulated nations trade infrequently and in small amounts. We have already discussed the factors that account for this deficiency, but they bear repeating here. We have noted that one of the main parametric constraints on trade, the international trade autarky factor (ITAF) was given, by contemporary standards, too low a value. However, we must bear in mind that the value for ITAF, 0.10, would permit a trade—product ratio of approximately 0.20, if all nations were to trade up to their import limits. In the systems with unbiased export pricing, i.e., the high trade systems, the trade product—ratio is about one—half of what we would find if all permitted imports were made. Deutsch and Singer have noted that as N, the number of nations in an international system, grows, the number of interaction opportunities grows by (N2—N)/2. They indicate that "every nation's needs and supplies differ, and the more nations there are, the greater will be the diversity of trade-offs available to the system."1 One of the major reasons for the relatively low level of trade in all the systems is the lack of interaction opportunities. Increasing 1Karl W. Deutsch and J. David Singer, ”Multipolar Power Systems and International Stability," World Politics, XVI, No. 3 (April, 1964), p. 395. .238 the number of nations from five to ten will raise the interaction opportunities from ten to forty-five. We think this revision can be counted upon to increase the i amount of trade, but we do not know by how much. We know as a result of our studies in Chpater V, that non-economic factors, when they are allowed to influence trade prices, can have a tremendous depressive effect on international trade. We have yet to explore in a systematic way the effects of the trade—price—bias parameters on the direction and level of trade. There are, however, encouraging developments in the field of inter- national economics which indicate that the field is becom— ing increasingly aware of the interaction between inter— national economic and political phenomena.2 We foresee that we will have the benefit of this work in the future. One more point deserves further elaboration. The model as it stands is non—spatial. There is no geographic reality, and as we proceed to adapt the model for simulating particular systems or subsystems, the need to introduce the factors of national proximity and continuity becomes more pressing. Spatial proximity and continuity are important factors in both international cooperation and international 3 conflict, and we shall have to think seriously about adding 2A very recent example is Charles P. Kindleberger, Power and Money (New York: Basic Books, 1970). 3See, for example, James P. Wesley, "Frequency of Wars and Geographical Opportunity," Journal of Conflict Resolu— tion, VI, No. 4 (December, 1962), pp. 387—389. 239 a spatial dimension to the simulated international systems. Ragnar Nurske has pointed out that international trade has served as "the engine for growth" in the inter— national system.Ll It has encouraged national specializa— tion and disseminated technology throughout the system. The low level of international exchange we observed in the simulated international systems explains in part the next general problem we will discuss, i.e., economic stagnation. 3. Economic Stagnation Our observation concerning the dynamic nature of the model and the systems it produces indicates that the model‘s performance with respect to economic growth is less than satisfactory. Part of this can be explained by the low levels of investment which we found characteristic of the simulated nations. However, the rate of economic growth, given a certain level of investment, is also dependent upon the depreciation rate of national productive resources. Currently, the depreciation rate is either 2, 5, or 10 per cent, depending upon a stochastic determination where each value has an equal probability of being selected. This procedure seems questionable from several points of view. It seems likely that depreciation rates, rather than being random, are systematically related to the level ”Ragnar Nurske, "Patterns of Trade and Development," Economics of Trade and Development, ed. James D. Theberge (New York: Wiley, 1968), pp. 85—102. 240 of economic development. We suspect that the more developed a country is, the higher the depreciation rate. If this suspicion is borne out in future empirical work, we anticipate reformulating the depreciation process so that depreciation rates are related to generation rates. The generation rates themselves, however, pose serious questions. They have been held constant throughout these runs, and the model contains no process by which they can be changed. Here we face fundamental conceptual diffi— culties. There are certain conceptual ambiguities in the economic system which have serious long term consequences. In the present formulation gross simulated product is equal to the total consumption, investment, and defense goods produced by a national system. GSP = CS + BC + EC *BCGR*TBC + a *FCGR*TBC = x * al CSGR TBC + a2 3 There are only two ways in which national product can expand in this formulation if we hold the generation rates constant. The set of ak values may be altered so that TBC is transferred from sectors with lower generation rates to sectors with higher generation rates. In the referent world this kind of process is encouraged by the forces of comparative advantage, and growth by specialization is common. However, there are limits to how much growth may be generated in this way, and when trading activity is low, 241 as it is in the simulated systems, the amount of growth that can be generated in this way is quite small. The other possible way to expand national product is to increase the national resource capability (TBC). In the present formulation, the production of BC value increases the national resource capability. TBC(I,T) = TBC(I,T—l) + a *BCGR*TBC(I,T-l) — DPTBC*TBC(I,T—1) 2 Total basic capability is the total of human and non—human productive resources available to a nation. While we can readily see how investment through the production of BC value, can increase the non—human productive factors, the conceptualization falters when we examine the connection between investment and the expansion of human resources. In other words, investment contributes directly to the quantitative and qualitative expansion of technology, but its relationship to population growth seems, at best, an indirect one. Moreover, if we were to assume that TBC represents only non—human production factors, we would be forced to argue that the generation rates, which are constant, represent the quantity and quality of human resources. This would clearly be an argument that we are not prepared to make. If we reverse the conceptualization and suggest that TBC represents human resources and the generation rates represent the quantitative and qualitative aspects of non— 242 human resources, we have a formulation that is more satisfying for a variety of reasons. A very large propor— tion of short term economic_growth is due to population growth, and, consequently, if we are to hold anything constant it seems more reasonable to freeze the level of technology, as embodied in the generation rates, rather than the size of the population. Furthermore, the assump— tion that TBC represents population and the generation rates represent the level of development allows us to take advantage of much work that has been done by economists and estimate values for these variants from real world data. We still have the original problem of how investment causes population growth; that is, how BC production increases TBC. We propose to reformulate this relationship so that BC production contributes to the growth of genera— tion rates, and CS production contributes to the growth of TBC, or population. While this reformulation is still in an early stage of development, we believe that it will enable us to intro— duce some interesting dynamic aspects. For example, we will be able to simulate nations with high population growth rates and low developmental growth rates, like contemporary India, or low population growth rates and high developmental growth rates, like late—nineteenth century France. The separation of human from non-human productive resources will enable us to introduce diffusion of technology and population migration dynamics as well. We foresee, ultimately, 243 that we will be able to generate a greater variety of economic growth patterns. However, it is likely that this reformulation, given the same level of investment that we observed in these systems, would not produce the kind of growth dynamics desired. This, for the most part, is due to the tendency for simulated nations to devote larger and larger proportions of their national resources to defense. 4. Military Escalation An arms race has been defined as the situation where "rival states stimulate one another to divert increasing proportions of their national income to military prepara— 5 ll tions. A review of Table 25 in Chpater V indicates that all of the simulated systems are characterized by arms races. The slowest of these races is found in system 21 where the average proportion of national income devoted to defense grew by 7.1 per cent per period. The fastest race, 31.4 per cent per period, is found in system 5. We found in Chapter V that much of the variance in the speed of arms races could be explained by the informa— tion processing rule used, as we noted the need to study systematically the effects of these rules. Beyond this, however, we must recognize the fundamental tendency for the model to generate arms races. We suspect that these rules 5Deutsch and Singer, op. cit., p. 391. 244 may increase or decrease the speed of arms races, but they will not effect the occurrence of arms races. To find the causal agent responsible for the arms races we must look elsewhere. The first factor we must consider is the rather obvious one that little provision has been made in the present model for major powers to lower defense expenditures. The major policy alternative that is evaluated is whether to keep defense spending where it is or to increase it, not to decrease it. This leaves us with only equilibrium or escalation as possible outcomes. Why don't we find equilibrium in any of the systems? The argument for the emergence of equilibrium seems particularly sound when we recall that the major powers are motivated by what may be called "parity deterrence.” That is, the power that is behind in the arms race does not seek superiority over the leader, but rather, it desires only to match the leader's power. When equality between the two powers is attained, they are both satisfied and feel no impetus to increase the proportion of resources devoted to defense. Once the parity point is reached the two powers will be in a stable equilibrium. We do not observe such equilibria principally because of the depreciation of total force capability (TF0). The stochastic choice of depreciation rates introduces shocks into the interaction system of the two competing powers. 245 These distrubances push the system off its equilibrium point in an escalating direction in the following way. Suppose powers A and B are in an equilibrium posi— tion; that is, both have the same level of total force capability, and neither wishes to increase the proportion of resources allocated to defense. TFC depreciates each period, and the depreciation rates are quite high. The rates 0.2, 0.3, and 0.4 have an equal probability of being selected. There is a one—third probability that, in a given period, powers A and B will suffer the same rate of depreciation. If this happens, the equilibrium will not be disturbed since a gap between the powers will not appear. If, on the other hand, they experience different deprecia— tion rates, which may be expected to happen in two out of every three periods of time, a gap may develop between the powers such that one of them is encouraged to increase the proportion of resources devoted to defense. Since this proportion will not drop, it seems highly unlikely that the powers could avoid military escalation for any length of time. It does not seem reasonable to maintain the deprecia— tion mechanism in its present form. Referent decision— makers have sufficient information about the peace—time depreciation of their force capability, and they can adjust their defense allocation accordingly. Some unpredic— table depreciation remains, but the stochastic variability 0f depreciation rates should be reduced and, perhaps, —‘ 246 normally distributed. It is clear that we shall have to reformulate the depreciation processes throughout the model. The ability to anticipate and compensate for force depreciation will permit the attainment of equilibrium, but it will not make possible military deescalation. To do this we shall need to reformulate the aspiration—level— for—national—security mechanism such that provision is made for the lowering of defense expenditures. At this point we can only suggest some factors that we wish to include in the reformulation. The present simulated nations react only to the existence of a gap. Modifications should be made such that rates of gap closure are monitored also. Thresholds should be introduced where actions that increase, decrease, or maintain arms are activated. Furthermore, nations must be aware of approach- ing their maximum feasible allocation of resources to defense. All of these factors require some substantial theoretical and empirical work before they can be incor- porated into the model. We believe, however, that the reformulation will improve the military dynamics of the model. Often, in the past, nations have reacted to a military challenge by doing more than increasing arms. One frequent response has been to seek and secure allies. This strategy is not available to the simulated nations, and its absence points up an important deficiency in the model. Alliance 247 structures are not changed internally, but must be intro— duced as input. If the model is to have any long range predictive power, we must be prepared to produce these changes internally. Therefore, we resolve to add the necessary processes to produce dynamic alliance structures in the near future. However, we must bear in mind that approximately one—half of recent arms races have ended in war.6 The model includes no procedures by which the simulated nations can bring their arms races to an end in this manner. Consequently, until the addition of military conflict to the model, we are likely to observe in some cases the generation of arms races which do not end short of the collapse of the system. Finally, it is clear to us that, for reasons of assessing validity, we must set up simulated worlds as much like a particular referent time and place as we can. This approach will demand an extensive data base, and the compilation of such a data base is as important as reformulat— ing the model. 6Samuel P. Huntington, "Arms Races: Prerequisites and Results," Approaches to Measurement in International Relations, ed. John E. Mueller INew York, Appleton-Century— Crofts, 1969), pp. 15—33. l APPENDIX THE COMPUTER PROGRAM The computer program that follows is written in FORTRAN IV, and is designed to operate on the Control Data Corporation 6400—6500—6600 computing systems. The use of other computing systems might require slight alterations. The program is composed of eight subroutines and a main executive routine. Figure 19 indicates the sequence in which the subroutines are executed in one period of simulated time. Two solid lines represent iterative passage of control from one routine to another while a single solid line indicates that a routine is called only once in each cycle. A period of simulated time begins at about one o'clock on the execution clock in Figure 19. The first routine called, DMERl, is concerned with revising aspiration levels and setting export prices. TRADER, the next routine called by the executive routine, negotiates and concludes trades between the simulated nations. DMER2 revises the national product decisions in accordance with trade commitments, resolves budget crises should they arise, and assesses the need for foreign aid. The granting of aid and the expression of hostility is handled by DMER3. CALCER calculates the consequences of decisions made and prepares the system for 248 249 FIGURE 19 PROGRAM EXECUTION SEQUENCE 12 ll\ /1 OUTPUT 10 2 \ / CALCER DMERl 9 ——. FORCST SIPER FORCST -——3 DMER3 TRADER 8/ \u DMER2 CORREL 250 the next period. OUTPUT, as the name implies, transfers summary information about the period to the appropriate output device. The program has several input/output options. The user may specify at what intervals data is to be read in and the model set back on some prescribed track. For example, it is possible to specify that after each period of decision—making, data is read from the referent input tape, and this information is substituted for the model—generated data before the next period begins. In this way single— period differences between referent and simulated behavior may be studied. On the output side, there is provision for specifying detailed print—out of all decision—making or specific por— tions of the decision—making cycle. In addition, summary information may be printed and/or written on tape at the end of each period. Our estimates of space, time and cost are as follows. The program requires approximately 50,000 octal memory locations, and one period of decision—making for a five- nation world requires approximately one—half of a second of central processer time. Our cost estimate for a Control Data Corporation 6500 computing system is less than one cent per nation per period. This estimate does not include compilation costs or peripheral processing costs, which may vary. We would estimate the basic cost of generating the twenty—four systems we have analyzed to be under twenty dollars. 251 I—INMJ‘LOKONCDG <11.szmo.mmo.HHao.m mamo.emom.m.mom>.mpoom.ommHH.ome<.momH<.mH<.mm.HHH<.ome.>mm.m<3.onm.Hm.om.om.mo.H .mom.mom.mmo.om<.ome.zHHzH.emomzH.emomxm.ommxm.mo.zpe .AmvzHZmo .AmvmmHQ .Amvmmc .Amvw .w AOHVmom> .AomvaE .Amvmqmmpm .Amvs .AmvmpmmmH .Amv>Hom .Aom.mvem .s Amvmzmo .ANH.o.ovBmom .Am.m.mvmpomH< .Am.mvmpoom .ANH.mvommH< .ANHS .mvomcHH .ANH.mvmomH< .ANH.m.meH< .ANH.m.mvmm .AoH.OHVHHH< .ANH.OHVome .ANH.oHv>mm .AOHVmHZ .AOHvon no .AmH.OHVHQ .AOHVQE .AOHVom .AOHVmo .ANH.OHVEOHE .ANH.OHVmom .Amm H.0vaom .AmH.OHmeo .AOHvomH .ANH.OHVome .Am.oHvzHH2H .Am.OH.OHVBH momzH .Am.m.mvemomxm .Am.m.mmvommxm .Am.Ovam .Aomvzpeam ZOHmzmsz Am HEHB .H mmae .epmepo .emmsz mmmHm EHmoomm 252 Em .AE.HVEE.AE.HVHE.AE.HVEOE.AE.Hvzm>.AHVm2m>.Avaom>.AHv>HEE.AHVH HHEEE.AHVszmo.AE.HVoEHQE.AE.HVomEEE.AE.HvoEE.AE.HvomE Amm.mv EHEE z.HuH m om m.HuE m OE m+NNHuzsmH ANH.HuH.AHVEV .wm E .EEEHE*.HEHV EHEEOE Hm O O EEzHEZOO ON EOzHEZOO EH EOzHEZOO OH E.AEOEE EEEEH*.EH\\\H.OHE.EHH.H.OHE.EOH.H.OHE.Em\\*HzEHOHEEEOO*m .EmH.*HzEHOHEEEOO *.EmH.*HzEHOHEEEOO*.Em\*EOzEE.AH+E.HOEE.HE H+E.HOEE.HH+E.HVEOE.AH+E.HO2E>.HHOmzm>.HHOEOE>.HHO>HEE.AHOH.szmO.EEO.HHEE.E EEEO.HEOE.E.EOE>.EEOOE.OEEH<.OEOH<.EOEH<.EH<.EEEEH.E.EHHE.OEE.>EE.E<3.OHEE.HE.OE.OE.EO.H EOHE.EOE.EOE.EEO.OEE.OmE.EHHEH.EEOEEH.EEOExE.OEEHE.EO.2OE .HEOEHEEO .AEOEEHO .HEOEEO .AEVHEEEE .HEOOE EOE .AEVOEOm .HEOEOOE .AEOOEOE .HmVOEOm .AEOOEEO .AE.EOEEOEH .AE.EE VEEOEE .AE.EOOmEH .HE.EOOEHE .HE.EVEEOEEE .AE.EVEEOEHH .HOHOEzm> .E HOHOEOE> .AOEOEHE .HmOEHEEOm .HEOE .HEOEOEEEH .AEV>HEE .HOE.EOHE .E HmVEsz .ANH.O.EVHEOE .HE.E.EOEEOEH< .HE.EOEEOOE .HEH.EVOEEH< .AEHO .vaEOHE .AEH.EOEOEH< .HEH.m.mOEH< .HEH.E.EOEE .HOH.OHVEHH< .AEH.OHOOEE .AEH.OHV>EE .AOHOEHE HOHvoHE EE .ANH.OHOHE .HOHVOE .HOHOOm .HOHOEO .HEH.OHVEOHE .HEH.OHOEOE .Amm H.OHOEOm .AEH.OHOEEO .HOHOOEE .ANH.OHOOEE .HE.OHOEHHEH.HE.OH.OHOEH EOEEH.HE.E.EVEEOEEE .Hm.m.mmvoEEHE .HE.OHVEO .HOEOEEEEE EOHEZEEHE HzHEEzH.EzHEEH.Oz.E.z.HO HEEEE EzHepoEEEE Ezm O EEOHEE HnmEH AEOEEEEEHEEE EH ZEOHEE HOEEanmEH HHEEEzHOHHEEnHEE EH 258 ODOC.)OOOUOOOOOOOOOOOOOOOOUOOOOOOOOOO EOzHHzOO E HH.HOEHH<*HHIE.HOEOEHHH.vapooE z.HnH E OE E E OE OO EEzHHzOO E AH.HOHHH<*HEE+HH-E.HonEOuHH.HVEEOOE H<.E<.H.H .EE HzHEE HH.OE.H2HEEHO EH HE.HOOEE\HHE.HVEO\EEH ZOHEEEHEE<*.EOE.OEHO HEEEOE Em HH.OHE.* EH3 EEHEEHHEE EOEH EEHEE EEH .*.EHE\.HH.* EH OE EOH<> EEEEHHEE OH EEEO EH: *.HH.* EHOE zOHHEEEOEzH*.EHEO HEEEOE Em O O EEEHEE OuHHVOHEE EEzHHzOO Em EOzHHzOO Em EOEHHzOO Em HH.E..EHH<.OEH.>EE.E<3.OHEE.HE.OE.OE.EO.H EOHE.EOE.EOE.EEO.OEE.OEH.EHHEH.HEOEEH.HEOEEE.OEEEE.EE.EEH .HOH.OHOHHH< .HOHOOEE .HNH.OHO>EE .HOHOEEE .HOHOOHE EE .ANH.OHVHE .AOHvoE .AOHvOm .AOHVEO .ANH.OHOEOHE .ANH.OHOEOE.HNN H.OHOEOE .ANH.OHOEEO .HOHOOmE .HNH.OHOOEH .AE.OHVEHHEH .HE.OH.OHOHH EOEEH .HE.E.EVHEOEHE .HE.E.ENvoEEHE .AE.OHVEE .HONVEEHEE ZOHEEEEHE HHzHEEH.Oz.zO EEEHEH EzHHEOEEEE Ezm HE.OHE.*uH NE *.E.OHE.*n:< *.E.OHE.*nEHHOE *.HH.* EOHE<2*.EOE\EHV HMEEOE m: E.OHE H *nH<* .E.OHE.*uE< *.HH.* HHHE *.HH.* EEEEEH*.HOE\.EHO HEEEOE EE AE.OHE.*HOEEH< 3E2 .*.HH.* EOHHHEoEE EEE *.HH.* EOOE EO EHHzO *.O.EE.HH.* EEzEE *.HH.EEO HEEEOE EH HHH.* EOH<> EOE *.H HH.* EOHHEE EHHE *.HH.* EEH<> EEEzEEOxE *.HH.* EOHHEz*.OEHO HEEEOE EH HNH.* EEEOE EEEEH*.EON.OEHO HEEEOE EH O O EEOHEE EEzHHzOO NH EEzHHzOO HH NH OH OE AAHIOz*zv.OE.EzO EH EEzHHzOO OH H+Ezu.EEE.HEEEE.OEOE.OEOE.EOOE.OEOE.OEOE.OEEE O.EEEEE.EEEH.EEEE.EEEEH.HEOEEH.HEOEEH.3.EHEEEE.EOOEHE.EOEEEH.>HEE.E EEEO.HEOE.E.EOE>.EEOoE.OEEHE.OEEH<.EOEHE.EH<.EEEEH.EEEH.EEEN EHH.EEHEE.OEHEE.EEOE.EE.EOE.EE>.EHHE.EEH.>EE.E<3.OHEE.EE.OE.EE.EO.H EOHE.EOE.EOm.EEO.EEE.OEE.EHHEH.EEOEEH.HEOEEE.OEEHE.EE.2OH .E HOHVEOE> .HONOEEE .HEEEHEEEE .HEVE .HEOEOEEEH .HEO>HEE .HOE.EOHE .E HEVEEEO .HNH.E.EVHEOE .HE.E.EVEEOEH< .AE.EOEOOOE .HNH.EVOEEHE .HNHE .EVOEEHE .HNH.EOEOEHE .HNH.E.EOEH< .ANH.E.EEEE .HOH.OHOHHHE .ANH.OHvoEH .ANH.OHO>EE .AOHOEEE .HOHOOHE EE .HNH.OHVHE .HOHOOE .HOHOOE .HOHVEO .ANH.OHOEOHE .HNH.OHOEoE .HNN 267 mmmmmmmmmmmmmmmmmmmmmmmmmmLammmmmmmm HOz.HuH.AHVEO .EN HzHEE AH.OE.EzHEEHO EH EOEHHZOO E HHEEOEEIO.HV*EHOH\HAEVEEAHVEVHHHVE E OH OE HO.OE.HHO3O EH E.HnH E OE HOz.HnH.HHO3O .EE HzHEE HH.OE.EzHEEHO EH EEzHHzOO E O.HuHEEOHE H.H.HE.HEEOxEO EH AHVEHHEEOEE HO.OE.HHO3 EH AHVEEHHO3+EHOHHEHOH Oz.HuH E OE OuHEEOHE OuEHOH E AOz.HuH.AHO MOH.OHOHMHE .ANH.OHOOEH .HNH.OHO>EE .HOHOEEE .HOHvoHE Em ANH OHOHE. HOHOOE HOHOOE .HOHVEO .HNH.OHVEOHE .ANH.OHVEOE .ANN H OHVEOm .MNH OHOEEO .AOHOOEE .ANH.OHVOEE .HE.OHVEHHEH .HE.OH.OHVHH EOEEH .AE E EOHEOEEE .HE.E.ENvoEEEE .AE.OHVEE .HONOEEHEE EOHEzEzHE HHzHEEH.Oz.z.H.HV EEEEE EzHHEOEEEE Ezm HEOE*.EE.*OE*.HE.*EO*.EEN\*zOHHEEHE>E*.EEN\\.OEHO EEEEOE EE HHHE.E.EEOE.EE.*EEOEEE*.HHO HEszE WE H*OE* ENH H *Om*.ENH.*EO*.EE.*OEH O\E *.HE\NH.* zH EHEHEO HEEEOE*.OEHV HEEEOE EE AHEE.E.EEOE.EE.*EEEEHEE*.EHV HEEEOE EE HH.OHE.EE.NH.*zOHHEz EOE EOHEEEE*.OEHO HEEEOE EE . . . HEOE *.H.OHE.*OE *.H.OHE.*EO H * H.OHE EE\\HH * EOHHEz OH EEHOEEHE HEEOEEE EHE*.EOH\\.OEHV HEEEOE EE MH.OHEmM* HHOHEEE*MEEO HEEEOE NE H.OHE * HE>EH EENHHEEE* EEO HEEEOE HE AH.OHEE.* HE>EH EEEHEEE*.EEO HEEEOE OE AHHE.E.EEVE.HE.*EEEEE*.EHO HEEEOE EN AHE.OHE.H.OHEON.EE.*HOEEOEE HEz*.HH\*EszHEEEE*.HHO HEEEOE EN AAE.OHE.H.OHEON.EE.*EEEOEEH*.EEO HEEEOE EN AA AHEMOHE.HWOWEVW.EE.*EHEOEEEEMEEV HEEEOE EN E.OHE H OHE N EE *EHEOEEE Hmz* EHO HEEEOE EN AHE.OHE.H.OHEON.EN.*HOOEOEE HEHOH*.EHV HEEEOE EN H*OEH O\E*.HE.*Omz*.EE.*EEE O\E*.HH E.*OE*.ENN\A* zH EEH<>*VE.EHN\\* EEEEEEE>OE *EEH.HON.OEHO HEEEOE EN . . HEOEH O\E*.EE.*OEZ*.EE.*EEE O\E*.EH E *OE* ENN\H* zH EEE<>*OE.EHN\\* HEEEHEEEEH *EEH.HON.OEHV HEEEOE NN HEOEH O\E*.EE.*OEZ*.EE.*EEE O\E*.HH E.*EO*.ENN\H* 2H EEH<>EOE.HHN\\* EOHHEEOEEOO *EEH.EON.OEHO HEEEOE HN ANH.*zOHHEz EOE EEOHEHOEE EHEHEEEEEHEHE.HOH\\\\.HEHV HEEEOE ON O O EEEHEE 271 mmmmmmmmmmmmmmmmmmmmmmmmhmmmmmmmmmmm HE.H.HOEEEEH+E.EEE.EEEEE.OEOm.OEOE.EOOE.OEOE.OEOE.OEEE O.EEEEE.EEEH.OEEE.EEEEH.HEOEEH.EEOEEH.3.EHEEEE.EOOEHE.EEEEEH.>HEE.E EEEO.HEOE.EEE.EEE>.EOE>.EEOOE.OEEHE.OEEHE.EOEH<.EHE.EE.HHHE.OEH.>EE.E<3.OHEE.HE.OE.OE.EO.H EOHE.EOE.EOE.EEO.OEE.EEH.EHHEH.HEOEEH.HEOEEE.OEEHE.EE.EEH .E AOHVEOEE .AONOEEE .AEOEHEEEE .HEOE .HEOEEEEEH .HEOEHEE .AOE.EOHE .E 272 mmmmmmmmmmmmEEEEEEEEEEEEEEEEEEEEEEEE EEHHHE+.NO\HOHNV\HHE+.NO\HAHOEH+HHEOEE HHH: *.NH.* EEEE *.NH.*OOOO zH OHE*AHHV EEEEOE 3m HH.OHEn* E>HEOmE EHHE *«NHQ*EHmEEOO*.OEHO EEEEOE Hm HH.OHEn*n mOH HEEEm *AHH\*OE *«H.OHE.Hm.NH.* EOE EEEOOEE OHE EEEOE*.OEHV EEwEOE om NH *EH OEE OE*.H.OHE.* EO EOEOOEE OHE OE>HEOmE *.NH«*zOHEHO OHE OzlquEEpm Oz *.OEHV EEEEOE EN 277 waswwwmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm H+OHuNH AAHH.H+OHV.EH.x.EO.E.Eq.x H.OOOOEm\HHOOnm EsznEmOEnEEEnmzw>nmom>.EEOOEAOEEHEAOEOEE.EOEE<.OHAEEHEQOEEQ>EEQE nHmvszmO nAEVOEEO “AEVEEO .AEVEEEOE «AEVOE EOm «HEVOmOm .HmOmOOE «AEOOEoE nGOOEOm nGVOEmO “AO.mvEmOzH «AOQEE vEmOEE .AEAEVOEEH “HOnEOOEEE aAmnmvEEOEzE nHmnmvEEOEEE .AOHVOZEE “E HOvaOEE nHONVEEE «AEVOEEEEE .Hmvz nAEOEpmmEH «AEV>EOE «AOEQEVEE “E HmszmO .ANHnO.EVEmOm .Hm.m.mvEOOaH< nHmanEOooE AHNHfivamEEE .HNHO «EOOEOHE .ANH.EVEOEEE .HNHQE.EOOH< aHNHKQEVEOEEE .AEVOEEE “Am.mvmm ZEEEH nANHAOHOOEEOE HNH.OHVOEEEE .AOHVEEOE nANH.OHVEE .ANH«OHVEOE: aANH.OH:Em> .AOH.OHVEEEE nHNHQOHvoEE “ANH.OHO>EE nSHOE: n83qu OO nHNH.OHEO .AOHvoE nHOHVOE .AOHVOO nHNHnodEOHE .HNHQOHVEoE nANN H.OHOEOm nANH.OHOEmO nAOHOOmE nANH.OHOOmE nHEQOHVEHEEH nHmnOHnOHOEH EOEEH aAQEQEVEEOEHE aAmqmuvaoEEEE nAm.OHvEO aHONVESEEO ZOHmZEEHO AszmanEnzv EEOEEO EzHEOOEEEm EZE ZEOEEE O.HuEEOO N ZEOEEE HHN**EEEOOO.Oz<.H.H.OE.HHquOOVV EH .HuHHOEqOO AAO.OE.AHO>EOEV.Oz<.A.H.OE.HHVOHOEVO EH OuAHVEqOO .HuAquOO A.H.EO.AHVEEOV EH AHO>EOO+HHVOqOOuHHVqOO EOzHEZOO : EOzHEZOO m leHuHHOmOm> H HHAEH.NHO.EE.AE.HOEmOV.OZE.AHHIEHQNHVEOmom>.EO.HE.HOEmOVV EH NH.mqu m OO ouAvaom> HANQNHOEOmOm>.EH.AEnHvEmOO EH EEzHEZOO N 279 I::IIICC'IE513131111IEEEEEEEEEEEEEEEEEEEEEEEEEEEE Ezm ZEEEEE EEzHEZOO HH EEEEO*E. "HHVEHEEO A. OOONEH. EE. EEEEOEEH EEEEO*E. "AHVzHEEO A. OOOEHH. EE. EEEEOVEH EEEEO*E.HAH VzHEEO A. OOOEE EE. EEEEOVEH EEEEO*E. "AHvzHEEO A. OOOEE. EE. EEEEOVEH EEEEO*E. “HHVzHEEO AH.HVEE*AH+EnHVOEEuEEE*AE.HVEOHEIAH+E“HVOEEHHH+E.HVOEE OuAHVEEEEE EEzHEZOO E .HIHE“HOEEE\AH+EaHVOEEnHH+EnHVOmEEE E OE OE AO.EE.AE.HVOEEV EH OuHH+EnHvOEEEE AO.EE.AE.HvOmEE EH HOIEHHH+EQHVOEE HO.EE.AH+EAHVOmEV EH AH+E.HVOEE*AH+EAHV>EE*N.IAH+EQHEOEEuAH+EnHVOmE .HuAH+E«HV>EE AHH+EaHv EE.EE.AEVEZEEV EH OuHH+E.HV>EE OuHH+EnHVEE H.E.EE.AH+EAHVEE>E EH .N\HEEE+AH+E.HVEE*H.VnHH+EaHVEE AEnHEEOHEI.HuEEE A.HIAH+EAHVEEE*H.+HH+E.HEEE>*AHH+EQHVHEI.HHO*HO.uAHMEHHvEOE H E HVH >EE+AE HVEEE*. N+AHVE<3*. N+AHVEHEEI HHVE2E>*E +AHVEOE>*E. ImH+E HVEE> 2 HI H HH OE EEzHEzoO E .OHuHHszm> H.OH.EE.HHEEZE>E EH AHEEOEEnAHVEzE> AEE..EE.A3OEEO\3OEEEE.AH+E.HOEEnAH+E.HVEEN aHH+E.HEEOE.AH+E.HVEE>.HHVEZEEAHVEOEEEEHAHHVEHEEAHVEEEEEXHEZHEH EO.AH+E.HVOEEEE.AH+E.HVOmEEEnHH+EaHEOEEnAH+EnHEOmE.EEH HE.HV EEHEE HIEEHHZEH H+E+H*OOH+2EEH*OOOHHEEH N OE OE AO.EE.EEEE£H+E.HOEN EnAH+E.HVEE.AH+E.HEEOE.AH+EnHEzE>nAHVEzE>nAHVEOEE.AHOEHEE.HHVE.ZHEEO«EEEnEEEEE.OEOEnOEEE.EOOE.OEoE.OEOm.OEEE O“EEEEE.OEEH.OmxmaEEEEH.EEOEEEAEEOEEEQE.EEEEEEAEEOEH<.EEEEEH.>HEE.E E2EO.EEOE«E.EOE>.EEOoEnEEEEEnOEEEEAEOEEE“EHEAEEEEEnEEEE.Ez“EEE<.OEE.>EEQE .HEEszEO aAEVEEEO nHEEEEE nHEVEEEEE .HEVOE EOE nAEEOEOE nHEEEOOE aAEOOEEO «HEVOEOE aAEOOEEO “AE.EVEEEEH aHEQEE VEEEEE nAEnEVOmEH nHEnEVOExE "AE.EVEEOEEE .AE.EOEEOEEE .HOHVE2E> nE HOHVEOEE «AONVEEE .AEVEEEEEE .AEEE .AEEEEEEEH .HEVEEEE nAOm.EvEE .E AEVEEEO nHNHaEnEVEEOE .AE.EAEVEEoEH< .AE.EVEEOOE nHNH.EVOEEE< .ANHE «EVOEEEE .ANHAEVEOEEE .HNH.EAEOEH< nANH.E.EEEEEE .AOHVEEE .HOHVOEE EE «ENHQOHVEE “HOHVOE .AOHVOE .AOHVEO .HNH«OHVEOHE .ANHAOHVEOE nANN HnOHVEOE AHNH.OHVEEO .AOHVOEE .ANH.OHVOEE nE.OHEHEEH nHEaOH.OHVEH EOEEH .AE.EQEVEEOEEE .HE.EAENvoEExE nE.OHEEE .HONOEEEEE ZOHEZEEHE AEEEEEH.EzHEEHamEEEH.2EEHnEnHE EEEEEO EEHEEOEEEE 281 EZE a a n H*H*AEEE.EH\*E*QEEHE H.OHEE.EH\*E* EEH H.OHEE.EH\*H*QEEH"H.OHEEnEH\*E*nEEH.H.OHEE«EH\N *E*.x:nH.OHEEnEH\*E*.xE.H.OHEE«EH\*E*AEE.E.OHEE.EH\*E*AEEE\*O*nx:HH nE.OHEE.EH\*E*£E.E.OHEE«EH\*<*JS.H.OHEE.:.OHEN«H.OHEN«EHE EEEEOE E A*NH*«EEEAEE.EH\*HH*QEEHQH.OHEE.EE.EH\*OH*QEEHAH.OHEEAE EE.EH\*E*anH.H.OHEEQEEAEH\*E*.EEHaH.OHEEAEEAEH\*E*.E:.H.OHEEQEE.EN H\*E*.EE«H.OHEE.EEQEH\*E*AEEAE.OHEE«EE.EH\*E*«EOE\*E*QEEH.E.OHEEAEH EQEH\*N*«EEEE.OHEEQEEQEH\*H*AEE.H.OHEEQE.OHEN«H.OHENan.EHV EEEEOE E A\\\\\NHQ* EOHEEE EOE EEEEEEE*AEOHAOmHE EEEEOE E O O EEEEEE EEzHEEOO N EEH.AE.HuE.AE.HVEEEEHV.2EH.AEanEQAEQHEEEEEEEQZEHQAE.HHEE EHE.HVEE2HV.zEH.HEQHHEAAEnHvOmEEVQZEH.AHOEEE.AEQHIE.HEAHVEEOEEEVQHE E“HIE«AE.HVEEOEEEV«2EH.AHVOqEE.AHEOE.AHEOEAAHEEO.HHEOEOmnAHVOmOmAHE ABUFF AD AF AID AIDREQ ALGA ALGAAS ALGE ALGI ALGOR GLOSSARY The amount of received hostility that a nation will ignore if its source is an ally. (59)* A non-leader's own estimate of expected threat. (39-40) The degree to which an alliance member has followed its alliance leader's suggestions concerning defense expenditures. (55—56 International assistance in the form of grants which may include consumption, investment, or defense goods. (53—56) The request made by an alliance member to its alliance leader for assistance. (56) The aspiration level for economic growth adapta— tion rate; the speed with which the growth goal will rise or fall given success or failure. (33— 34) The aspiration level for economic growth adapta— tion asymmetrygthe degree to which a nation will more readily raise its growth aspiration level given success and more reluctantly lower it given failure. (33—3”) The aspiration level for economic growth emulation factor; the degree to which the achievement of a significant other nation with respect to economic growth will influence the nation's own growth goal. (33—34) The aspiration level for economic growth inertia factor; the propensity of the growth goal to remain constant given no stimulus for change. (33—34) The aspiration level for economic growth operational- ization rate; the rate at which closure is sought between the desired rate of growth and the current rate of growth. (AA) *Numbers in parentheses refer to the pages in Chapter II where a fuller discussion of the concept is to be found. 282 ALGRO ALLY ALPA ALPAAS ALPE ALPI ALPOH ALPOR ALSEC ALSID APOW 283 The aspiration level for economic growth; the rate of growth which the nation considers desirable at a point in time. (33—34) The relationship between nations where each is committed to assist the other in the event of an attack on one of them. The aspiration level for political stability adaptation rate; the speed with which the political stability goal will rise or fall given success or failure. (32—33) The aspiration level for political stability adaptation asymmetry; the degree to which a nation will more readily raise its political stability aspiration level given success and more reluctantly lower it given failure. (32—33) The aspiration level for political stability emulation factor; the degree to which the achieve— ment of a significant other nation with respect to political stability will influence the nation's own political stability goal (32—33) The aspiration level for political stability inertia factor; the propensity of the political stability goal to remain constant given no stimulus to change. (32—33) The aspiration level for political stability; the desired probability that the decision—makers will retain their decision—making positions. The aspiration level for political stability operationalization rate; the rate at which closure is sought between the desired level of political stability and the current level of political stability. (AA) The aspiration level for national security; the proportion of its resources a nation desires to allocate to defense needs to counter expected threat. (3H-A3) The propensity for nations to discount hostile verbal statements when estimating hostile inten— I tions. (39) g The expected threat—countering capability of a leader's alliance. (38) APPF BC BCGR BCP CS CSGR CSmax CSmin CSP DBC DFC DL EBUFF ECAF 284 The alliance preference pricing factor; the proportionate export price decrease or increase given allies and non—allies respectively. (48) Basic capability; value which has the characteristic of being able to produce more value. Basic capability generation rate; the basic capability value produced by the commitment of one unit of resources (TBC) to the investment sector. (13) The proportion of national resources (TBC) devoted to the investment sector for the production of BC value. (14) Consumption satisfaction; value which has the characteristic of being able to satisfy population wants and needs. (12) Consumption satisfaction generation rate; the consumption satisfaction value produced by the commitment of one unit of resources (TBC) to the consumption sector. (13) The amount of CS value that would be produced if all national resources were devoted to the consumption sector. (17) The minimum level of CS value production necessary to maintain the nation.(17) The proportion of national resources (TBC) devoted to the consumption sector for the production of CS value. (14) The depreciation rate for national productive resources. (14) The depreciation rate for total force capability. (14) Decision latitude; the degree to which decision— makers are dependent upon validator support for their continuation as decision-makers. (20—21) The degree to which the expression of hostile feelings will be suppressed for the sake of maintaining close economic ties. (60) The emergency CS allocation factor; the propensity of the regime to react to a crisis of support by acceding to the validators' wishes with an emergency CS allocation. (44) EGPRI ESPF EXPRC FC FCCUE FCGR FCic FCP FicP GR HOST IEFCT IMLIM IP RULE 4 285 The national priority assigned the goal of economic growth. (53) The economic strength pricing factor; the degree to which export prices are raised or lowered in response to differential abilities to pay. (48— 9) The set of national export prices indicating the terms of trade for each commodity pair with each potential trade partner. (47-50) Force capability; the value which has the characteristic of being able to destroy other value. (12) The suggestion as to level of defense expenditure which an alliance leader makes to an alliance member (36—39) Force capability generation rate; the force capability value produced by the commitment of one unit of resource (TBC) to the defense sector. (13) The amount of total force capability (TFC) devoted to internal security. (19) The proportion of national resources (TBC) devoted to the defense sector for the production of FC value. (14) The proportion of total force capability (TFC) devoted to internal security. (45) The set of national generation rates including CSGR, BCGR, and FCGR. (13) The aggregate flow of threats, accusations, and protests between nations. (56—60) The effectiveness of identive power to cause an alliance member to conform to its alliance leader's defense allocation suggestion. (42—43) The national import limit for a specific good. (46) Information processing rule number four, the rule by which expectations as to future behavior are derived from present behavior; also referred to as the null rule. (24) I I I | IP RULE 3 IP RULE 2 IP RULE 1 IP RULE O IPOW IRPF ITAF NSPRI OPOW PBUFF PDTBC POH PR 286 Information processing rule number three, the rule by which expectations as to future behavior are derived from the present behavior and the last change in behavior; also referred to as the incremental rule. (25) Information processing rule number two; the rule by which expectations as to future behavior are derived by averaging past behavior; also referred to as the mean rule. (25) Information processing rule number one; the rule by which expectations as to future behavior are derived from the detection of trends. (25—26) Information processing rule zero; the rule which specifies the use of rules one through four depend— ing on which rule would have yielded the best prediction had it been used in the previous period; also referred to as the pragmatic rule. ) Identive power; the amount of moral suasion that an alliance leader can exert over an alliance member by the manipulation of symbolic rewards. (Lil-43) International risk pricing factor; the degree to which export prices are raised in response to the reception of hostile communications. (49—50) International trade autarky factor; the propensity of a nation to import goods. (46) The national priority assigned the goal of main— taining national security. (53) The expected threat that an alliance leader believes its alliance should be able to counter. (38) The degree to which the transmission of hostile feelings will be suppressed in response to differ- ences in national power between actor and target. (58-59) The rate of growth in productive resources (TBC) from one period to the next. 4) Probability of office—holding; the overall measure of the stability of the national political system. (21) Probability of revolution; the degree to which the validators are organized in opposition to the decision—makers. (22—23) _O PSPRI REACT SURPLUS TAID TBC TF TFC TOTIM TRADE UPOW VScs VSm VSns 287 The national priority assigned the maintenance of political stability. (53) The propensity to react to the reception of hostility by transmitting hostility. (56-57) The amount of resources left for allocation after primary goal needs have been met. (55-56) The total value of aid that is to be sent to another nation. (56) Total basic capability; national resources avail— able for the production of value. (13) The degree to which two nations are economically coupled through trade linkages. (55) Total force capability; the amount of national force capability available for use in any given period of time. (13—14) The total amount of imports that will be allowed to enter the nation in any given period. (46) The exchange of commodities by two nations. (45— 51) Utilitarian power; the degree to which an alliance member is economically dependent on its alliance leader. (41) Validator satisfaction with regard to consumption satisfaction; the degree to which validators give support to decision—makers in response to CS value flow. (18—19) Validator satisfaction overall; the aggregate support of decision—makers by validators. (20) Validator satisfaction with regard to national security; the degree to which validators are content with the international position of their nation. (19—20) BIBLIOGRAPHY Adelman, Irma, and Morris, Cynthia Taft. Society, Politics and Economic Development. Baltimore: John Hopkins Press, 1967. Alker, Hayward R., Jr., and Russett, Bruce M. "Indices for Comparing Inequality," Comparing Nations, ed. Richard L. Merritt and Stein Rokkan. New Haven: Yale University Press, 1966, 349—372. Alker, Hayward R., Jr., and Brunner, Ronald D. "Simulating International Conflict: A Comparison of Three Approaches," International Studies Quarterly, XIII, No. 1 (March, 1969), 70—110. Almond, Gabriel. "Research Note: A Comparative Study of Interest Groups and the Political Process,” American Political Science Review, LII, No. 1 (March, 1958). Aron, Raymond. Peace and War. New York: Praeger, 1967. Banks, Arthur S., and Textor, Robert B. A Cross—Polity Survey. Cambridge, Mass.: MIT Press, 19 3. Banks, Arthur S., and Gregg, Phillip M. "Grouping Political Systems: Q—Factor Analysis of a Cross—Polity Survey," American Behavioral Scientist (November, 1965), 3—6. Benson, Oliver. "Simulation of International Relations and Diplomacy," Computer Applications in the Behavioral Sciences, ed. Harold Borko. Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1962, 574-595. Berrien, F. Kenneth. General and Societal Systems. New Brunswick, N. J.: Rutgers University Press, 1968. Bhagwati, Jagdish. The Economics of Underdeveloped Countries. New York: World University Library, 1966. Blalock, Hubert M. Social Statistics. New York: McGraw— Hill, 1960. Bonini, Charles. Information Processing and Decision—Making in the Firm. Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1957. 288 289 Boulding, Kenneth. "Toward a General Theory of Growth," The Canadian Journal of Economics and Political Science,XIX, No. 3 (August, 1953), 326—340. Braybrooke, David, and Lindbloom, Charles. A Strategy of Decision. New York: The Free Press, 1963. Brody, Richard A., Benhour, Alexandra H., and Milstein, Jeffrey S. "Hostile International Communication, Arms Production, and Perception of Threat: A Simula— tion Study." Stanford, Calif.: Institute of Political Studies, Stanford University, July, 1966. Brody, Richard A. "Some Systemic Effects of the Spread of Nuclear Weapons Technology: A Study Through Simulation of a Multi—Nuclear Future," Journal of Conflict Resolution, VII, No. 4 (December, 1963), 665—753. Burgess, Philip, and Robinson, James. "Alliances and the Theory of Collective Action: A Simulation of Coali— tion Processes,” International Politics and Foreign Policy, ed. Rosenau, James N. New York: The Free Press, 1969, 640—653. Chadwick, Richard W. "Definition of Simulation Threats Accusations and Protests." Evanston, Ill.: North— western University, Department of Political Science, April, 1965. Chadwick, Richard W. "Developments in a Partial Theory of International Behavior: A Test and Extension of Inter—Nation Simulation Theory." Unpublished Ph.D. dissertation, Northwestern University, 1966. Chadwick, Richard W., and Deutsch, Karl W. "Doubling Time and Half Life," Comparative Political Studies, I, No. 1 (April, 1968), 139—1u5. Chadwick, Richard W. "An Inductive, Empirical Analysis of Intra— and International Behavior, Aimed at a Partial Extension of Inter—Nation Simulation Theory," Journal of Peace Research, No. 3 (1969), 193-214. Chenery, Hollis B. "Patterns of Industrial Growth,” American Economic Review (November, 1960), 124—154. Coplin, William D. World Politics Simulation, II. Detroit: Wayne State University, Department of Political Science, 1967. Cyert, Richard, and March, James. A Behavioral Theory of the Firm. Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1963. 290 Deutsch, Karl W. The Analysis of International Relations. Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1968. Deutsch, Karl W., and Singer, J. David. "Multipolar Power Systems and International Stability," World Politics, XVI, No. 3 (April, 1964), 395. Deutsch, Karl W., and Eckstein, Alexander. "National Industrialization and the Declining Share of the International Economic Sector, 1890—1959," World Politics, XIII, No. 2 (January, 1961), 267—299. Deutsch, Karl W., Bliss, Chester I., and Eckstein, Alexander. "Population, Sovereignty, and the Share of Foreign Trade,” Economic Development and Cultural Change, X (July, 1962), 353-366. Driver, Michael J. "A Cognitive Structure Analysis of Aggression, Stress, and Personality in an Inter-Nation Simulation," Lafayette, Ind.: Purdue University, August, 1965. Durkheim, Emile. Professional Ethics and Civic Morals. London: Routledge and Paul, 1957. Easton, David. A Framework for Political Analysis. Engle— wood Cliffs, N. J.: Prentice—Hall, Inc., 1965. . A System Analysis of Political Life. New York: John Wiley and Sons, 1965. Elder, Charles D., and Pendley, Robert E. ”An Analysis of Consumption Standards and Validation Satisfaction in the Inter—Nation Simulation in terms of Contemporary Economic Theory and Data," Evanston, Ill.: Simulated International Processes Project, Northwestern University, November, 1966. "Simulation as Theory Building in the Study of International Relations," Evanston, Ill.: Simulated International Processes Project, Northwestern University, July, 1966. Etzioni, Amitai. Political Unification. New York: Rinehart and Winston, 1965. Feierabend, Ivo K., and Feierabend, Rosalind L. "Aggressive Behavior Within Polities, 1948—1962: A Cross—National Study," Journal of Conflict Resolution, X (September, 1966), 249—271. Forcese, Denis. "Power and Military Alliance Cohesion." H/U’pfiblished Ph.D. dissertation, Washington University, \a 1968. 291 Forrester, Jay. Urban Dynamics. Cambridge, Mass.: The MIT Press, 1969. Gift, Richard E. "Trading in a Threat System: The U.S.— Soviet Case," Journal of Conflict Resolution, XIII, No. 4 (December, 1969), 418-437. Gregg, Phillip M., and Banks, Arthur S. "Dimensions of Political Systems: Factor Analysis of a Cross—Polity Survey," American Political Science Review, LIX (1965), 602—614. Guetzkow, Harold, and Cherryholmes, Cleo H. Inter—Nation Simulation Kit. Chicago: Science Research Associates, Ind., 19 . Guetzkow, Harold, et a1. Simulation in International Rela— tions. Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1963. Guetzkow, Harold. "Some Correspondences between Simulations and 'Realities' in International Relations," New Approaches to International Relations ed. Morton A. Kaplan. New York: St. Martin's Press, 1968, 202—269. Gurr, Ted R. "A Causal Model of Civil Strife: A Comparative Analysis Using New Indices," American Political Science Review, LXII (December, 1968), 1104—1124. Why Men Rebel. Princeton, N. J.: Princeton University Press, 1969. Haas, Ernest, and Whiting, Allen. Dynamics of International Relations. New York: McGraw—Hill Book Company, 1956. Heilbroner, Robert L. Understanding Macro—Economics. Englewood Cliffs, N. J.: Prentice—Hall, Inc., 1968. Hermann, Charles F., and Hermann, Margaret G. "An Attempt to Simulate the Outbreak of World War I," American Political Science Review, LXI, No. 2 (June, 1967), 400—416. . "Validation Studies of the Inter—Nation Simula— tion." China Lake, Calif.: U. S. Naval Ordinance Test Station, December, 1963. Hermann, Charles F. "Validation Problems in Games and Simulations with Special Reference to Models of International Politics," Behavioral Science, XII (May, 1967), 216—231. 292 Homans, George C. Social Behavior: Its Elementary Forms. New York: Harcourt, Brace and World, 1961. Huntington, Samuel P. "Arms Races: Prerequisites and Results," Approaches to Measurement in International Relations, ed. John E. Mueller. New York: Appleton— Century-Crofts, 1969, 15—33. Kaplan, Morton. System and Process in International Politics. New York: John Wiley and Sons, 1957. Kindleberger, Charles P. Power and Money. New York: Basic Books, 1970. Kuh, Edwin, and Meyer, John R. "Correlation and Regression Estimates When the Data Are Ratios," Econometrica, XXIII (October, 1955), Moo—416. Kuznets, Simon. "Economic Growth of Small Nations," The Economic Consequences of the Size of Nations, ed. A. Robinson. New York: St. Martin's Press, 1965, 14—32. . Modern Economic Growth. New Haven: Yale Univer— sity Press, 1966. "Quantitative ASpects of the Economic Growth of Nations: VI. Long—Term Trends in Capital Formation Proportions," Economic Development and Cultural Change, IX (July, 1961, Part II). Lagerstrom, Richard P., and North, Robert C. ”An Antici— pated—Gap, Mathematical Model of International Dynamics," Stanford, Calif.: Institute of Political Studies, Stanford University, April, 1969. Lamb, George G. "Basic Concepts in Subjective Information Theory, Thermodynamics and Cybernetics of Open Adaptive Societal Systems.” Evanston, Ill.: Northwestern University, Department of Chemical Engineering, August, 19 9. "Engineering Concepts and the Behavioral Sciences," General Systems, XII (1968), 165—169. ”A Frame for Analysis of Societal Change: Basic Presuppositions," Evanston, Ill.: Northwestern Univer- sity, Department of Chemical Engineering, December, 1968. Lasswell, Harold D. ”Introduction: The Study of Political Elites,” World Revolutionary Elites, ed. Harold D. Lasswell and Daniel Lerner. Cambridge, Mass.: The MIT Press, 1965, 3—28. 293 Meier, Dorothy L., and Stickgold, Arthur. "Progress Report: Event Simulation Project— INS l6," Evanston, Ill.: Simulated International Processes Project, Northwestern University, 1965. Milstein, Jeffrey S., and Mitchell, William C. "A Quantita— tive Analysis and Predictive Computer Simulation," Peace Research Society (International) Papers, X (1968), 161—213. Naylor, Thomas H., et al. Computer Simulation Techniques. New York: John Wiley and Sons, 1966. North, Robert C. "Decision—Making in Crisis: An Introduc- tion," Journal of Conflict Resolution, VI, No. 3 (September, 1962), 197—200. "Steps Toward Developing a Theory." Stanford, Calif.: Studies in International Conflict and Integra— tion, April 24, 1967. (Mimeographed.) Nurske, Ragnar. "Patterns of Trade and Development," Economics of Trade and Development, ed. James D. Theberge. New York: John Wiley and Sons, 1968, 85—102. "The Size of the Market and the Inducement to Invest,” Development and Society, ed. D. E. Novack and R. Lekachman. New York: St. Martin's Press, 1964, 91—96. Organski, A. F. K. The Stages of Political Development. New York: Alfred A. Knopf, 1965. World Politics. New York: Alfred A. Knopf, 1958. Pen, Jan. A Primer of International Trade. New York: Vintage Books, 1966. Pendley, Robert E., and Elder, Charles D. "An Analysis of Office Holding in the Inter—Nation Simulation in Terms of Contemporary Political Theory and Data of the Stability of Regimes and Governments," Evanston, Ill.: Simulated International Processes Project, Northwestern University, November, 1966. Pruitt, Dean G. "An Analysis of Responsiveness between Nations," Journal of Conflict Resolution, VI, No. 1 (March, 1962), 5-18. Pryor, Frederic L. Public Expenditures in Communist and Capitalistic Nations. Homewood, Ill.: Irwin, 1969. 294 Raser, John R. Simulation and Society. Boston: Allyn and Bacon, Inc., 1969. Raser, John, and Crow, Wayman. WINSAFE II: An Inter—Nation Simulation Embodying Capacity to Delay Response. La Jolla, Calif.: Western Behavioral Sciences Insti— tute, July, 1964. Rashevsky, Nicholas. Mathematical Theory of Human Relations: An Approach to a Mathematical Biology of Social Phenomena. Bloomington, Ind.: Principia Press, 1947. Rosecrance, Richard. Action and Reaction in World Politics. New York: Little, Brown, 1963. Rosenau, James N. "Pre—theories and Theories of Foreign Policy," Approaches to Comparative and International Politics, ed. R. Barry Farrell. Evanston, Ill.: Northwestern University Press, 1966, 27—92. Rummel,Rodolf J. "Dimensions of Conflict Behavior Within and Between Nations," General Systems Yearbook, VIII (1963), 1-50. . "Indicators of Cross-National and International Patterns," American Political Science Review, LXIII (March, 1969), 127—147. "The Relationship Between National Attributes and Foreign Conflict Behavior," Quantitative Inter— national Politics, ed. J. David Singer. New York: The Free Press, 1968, 187—214. "A Social Field Theory of Foreign Conflict Behavior," Peace Research Society (International) Papers, IV (1965), 131-150. Russett, Bruce M. International Regions and the International System. Chicago: Rand—McNally, 1967. Russett, Bruce M., and Taylor, Charles L. "Is Political Instability Related to Per Capita Income?" Bendix Corporation, Working Paper No. 97. Ann Arbor, Mich.: Bendix Corporation, January 23, 1967. Russett, Bruce M. "Measures of Military Effort," American Behavioral Scientist (February, 1964), 26—29. "Who Pays for Defense?" American Political Science Review, LXIII, No. 2 (June, 1969), 412—426. World Handbook of Political and Social Indicators. New Haven: Yale University Press, 1964. 295 Salmore, Stephen A., and Hermann, Charles F. "The Effect of Size, Development and Accountability on Foreign Policy," Paper delivered at the Seventh North American Peace Research Society (International) Conference, University of Michigan, Ann Arbor, Mich., November 11—12, 1969. Sawyer, Jack. "Dimensions of Nations: Size, Wealth, and Politics," American Journal of Sociology, LXXIII (September, 1967), 145—172. Schelling, Thomas C. The Strategy of Conflict. Cambridge, Mass.: Harvard University Press, 1960. Sherman, Allen W. "The Social Psychology of Bilateral Negotiations." Unpublished Masters thesis, Department of Sociology, Northwestern University, 1963. Simmel, Georg. The Sociology of George Simmel. Translated and edited by Kurt H. Wolf. Glencoe, Ill.: The Free Press, 1950. Singer, J. David. ”The Level—of—Analysis Problem in Inter— national Relations," The International System: Theoretical Essays, ed. Klaus Knorr and Sidney Verba. Princeton, N. J.: Princeton University Press, 1961, 77-92. "Threat—Perception and the Armament—Tension Dilemma," Journal of Conflict Resolution, 11, No. 1 (1958), 90—105. Skinner, Donald D., and Wells, Robert D. Michigan Inter— Nation Simulation. Ann Arbor: The Department of Political Science and the Center for Research on Learning and Teaching, University of Michigan, 1965. Smoker, Paul L. "An International Process Simulation: Theory and Description," Evanston, Ill.: Simulated Interna— tional Processes Project, Northwestern University, 1968. Sombart, Werner. Die Deutsche Volkswirtschaft in Neun— zehnten Jahrhundert. 3rd Edition. Berlin: Bondi, 1913. Stockholm International Peace Research Institute. SIPRI Yearbook of World Armaments and Disarmaments, 1968/69. New York: Humanities Press, 1970. Tanter, Raymond. ”Dimensions of Conflict Behavior Within and Between Nations, 1957—1960," Journal of Conflict Resolution, X (March, 1966), 41—64. TEMPER: Technological, Economic, Military, and Political Evaluation Routine. Vol. 1—7. Bedford, Mass.: Raytheon Company, 1965—66. 296 Tsantis, Andreas C. "Political Factors in Economic Development," Comparative Politics, II, No. 1 (October, 1969), 63—78. Turing, A. M. "Computing Machinery and Intelligence," Computers and Thought, ed. Edward A. Feigenbaum and Julian Feldman. New York: McGraw—Hill, 1963, 11—35. United Nations. Statistical Office. Statistical Yearbook, 1967. New York: United Nations Publications, 1967. U. S. (1963—1969). President (Johnson). Economic Report of the President. January, 1967. Washington, D. C.: United States Government Printing Office, 1967. Viner, Jacob. "Barriers to Economic Development," Develop— ment and Society, ed. D. E. Novack and R. Lekachman. New York: St. Martin's Press, 1964, 81—90. Walker, Helen M., and Lev, Joseph. Statistical Inference. New York: Rinehart & Winston, 1953. Weede, Erich. "Conflict Behavior of Nation—States." Paper delivered at the Midwest Meeting of the Peace Research Society (International), Northwestern University, Evanston, Illinois, April 17, 1969. Wesley, James P. "Frequency of Wars and Geographical Opportunity,” Journal of Conflict Resolution, VI, No. 4 (December, 1962), 387—389. Wright, Quincy. The Study of International Relations. New York: Appleton—Century—Crofts, Inc., 1955. A Study of War. 2nd Edition. Chicago: Univer— sity of Chicago Press, 1965. Zinnes, Dina. "A Comparison of Hostile Behavior of Decision— Makers in Simulate and Historical Data,” World Politics, XVIII, No. 3 (April, 1966), 474—502. MICHIGAN STATE UNIV. LIBRRRIE WIHWHHWIHWIWI‘HI’IWIIWIIHIIHIIHIWI 31293000870026