A COMPUTER SIMULATION EXPERIMENT WITH SELECTED COMMUNICATION AND INDIVIDUAL BEHAVIORAL VARIABLES IN THE BUSINESS FIRM BY Henry Paul Sims, Jr. This research is a computer simulation experiment involving a limited number of factors that affect the productivity of a business organization. The focus is on the psychological motivation of the individual in a business organization, and how that motivation influences the productivity of the organization. The primary objective of the research is to demonstrate how computer simulation can be useful as an experimental vehicle to develop additional insight into the behavior of individuals in organizations. The specific setting is a hypothetical business firm that manufactures and markets a multi-product line. The primary decision considered in the model is the allocation of sales effort among different products in the product line, and the effect of that allocation on the firm's profitability. Henry Paul Sims, Jr. For behavioral theoretical background, the research relies primarily on the work of Vroom regarding partici- pation in decision-making. The Specific mechanisms through which participation may influence productivity are examined, particularly the classification of participation effects into "decision quality" and "ego- involvement." The model also draws upon the work of Likert regarding the effects of communication links in an organization. Selected communication links are incor- porated for study. In particular, the link between production and sales concerning knowledge of inventory position is considered. Also included for investigation is an intercommunication link between territorial salesmen regarding characteristics of the market. The primary reSponse variable that is selected for analysis is the profitability of the firm. A complete 26 factoral experimental design is utilized, concentrating on the following experimental variables: ° Stability of demand: stable vs. volatile. ' Market reSponse to sales effort: uniform response vs. non-uniform response. - Knowledge of inventory by sales department: no knowledge vs. knowledge. ° Salesmen intercommunication: no information exchange vs. information exchange. - Sales manager style: authoritarian vs. equalitarian. Henry Paul Sims, Jr. ' Salesmen personality: low vs. high need for ’ independence. Two factors, stability of demand and the inventory communication link, were found to have little effect on profitability, primarily because of counterbalancing mechanisms that tended to result in Opposing effects. The intercommunication of salesmen was found to be insignificant because of the lack of substantive infor- mation transfer. The strongest effect on profitability was caused by participation in the sales-effort allocation decision, eSpecially when considered in conjunction with valid market information and salesmen characterized by a high need for independence. The model demonstrated a structure which explicated the partialling of the effects of participation into an information transfer component and an ego involvement component, each acting independently of the other. The experiment has provided an initial step toward the integration of "macro" and "micro" type simulations, as well as demonstrated the viability of simulation as a means of explicating the structural mechanism underlying behavioral theory. The major contribution of the experiment is the demonstration of the usefulness of computer simulation as a vehicle for the exploration of behavioral theory. A COMPUTER SIMULATION EXPERIMENT WITH SELECTED COMMUNICATION AND INDIVIDUAL BEHAVIORAL VARIABLES IN THE BUSINESS FIRM BY Henry Paul Sims, Jr. A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Management 1971 DEDICATION To My Parents ii ACKNOWLEDGMENTS I wish to extend my sincere appreciation to several people who have contributed to this thesis: To Dr. Richard F. Gonzalez, for his continuous understanding and support throughout this course of study. To Dr. John Gullahorn, for his stimulating ideas, eternal optimism, boundless encouragement, and sincere friendship. To Dr. Jeanne Gullahorn, for her penetrating insight and advice. Above all, my special gratitude is extended to my wife Laurie--who faithfully believed in this project and accepted my frequent 5:00 a.m. returns from the Computer Center. This research was supported by PHS Research Grant No. MH-16935, National Institute of Mental Health, and by the Computer Institute for Social Science Research, Michigan State University. iii Chapter 1 TABLE OF CONTENTS INTRODUCTION Theoretical Background Methodology Background Macro and Micro Models The Nature and Purpose of this Experiment STRUCTURE OF THE MODEL The Organization Program PAJAMA Subroutine FCAST Subroutine INVENT Subroutine GENPRO Subroutine PRREC Subroutine SLSMEN Subroutine SLSGEN Subroutine BOOK Subroutine NEWPRO Subroutine EXEC Utility Subroutines Summary EXPERIMENTAL DESIGN DESCRIPTION OF RESULTS Factor A: Stability of Demand Factor B: Market Response Interaction Effect: B-E Factor C: Inventory Link Factor D: Salesmen Intercommunication Factor E: Participation Interaction Effect: B-E Factor F: Need for Independence Interaction Effect: E-F Summary of Results The Problem of Realism in Simulation iv Page Chapter 5 SUMMARY AND CONCLUSIONS Effect of Demand Stability Effect of Inventory Communication Link Effect of Salesmen Intercommunication Effect of Participation Future Research Summary BIBLIOGRAPHY APPENDIX I: Detailed Flow Chart APPENDIX II: Glossary of Variable Names APPENDIX III: Program PAJAMA APPENDIX IV: Initial Conditions Program APPENDIX V: Analysis of Variance Output Page 95 95 97 101 104 107 108 110 117 140 145 168 169 Table 3-1 4-2 5-1 5-2 LIST OF TABLES Experimental Variables Summary of Analysis of Variance Summary of Results Variation of Promotion Expenditures with Factor A Effects of Standard and Alternate Levels of Factor C Profit Under Different Conditions of Factor D vi Page 75 78 89 97 102 103 Figure 1-1 2-10 2-11 2-12 5-1 LIST OF FIGURES Vroom's Concept of the Relationship Between Motivation and Performance Organization Chart Organizational Activities Information and Decision Flow Overall Flow Chart Subroutines FCAST and INVENT Flow Chart: Subroutine GENPRO Flow Chart: Subroutine PRREC Sales Effort Allocation Process Stages in the Product Life Cycle Hypothetical Wiebull Distribution Flow Chart: Subroutine SLSGEN Flow Chart: Subroutine EXEC Effect of Inventory Communication Link The Effects of Participation vii Page 28 29 30 34 36 39 40 43 48 50 56 64 100 107 CHAPTER 1 INTRODUCTION What causes people to willinaly contribute to the pro- ductivity of an organization? The quest for the answer to this question is a continuinr challenge to both managers and behavioral scientists. Occasionally relatively simple answers are proposed, e.r.. a "Theory Y" versus a "Theory X." But simplified theories cannot adequately explain the ex- ceedingly complex interactions of the myriad of factors that influence an organization's performance. Indeed, even the concept of "productivity" has become rather blurred in recent years. In the past, profitability was normally a sufficient measure of productivity. Today, however, a business organi- zation is accountable not only to stockholders or owners, but also to employees, customers. and ultimately to society as a whole. In addition to the responsibility of achievinm prof- it objectives. management must also face the problem of ful- filling a multitude of objectives which are frequently ill- defined and difficult to measure. In an overall sense, a business firm continually attempts to optimize an ill-defined multiple criterion function, with constantly changinr oper- ating constraints. This research is an experiment involving a limited 2 number of factors that affect the productivity of a business organization. The focus is on the psychological motivation of the individual in a business organization, and how that motivation influences the productivity (as measured by prof- itability) of the organization. A secondary focus is con- cerned with the effect of selected communication links on productivity. The level of technology as well as the tech- nical competence of individuals is assumed to be constant and is not directly considered. The primary objective of the research is to demonstrate how a relatively new methodol- ogy-~computer simulation--can be useful as an experimental vehicle to develop additional insizht into the behavior of individuals in oraanizations. It is an exploration aimed at elucidating behavioral theory by subjecting that theory to a simulation process and observing whether or not the simula- tion approach leads to an understanding of gaps. or incon- sistencies, that may exist within the formulation. Theoretical Background In discussing the interrelationship of work and motiva- tion, Vroom (196a, p. 6) specified three phenomena which have attracted the attention of behavioral scientists: 1. The choices made by persons among work roles. 2. The extent of their satisfaction with their chosen work roles. 3. The level of their performance or effectiveness in their chosen work roles. This research concentrates on the third phenomenon. In 3 particular. it focuses on the effect participation in deci- sion—making has on productivity. The specific setting is a hypothetical business firm that manufactures and markets a multi-product line. The primary decision considered in the model is the allocation of sales effort among different prod- ucts in the product line, and the effect of that allocation on the firm's profitability. For behavioral theoretical background, the research re- lies primarily on the work of Vroom (1960. 1964) regarding participation in decision-making. In a study of supervisors, he found that individual personality differences constitute important moderators in the relationship between participa- tion in decision-making and performance. The amount of psychological participation was found to be highly related to performance in those subjects who were high in need for independence: however, such participation was unrelated to the level of performance in those scoring low in measured need for independence. Vroom's findings regarding the mod- erating effects of subordinates' need for independence are incorporated into this model. In this experiment. two types of individual salesmen are specified. The first type has high need for independence and derives intrinsic satisfaction from the opportunity to exert perceived influence on the decision-making of his superiors. The second type has low need for independence and places no personal value on the opportunity to influence decision-mak- ing. It is recognized that in a given population of actual n individuals, the attribute of need-independence will be dis- tributed along a continuum. For purposes of experimental control, however, the model assumes dichotomous polar levels of need-for-independence. The specific mechanisms through which participation may influence productivity are also examined by this research-- particularly the classification of participation effects into "decision quality" and "ego-involvement." In discussing decision quality, Vroom (i96h, p. 7) observed that: Differences in the quality of decisions reached demo- cratically and autocratically are probably dependent on a large number of factors. including the extent to which supervisors and subordinates have information relevant to judging the organizational consequences of different courses of action and the extent to which the interests of each are in harmony with organizational objectives. It is extremely unlikely that democratic decisions are always higher in quality than autocratic ones or vice versa. In addition to the information quality aspects of deci- sion-making, this experiment also attempts to demonstrate the relationship between participation and ego-involvement in de- cisions. and how this relationship can influence the produc- tivity of the organization. It is also possible that people become "ego involved" in decisions in which they have had influence. If they have helped to make a decision it is "their decision," and the success or failure of the decision is their success or failure. Intuitively it would appear that the amount of personal involvement of people in de- cisions is dependent on the amount of influence they have had in the decision and on the extent to which they pride themselves on their ability to make that kind of decision. If, for example, a person who con- ceives himself to be a brilliant scientist shares in the making of a decision which he believes requires scientific judgement. the outcome of that decision is a test of the adequacy of his self-conception. A 5 successful decision confirms his self-concept: an unsuccessful decision threatens it. On the other hand, when he helps to make a decision on an ad- ministrative matter he has less "at stake." Nei- ther a successful nor an unsuccessful decision would be greatly inconsistent with his self-con- cept (Vroom. 196a, p. 7). The model also relates individual motivation and per- formance to the overall performance of the organization. in- corporating an elementary theory of individual motivation. the Vroom concept of "Expectancy Theory" (1964). The struc- ture of the model thus is dependent on the theoretical and empirical framework used to portray individual behavior in organizations . Vroom's conceptualization of Expectancy Theory may be referred to as the instrumentality-valence model. which is stated in two propositions: Proposition 1. The valence of an outcome to a person is a monotonically increasing function of the alge- braic sum of the products of the valences of all other outcomes and his conceptions of its instrumentality for the attainment of these outcomes. Valence is defined as "anticipated satisfaction from an Outcome." Proposition 2. The force on a person to perform an act is a monotonically increasing function of the algebraic sum of the products of the valences of all outcomes and the strength of his expectancies that the act will be followed by the attainment of these outcomes. It is also assumed that people choose from among alternative acts the one corresponding to the strongest positive (or weakest negative) forces (vrocm, i96u. p. 17). Thus, in Vroom's model, each individual has internal attributes such as expectancies. instrumentalities, valences 6 of particular states. forces to perform particular acts, and skill levels. An individual's performance may be regarded as a func- tion of his ability and motivation. In this model. all sales- men are assumed to have equal and constant ability and tech- nical competence. variations in performance. then. depend only on changing levels of motivation. vroom specified three possible relationships between motivation and performance: linear. negatively accelerated approaching an upper limit. and an inverted U function. vroom generally favors the inverted U function. quoting evidence, and offering two possible explanations. One reason for the decline of performance under very high levels of motivation may be a narrowing of the cognitive field. Another reason may be an association between anxiety and a high level of motivation. In this model. it is assumed that the levels of motivation in organizations are generally not the very high levels that are associated with a reduced cognitive field or high anxiety. Therefore. the declining portion of the in- verted U function is ignored. resulting in a negatively accelerated relationship between motivation and performance that reaches an upper limit. PERFORMANCE MOTIVATION Figure 1-1. Vroom's Concept of the Relationship Between Motivation and Performance 7 In the model. individual performance of a salesman is defined in terms of the total amount of personal effort that a salesman expends in selling his products to customers. The individual level of motivation (effort) is regarded as a choice made by the individual. In terms of the instrumental— ity-valence model, he selects a level of effort that is con- sistent with the highest possible force. For salesmen with high need-independence, the opportunity of influencing his superior's decision-making is regarded as possessing high in- strumentality because the outcome of perceived influence with the superior is intrinsically rewarding. Opportunity to parti- cipate therefore becomes a motivating factor for the sales- man. For salesmen with low need-independence. the oppor- tunity to exert influence in decision-making is regarded as possessing low instrumentality because the outcome of per- ceived influence with the superior is not regarded as re- warding in itself and is therefore not a motivating factor. The model provides a mechanism by which the effect of the salesman's motivation-effort reaction is related to the determination of the overall performance of the firm. This experiment also draws upon the work of Likert re- garding the effects of communication links in an organiza— tion. In particular, the model incorporates selected cross- communication links that are included as determinants in decision processes. Likert defined the "linking-pin" communication link concept as a key causal variable that differentiates a 8 "System 2" from a "System 4" type of managerial style. System h management...uses an overlapping group form of structure with each work group linked to the.rest of the organization by means of persons who are members of more than one group. These individuals who hold overlapping group member- ship are called "linking pins" (likert, 1967. p. 50). He concluded that the information transfer and motiva- tional effects of group decision-making are key behavioral elements that lead to high performance by an organization. Likert defined his System 2 as "benevolent-authorita- tive" while System h was described as "participative-group." System 2 is characterized by a traditional "chain-of-command" authority structure while System h is characterized by supportive relationships. group decision-making, and high performance goals. In this experiment. selected communication links are incorporated for study. In particular, the link between production and sales concerning knowledge of inventory posi- tion was considered. Also included for investigation is an intercommunication link between territorial salesmen regard- ing characteristics of the market. In summary. this research attempts to incorporate se- lected theoretical and empirical conclusions of Vroom and Likert as the primary theoretical framework for the experi- ment. Methodology Background The influence of human behavior on the performance of an organization is a complex. interactive process. Various 9 investigative techniques have traditionally been useful in the quest for knowledge about behavioral processes. One meth- odology that has been in the forefront in the development of social science theories is field research, or "survey" re- search (as typified by the work of the Institute for Social Research at the University of Michigan, or the Ohio State Leadership Studies). A second technique. laboratory experi- mentation. has been useful in isolating specific behavioral variables from random variations and confounding factors. thus facilitating the observation of outcomes that are strictly controlled by the experimental treatment. A.third methodology that has received significant and increasing attention is computer simulation. Simulation. as a technique, has been primarily utilized as a tool for the prediction, understanding. and explanation of existing sys- tems (system analysis) or systems undergoing design (system design). In the field of business. computer simulation has been most frequently applied to the analysis of production, transportation. or marketing systems by focusing on the dynamic flow of materials or products. It has served as a substitute for more precise analytical optimization methods when such methods were inapplicable to complex systems. This experiment, however, is concerned with the use of simulation as a means of exploring social science theory in the context of the business firm. Since the focus is on behavior, rather than on the flow of materials, the experiment is classified as'a "behavioral" simulation. 10 Simulation. of course. is far from new: "...man has been simulating things ever since he first scratched a pic- torial representation of some real object on the wall of a cave" (Conway, et al.. 1959. p. 92). Simulation is new, how- ever. if one considers the subtle changes in the definition of the word. Webster's (1967. p. 811) rather broad defini- tion of "simulate” ("to give the appearance of") can be applied to a wide variety of pictures. schematics, physical models, mathematical models. etc. In management and be- havioral science literature. however. the word simulation has taken on a variety of new connotations: "One fact is certain. There is a computer simulation semantic jungle.. ..no consensus exists among writers on a definition of com- puter simulation" (Tarter. 1970. p. 2). The following definition. however, best characterizes the intent of the present research: Computer simulation is an experimental method which utilizes a digital computer to operate a dynamic model for the purpose of theory development. The subtle distinctions between the terms theory. model. and simulation are not always apparent. Theories are pur- posely broad. They attempt to provide a generalization to explain many specific cases. In the physical sciences. theories are generally formulated in mathematical terms. but. in the social sciences, real-world phenomena are not so amen- able to mathematical expression. Social science theories are typically depicted verbally. although attempts are fre- quently made to reduce verbal expression to the more precise 11 and rigorous mode of mathematics. . . .some social scientists have sought refuge in the domain of mathematics. The introduction of strict. emotionally neutral definitions and rigorously logical reasoning has considerable appeal. . . (Gullahorns Emonograph. forthcoming] ). A theory. in entirety, may be too general to be tested in any meaningful way. A more specific expression may be required if a theory is to be tested: ergo, the need for a model. A model may thus be regarded as a specific manifes- tation of a theory. It is the unique mechanism by which theory is translated into a specific set of conditions and circumstances for the purposes of explanation and experimen- tation. Theory provides the basic framework for specifica- tion of a model and is the starting point for building the model. Barton (1970, p. 25) has discussed the relationship between theory and models: Theories are purposely broad; they try to generalize over many specific cases. This is really an economy of communication so that man does not need to pass hundreds and hundreds of detailed instances from one generation to the next. . . . To use theory. one must make the generalizations of theory specific enough to guide the making of observations and the taking of actions. Models serve this purpose. Theory, therefore. is frequently expressed as a mathe- matical model. For complex systems. however, a mathematical description may be intractable to a closed analytical solu- tion. Also, the restrictions imposed by the number of vari- ables that can be interrelated by a mathematical model are limited. Other alternatives are needed. Because of its speed and symbol-manipulating capabilities, the digital 12 computer provides a tool for the implementation of such an al- ternative. Not only mathematical. but logical models as well, can be specifically formulated for operation on a com- puter. The result is a computer simulation. A computer simulation is a unique kind of model. It serves not only as an explication of a theory, but also pro- vides the opportunity to set the model to work: to observe the behavior of the model over time and under varied sets of conditions. Since it typically incorporates the interrela- tionships of the system variables over time, a computer simulation is generally dynamic. A simulation is an abstrac- tion of not only the static structural relationships. but the dynamic, process relationships as well. "It is the exhibition of process that distinguishes simulation from such static models as blueprints. dolls. etc." (Crow. 1967. pp. 11. 12). A behavioral simulation may thus be thought of as a special kind of dynamic model. i.e.. an abstraction of reality that attempts to describe the socio-psychological' components of the reality and to specify the dynamic nature of the relationships among those components. Computer simulation provides a unique vehicle for the study, development, and explication of theory. Verba (196a. p. #99) discussed the role of simulation in the development of theory: One of the major contributions of simulation research L' 13:] to the development of theory. . . . The process of designing a simulation forces the designer to explicate his model. . . . One is forced to make explicit what may have been 13 the implicit assumptions about the subject matter --in order that these assumptions may be placed in the operating model . . . . a simulation differs [in] that it goes on from there. The explicated model is set to work and. over time. generates data on subsequent states of the simulation. A computer simulation can be regarded somewhat analagous to a laboratory experiment in that it provides the advantage of being able to attain a high degree of experimental con- trol. Like a laboratory experiment. the results may thus lead to further confidence in the theory. disprove the theory. or possibly lead to further investigation that may modify the theory. Frijda (1967) also supports the concept of computer simulation as a useful tool for the development of theory: . . .computer programs can serve as unambiguous formulations of a theory. The program language is precise: the meaning of a given process is fully defined by what it does. . . . [It] is a means to demonstrate and test the consistency and sufficiency of a theory. If the behavioral data which the theory wants to explain are in fact re- produced by running the program. the theory has been proved capable of explaining these facts. Moreover. running the program under a variety of conditions may generate consequences of the theory which can be tested against new evidence. These consequences may be unforeseen and they may be quite important. . . . Extensive experimentation is possible by running different versions of the program 0 e e 0 One of the byproducts of simulation is the necessity of guaranteeing sufficiency and completeness. "The computer simulation model will not operate if you forget anything. If you fail to take into account some necessary mechanism that. in the verbal description of a theory you might readily pass over. the computer simulation will not run" (Feigenbaum. 1963, p. 5). The Gullahorns found the same in their simulation of 1h George Homans' theory of social exchange: . . .the very process of translation [from verbal theory to computer programZJforces one to be pre- cise about variables and their relationships and thus helps one to recognize ambiguities in ex- pression and implicit assumption in the verbal mod- el. For example. if the verbal formulation con- tains qualifying phrases such as "other things being equal." in programming one must define pre- cisely what these "other things" are and what it means for them to be "equal“: otherwise the computer simply will not operate--the programmed theory will not generate the expected consequences (Gullahorns. 1962. p. 5). In the past. the primary development of theory has re- sulted from the collection of empirical data. The social scientists' objective was to devise a theory that would ex- plain the data. The more data explained by the theory. the more confidence in the validity of the theory. Computer simulation does not alter this objective, nor will simulation replace empirical data as the primary source of theory development. Rather, simulation becomes an addi- tional tool. one that can lead to further insight regarding the theory. Through the computer. it becomes possible to perform analyses that would otherwise be impossible: to make further deductions regarding the dynamic implications of com- plex systems. The use of computer simulation as a tool for theory de- velopment has been recently established in all the behavioral sciences, ranging from political science (which focuses on the nation as a behavioral entity) to psychology (which focuses on the individual as a behavioral entity). Selected examples may be found in Ableson and Bernstein (1963), Amstutz 15 (1967), Carroll (1969), Guetzkow (1962). Guetzkow, et. a1. (1963). McPhee (1961). Schelling (1961), and Tomkins and Messick (1963). Some attention has also been directed to- ward the use of computer simulation as a methodological technique in doctoral dissertation research. A number of dissertations that employ computer simulation as a research methodology can be found in Tuason (1965), Bellman (1969). Uhr (1969), Chervany (1968). Siemens (1967). Cleveland (1967). Miller (1962). Summit (1965). Desjardins (196a). Hutchinson (196h), Wallace (1961), Green (1960), Benson (1963). Hershaur (1969). Stallings (1970), Parker (1970), Hunt (1970), Fondren (1963), Kaczka (1966). Gensch. (1967), Norek (1970). Bright— man (1970), Srinivas (1970), Bettman (1969), and Weber (1970). Macro and Micro Models In this investigation of human behavior in organiza- tions. computer simulation is employed to generate an arti- ficial history of a hypothetical firm for purposes of an- alysis. The research attempted to integrate two major trends which have characterized behavioral simulation: the "macro" versus the "micro" approach. The Gullahorns (1969. p. 3) take note of these trends in their review of social system simulations: In endeavoring to organize the varied types of simulations that were retrieved in our biblio- graphic search, we have distinguished between models that emphasize universal processes affec- ting a social system as opposed to those that focus on micro-behavioral processes within socio- cultural contexts. The distinction is by no means clear-cut. However, while not all of the models 16 simulating total system process exhibit a "black box" approach to individual decision making, they nevertheless tend to trivialize autonomous infor- mation processing. The simulations we have termed micro-behavioral, on the other hand, incorporate social psychological considerations and tend to de- emphasize total system processes. In business-related simulations. the "macro" approach typically involves modeling the marketing. production, and financial functions as well as the flow of materials. prod- ucts. information. and profits along with associated de- cision processes. The decisions in such a large-scale simulation generally are not associated with an individual nor specified in detail. Examples of the large-scale ap- proach include the Cyert and March model of price and out- put behavior in the firm, the Bonini simulation of informa- tion and decision processes. and the Kaczka model of the behavior of work groups. The Cyert and March (1963) simulation dealt with the output and price decisions of a business firm operating in an oligopoly market. The basic decision processes of the firm reflected the March and Simon (1958) concepts of "satisficing." The firm in the model makes three basic decisions during each time period. First, the price to be charged for the product is determined. Second, the number of units to be produced in the next time period is decided. Finally, the firm decides on the amount of sales effort and the amount allocated to sales promotion. The decisions are made subsequent to a comparison of 17 past performance with goals. In effect. decisions are made on the basis of past results. Goals are adjusted according to a lagged relationship with performance. If problems exist, the firm may implement a search for alternatives to solve the problem. The firm operates under a type of manage- ment by exception since search is avoided if all appears to be going well. The experiment on the Cyert and Xarch model is imple— mented through the use of a number of runs. each with random variations in the values of the model parameters. The re- sults are analyzed by regression techniques to determine which parameters exert the strongest influence over the per- formance of the firm. In summary. the Cyert and Narch model presents empirically based behavioral processes that are rel- evant to the determination of the price and output deci- sions of the business firm. Bonini's work incorporated the simulation of the infor- mation and decision process in a hypothetical business firm. His model was an attempt to synthesize some of the relevant theory from several disciplines--economics, accounting, marketing, organization theory, and behavioral science. His :nodel of the business firm included the traditional functions 0f production and marketing along with an executive group in) oversee the operations. The objective of the simulation was: to study the effects of certain environmental, informa- tiomual, and organizational factors upon the performance of the firm. 18 The decision processes in the Bonini model were heavily influenced by the work of Narch and Simon (1958). and Cyert and March (1963). concerning the concepts of "satisficing" rather than "maximizing." The model was distinctive because of its emphasis on the behavioral aspects of the organization. In particular. Bonini incorporated the concept of an "index of pressure." which is a function of performance relative to goals. When an organization is failing to perform up to expectations. there is a tendency for pressure to build up within an organization. and this pressure generally results in attempts to achieve better performance. (Bonini. 1963. p. 19) The study was basically an experimental simulation that investigated the effects of changes in the external environ- ment. the information system. and the decision system. Among other findings. one major conclusion was that firms with relatively stable environments were slow to take advan- tage of profit opportunities. while firms faced with highly variable environments generally. after considerable search, ended up with higher sales and/or lower costs. An important conclusion generated by the study was the Point contrasting a priori versus a posteriori theorizing. IHe emphasized how it was often easy. after the fact. to ex- Dlain how a mechanism would work. but emphasized that such explanations should not mean that the results were necessarily trdsvial or obvious beforehand. Although the Bonini simulation was not intended to be appfilicable to any particular company's problems. it was 19 important because of the demonstration of the capability to abstract selected behavioral aspects of the information sys- ten and decision processes in the firm. Eugene Kaczka has constructed a computer simulation model that concentrates on the behavioral aspects of the production sector of the business firm. His experiment ad- dressed the question of whether a managerial climate which is employee oriented results in higher levels of performance than a task oriented climate. His research employed a fac- torial experiment on a complex model of a hypothetical in- dustrial organization that was composed of four structural components: the market. the executive level. middle and lower management levels. and work groups. As measured by both economic and socio-psychological criteria. he found performance to be significantly affected by managerial climate. His model demonstrated general sup- port of the hypotheses of Likert (1967): Several of the dimensions of managerial climate. and interactions of these dimensions yielded higher levels of organizational performance under employee-orientation than were realized under task-orientation. The notable exception was low cost emphasis which yielded poorer performance. In summary. the findings indicate that the most efficient levels of performance result when con- cern for cost performance is wed with a concern for the employees of the organization. (Kaczka, 1966. p. 230-231) The "micro" approach, on the other hand. has typically beern.concerned with the simulation of individual decision Processes. but generally ignores the process by which the outczome of these decisions affects the total operation of the 20 organization. Examples of the micro approach include the Smith simulation of the personnel selection decision process. the Clarkson simulation of the stock selection decision proc- ess. and the Gullahorn and Gullahorn simulation of George Homans' theory of exchange in interpersonal behavior. Smith's (1968: Smith and Greenlaw. 1967) simulation dealt with a decision process in personnel selection. In particular. the simulation attempted to emulate the decision process of a personnel psychologist who utilized test scores and other data about individuals for selection and placement into various types of clerical-administrative positions. The simulation was designed to produce not only a specific employment recommentation. but also various interpretive comments about each applicant. His model was proven capable of producing a strong correspondence between the human psychologist and the output of the computer simulation. In his experiment. he found a 9&1 level of agreement between the human decisions and the simulated psychological inferences. His research suggested the possibility of utilizing computerized interpretive programs as prescriptive models for personnel selection. Also mentioned was the possibility of utilizing a similar model for computer-assisted instruc- tion of industrial psychology students. In the long run. however. his major contribution lies in the fact that . . . Research of this type gives deeper insight into the manner in which people resolve problems. The methodology allows the researcher to map an 21 equivalent thought process at a particular point in time and could permit the study of the effects of aging and experience on decision- making capabilities of individuals and groups. (Smith. 1968. p. 329) Another research effort that involved the simulation of individual decision-making was the Clarkson model of the trust investment process. The principal objective of the study was to develop a behavioral. as opposed to normative. theory of trust investment. A.trust investment office is charged with the responsibility of allocating a given invest- ment amount among various stock investment alternatives. Clarkson's model was constructed by observing the decision logic of an investment officer and by incorporating that logic into a heuristic program for computer simulation. Clarkson's basic findings indicated that his model provided a reasonable prediction of both the actual portfolios as well as the decision processes by which selection was made. One of the better known micro-models is HOMUNCULUS. the Gullahorns' simulation of interaction between two (or more) individuals (196“. i969). HOMUNCULUS is not only a model of a social system but is also a micro-process because of the elaborate decision system of each individual in the model. Each individual is defined by a list structure representing personal attributes such as identity. values. attitudes. and a memory structure that may change with each interaction event. The decision that each individual makes is contingent upon his values. attitudes. and memory structure. The model is based on George Homans' theory of social 22 exchange. where two individuals act in face-to-face inter- action and provide rewards and punishments to each other. The construction of the model requires the translation of five verbal prepositions advanced by Homans (1961. pp. 53-111) into operational computer statements. The particular situa- tion described by the model is a dyad. where one person re- quests help from another in a Job related task. In the in- teraction event. both may benefit and both may pay a price in a sort of psychoeconomic exchange. The decision process that guides each individual's actions depends on the payoffs that his past interactions have elicited. The input is composed of the personalities and past histories. defined by attributes assigned to the two in- dividuals. The output data consists of the verbal responses of the two individuals as they proceed through a sequence of interaction events. HOMUNCULUS has also been used to investigate triads (196a) and has found that. for individuals who were initially strangers. an isolate and a dyad results. For triads who are initially friends. the results are similar to the dyadic interaction. The primary value of the Gullahorns' work is the fact that a social science theory. formerly expressed only ver- bally. was proven tractable to expression by a computer pro- gram which demonstrated the internal consistency of the theory. 23 The Nature And Purpose Of This Experiment Currently. the Gullahorns are undertaking a project whose long-term objective is to integrate a "microprocess" individual decision-making model (similar to their HOMUNCU- LUS model) into a "macroprocess" that models the functions of a large-scale firm. The interfaces between the microproc- esses and macroprocesses of the system will occur at the decision nodes. When the need for a decision is specified by the macrosystem. the inputs to the decision node will be- come the inputs (or stimuli) to the microsystem. The micro- system will emulate the decision process of an individual and will emit a decision (to the macrosystem) after being operated on by the socio-psychological attributes of that individual. The current simulation endeavor provides the challenging opportunity to advance. . .by building a model of an organization in which individual values. decisions. and behavior pro- duce system results; and in turn. organization values. adaption. and integration feed back to influence individual's values and behavior (Gullahorns. 1970. p. 1). As an intermediate step in this endeavor. this experi- ment involves a simplistic firm with a relatively simplified decision process in order to concentrate on the feasibility of integrating a macro and micro process into the same model. In essence. the model is designed to be complex enough to produce realism in terms of output. but simple enough to be manageable. Considering time constraints and considering the state of the art. the experiment. therefore. is designed 2b to meet three limited objectives: 1. To provide a means of incorporating and analyzing complex organizational relation- ships and selected communication and indi- vidual behavior factors that may. in com- bination. impinge on the performance of the organization. 2. To provide experimental control in order to isolate specific causal relationships. 3. To provide a means of evaluating the inte- gration of micro- and macro-processes into a single model. In summary. the research is a planned experiment on a hypothetical business firm. performed by computer simulation. to investigate the conditions under which selected person- ality variables and communication links can affect the pro- ductivity of the firm. The experiment is intended to be an exploration into the use of computer simulation for explica- tion of behavioral theory. CHAPTER 2 STRUCTURE OF THE MODEL The model developed for this research is a simplified representation of the operations of a hypothetical business firm that manufactures and markets a line of products. The model utilizes a "fixed-time" interval (of one month) of timekeeping and transcends a total time of 100 months for each run. The model was written in FORTRAN Iv and computa- tions were performed on a Control Data Corporation Model 6500 computer. Nominal demand for each of the firm's products is exogenously defined and may be described by a unique product life cycle (see Subroutine SLSGEN). Demand. of course. is influenced by pricing and promotion policies. sales effort. and random factors. Since the model assumes that the firm is a well estab- lished existing organization with a history. initial condi- tions (primarily attributes of the existing product line) must be assigned to simulate the history of the firm. A pro- gram which was used for random generation of initial condi- tions is shown in Appendix IV (PROGRAM IC). The model also assumes a given and constant production capacity function over the length of the run. Therefore. capital investment decisions are not considered. The 25 26 manpower system is also assumed to be constant (both in a qualitative and quantitative sense). and the flow of man- power is not considered.1 The Organization The firm in the model is composed of a chief executive officer. an accounting department. a production function. and a marketing function. Figure 2-1 shows a breakdown of these functions. This chart is a representation of the for- mal structure of the firm. The elements in the chart are restricted to only those elements of the firm which are actually incorporated into the model. It is recognized that a firm. which exists in the real world. even a small busi- ness. will be considerably more complex than the organiza- tion reflected by the chart. .A general summary of the organization activities is shown in figure 2-2. This chart specifies the primary responsi- bilities. decisions. and actions that are assigned to each position and department in the firm. Each of the individuals and/or departments in the model receives specific types of information and makes an explicit set of decisions or performs a specific set of actions. Figure 2-3 is a summary of the information-decision flow pro- cesses in the model. listed according to position. 1For a Capital Investment simulation. see Cleveland, (1967). For a Manpower Planning simulation. see Weber, (1970L 27 Program PAJAMA ProgravaAJAMA is the executive program that implements and controls each replication.2 After the values for the block initial conditions are read in. a loop is then estab- lished for each run with the values for the experimental variables defined by a zero-one generating algorithm. For each subsequent run. initial conditions are then reset. . The program also calls each of the subroutines that perform the dynamics of the simulation. An overall flow chart showing the executive program and subroutines is shown in Figure Z-h. 28cc Chapter 3 for definition of replication and run. 28 .253 20.5.5333! 7N 0.5m. u _ .2535 _ 363$ E<535 3535 $533 5333 N :2 8x _ _ 223,. azufifiewd 43 $— hzuxhxu_u :33 use... o» 5:! 32. 0333:. .m -9. e. 3.3 3303? .m Gabauuxuv co. a 30m 9.0.35.5 3e: sci—eon out.“ 3.8. .N 32.5359 :89 “to... 5.333 ._ Lounge: 9.3973: .5: assesses-58¢ ate...“ Bug—.95 2.0333 839.. :93 use out“. scene; m0 303:... «>323. .N Causation me: .5602 5...: noted an... .3395 .25”; .85 uCoLa unea .. Leu— ta 022690 $020 92.2:ng zo . .255 uz. mzo.._.u< 93 9.2m. uma own: 2t auzuuux 22:25.2. zo.h.moa .003... 0:00. mun 0.:3. .. .u 00.0.3.3 0.0.00 00 23:3. .m 30.. 0.0.0... :0 030.00.... :0.»00:0..:.000.. 5.00.53 0:0 00.... .N 300000.. 0.0. 00. 00 090.32 .. .55.... 8.0.3.... 0.0.0» ..0..0>0 00m “um «a... 3300:0300: :0. 00... :03 0:0 00.3 9.0.. .»m<0.. 00000.0. 0.0.00 0» 000.39. .m 0.030003 0... a 0.0.. 0000000.. 2.0.33 00 5.00.0: :. 0.0.8 30000.0. .~ 0:00.30 0:0 2.0.3... .:0:.00.0n :000 50.: 00.00 .. no.0. 00.00 00000... .M 0:00.00.:u00.> 0:.00x00z .u 0.002.00 :0.00:00.... .m 30.. 0.0.00. :0 0.530.... 0.200.?» 33:02.. .N 0...... 33:03. 0030.. u .E.. {00:02. . :03: . . 3.353 :0. 00:00.3 0. 300:0... .502... .26. 33:02.. 0.0.02.8 2.52... 0...... $00303. 00» 33:02.. 00.3028 3030):. 0:000... 0:00 .00....00... 9.00.00: .n 1.0:... «0000.0... 00.3 A.unoo 9:.0 3.532 .553 —0>0— >000:0>:. 0:000... .N 3...... >..0u:0>:. 3...... 00.0.. 00.00 .. 0.00.03 0.30.3... .m 300703.00. > 8.3.8... 2.02.353 mun 0.52 u .332): 3.3.3 3:0... 8... Sign... -3... 5.82.9... .8... 208.33 “.333 3.330.... .. 3.5.5.... 0.30323: .. 0.2.2.3 5.33.8... .. 5.9.2.3.... 530...... 82.3. 9.0.3 5.8.5... .2333 on... 3.3 .253 ... 3028.... «:3 J. 328.3 .0 03 ... euoa.u.u 3.2. 3 3.5:...80 255...... unseen to»... 3.3. .m {0:0 .23.... new .n some 5 0.530... .m 05.0.. «3.5 .o :3. 6 9.35.. Tonnes... no.8 32.... 9.3.3.. 3.3.... to...» .5543 3.3.0.... .23.... 3 02.83.. . 5.3.... an..." .N «0.3 .2032. «um .N .0 Sn... :0 9.3.5.. .N .53.... .33..» 339.0“. «0.3. .. 3.3 303.3... .. I... 3.3 «8.3.... .. 5.30.3 9.3.3.. 520.... 333...} {1 3.: 2.0.... no.3 “.2...“ 5‘28 .0 005...»... .... .> 838...... .. 5...; 3.3.3.... 3.3. .323. 8. a 33 2.0:...» :00. mcu 0.52 .... €030.60... .252 0.83 2.9.7.... 502.02.. 30.3.... .26. 33:02.. .~ -3352: 30.3.3 .N 0.0.8 .N «0.1.0.6.. c0308... ~30 3...... 380 8325.... 339 3.83 9:22. E352: .5521». 3.2.. .. 3...... 32.5.8 .. 8:38... .o .8. ._ 2.288.. 33:02.. 333 “:0... 2.05.82. 2 2.0.5.5 gave... .. 3:300 “52...... .. .30.: 3.9.... 9.09.5 .. 9:33“ 3h READ STARTING CONDITIONS EACH REPLICATION (Program PAJAEA) 1 EACH RUN ESTABLISH TIME LOOP GENERATE VALUES FOR EXPERIMENTAL FACTORS DEFINE INITIAL CONDITIONS (Program PAJAMA) 1 EACH PERIOD FORECAST SALES l (Subroutine FCAST) MA PR CE & P OMOTIO I s ULE O RECOMMENDATIONS [(Subroutlne INVENT) (Subrout inc PRREC ) ‘ 1 1 [SCHEDULE & GENERATEl PRODUCTION (Subroutine GENPRO)‘ ALLOCATE SALES EFFORT (Subroutine SLSMEN) I G L ACT AL LE (Subroutine SLSGEN) & BOOKKEEPING (Subroutine BOOK) EACH QUARTER NEW PRO DROP PRODUCT DECISIO gLSubroutine NEWPRO) PRICE & PROMOT O DECISIONS (Subroutine EXEC) EACH RUN & O SUMMARY STATISTICS (Subroutine SUMARY) Figure Z-h. Overall Flow Chart 35 Subroutine FCAST The objective of this subroutine is to provide a mecha- nism by which each salesman is able to forecast the next- period sales for each product in his territory. The fore- cast is based on the sales of the last period, adjusted for a trend factor. The forecasting algorithm is based on the double expo- nential smoothing method (Clark and Schkade. 1969. pp. 705-6; Brown. 1963, pp. 128-132). The double exponential method is used because of the capability to correct for trend. The steps in computing a double-smoothed forecast are: . Calculate new smoothed average SAVE(K.L) = ALPl * OHDERS(K.L) + (l-ALPI) * PSAVE(K,L) where: SAVE = New smoothed average ALFI = Smoothing constant ORDERS = Current orders PSAVE = Previous smoothed average . Calculate change in smoothed averages CHANGE = SAVE(K.L) - PSAVE(K.L) . Forecast change in orders DELOR(K.I) = AIPI * CHANGE + (i-ALPI) * DELOR(K.L) where: DEIOR = Smoothed change in orders . Compute forecast of orders for next period FORC(K,L) = SAVE(K,I) + (l-ALPI) * DELOR(K,L)/ALP1 36 PORECAST ‘PCRPCAST SALES TEICE‘SILESEAN (BASED ON LAST RACE PRODUCT) NONIR'S SALES P TREND CF SAIFS) INVENTORY (EACH PRODUCT) CALCULA E INVENTORY LEFT PRCE LAST MONTH ADJUST LINITS ON INVENTORY (BASED ON LAST MONTH'S SALES) EELON LL ' 1' EXCESS COMPARE INVENTORY WITH LIMITS ‘PROPOSE INCREASE WITHIN IN INVENTORY LIMITS ‘PROPOSE DECREASE ‘ VIN INVENTORY DECREASE PRESS E SE RESSURE ON SALES DEPT. | '19N SALES DEPT. PROPCSE SMALL INCREASE IN INVENTORY Figure 2-5. Subroutines FCAST and INVFNT Subroutine INVENT The level of inventory to be scheduled for the end of the next period is determined by the Production Vice-Presi- dent. The objective of this subroutine is to provide a mech- anism for establishing the scheduled inventory level. The logic used in this subroutine is adapted from the Cyert and March (1963. pp. 149-236) general model of price and out- put determination. Two limits are pertinent in the scheduling of inven- tory; UINL. the upper inventory level limit, and HINT. the lJDwer (or "runout") inventory level limit. The upper limit .143 a level that, when exceeded, leads to excessive cost of 37 maintaining unneeded inventory. The lower limit. when breech- ed. leads to excessive opportunity costs because of lost sales. The limits are a lagged function of sales--rising as sales rise. falling as sales fall. UINL(K) = .ALP? * UINI(K) + (i-AlP?) * 0.3 * TSAL(K) RINL(K) = .ALP7 * RINL(K) + (1-ALP7) * 0.1 * TSAL(K) where: UINL ‘= Upper inventory limit. ALP? = Smoothing constant. TSAL = monthly sales. RINL = Runout inventory limit. Subsequent to adjusting the limits to reflect the most recent sales information, the current inventory level is com- pared with the limits. If current inventory level is in ex- cess of the upper limit. a decreased level of inventory (for next month) is scheduled, pressure on the Sales Department is increased to sell the excess inventory, and the inVentory limits are decreased. If current inventory is below the lower limit, an in- creased level of inventory (for next month) is scheduled. pressure on the Sales Department is decreased, and inventory limits are raised. If current inventory is within limits, a small increase in the level of inventory is scheduled. This logic, defined by Cyert and Xarch, (1963, p. 151) 118 a.form of adaptive behavior in the face of incomplete in- formation: Each firm makes decisions on the basis of feedback 38 from past results. The firms adjust goals. . . on the basis of such feedback. In investigating the production planning procedures of a number of real firms. Wilson (1969, p. 152) found strong support for the Cyert and March model: We have been using. . .a technique, which is heuristic rather than a mathematical model. work- ing with upper and lower bounds. . .on inven- tory. . . The problem is based on trying to keep inventory between a maximum level and a minimum level. . . . Subroutine GENPRO The objective of subroutine GENPRO is to provide a mechanism for scheduling and generating the actual produc- tion of each product. The amount of production to be scheduled depends on the forecast, inventory-on-hand, and scheduled inventory: SCHED = TFORC(K) - INV(K) + SCINL(K) where: SCEFD Production schedule. in units -TPORC = Sales forecast INV = Current inventory SCINL = Scheduled inventory level Actual production for each product is randomly generated in the model from a normal distribution with mean equal to the scheduled production, and a standard deviation equal to 2&5 of the scheduled production. 39 zoo-2&9 PRODUCTION CALCULATE PROPOSED PRODUCTION SCHEDULING FOR NEXT PERIOD ‘“““""’ (BASED ON FORECAST. PROPOSED (FOR EACH INVENTORY & INVENTORY ON RAND PRODUCT) J PRODUCTION CENERATE ACTUAL PRODUCTION (FROM NORMAL DISTRIBUTION (FOR EACH WITH PROD. SCHEDULE AS MEAN) PRODUCT) ' ~ 250-299 Figure 2-6. Flow Chart: Subroutine GENPRO Subroutine PRREC In subroutine PRREC. the marketing vice president makes recommendations to the executive committee regarding price and promotion decisions. The logic underlying this section of the model is based on the work of Cyert and March (1963) and Kaczka (1966). From the marketing vice-president's viewpoint. his goal is to accomplish the sales forecasts that have previously been made, (subroutine FCAST). His decisions depend on a comparison of current forecast with previous forecast, and the actual performance of sales in relation to the previous forecast. The behavior he exhibits is, thus. dependent upon whether or not the sales goal of the previous fore- cast was met. If this forecast is higher than the previous one and the previous forecast was realized, then the situation is viewed as favorable. The pres- sure for sales effectiveness is decreased..... and he recommends that the promotion percentage be main- tained. If the previous forecast was not met. he increases sales effectiveness pressure and may rec- ommend either a price cut or an increase in pro- motion percentage. Uhder the condition where the current forecast is lower than the previous and the prev feel difficulty. pres moti fore for the lm ious forecast was met, he is inclined to that the new goal can be met with little As a result. he maintains the sure on his salesmen and maintains the pro- on percentage. However. if the previous cast was not met, he reacts with a request increase in sales promotion and (increases) current pressure on salesmen. In brief, the marketing vice-president's main functions are the development of sales fore tion of sales goals, to r the are pp- The motion ar subroutine EXEC). casts which serve as sales goals. the direc- the direction of salesmen ealize these goals and the recommendation of price and promotion strategies which he feels needed to achieve the sales goal. (Aaczka. 1966, 153-5h). actual decisions regarding changes in price and pro- e made quarterly by the chief executive officer (see The recommendation of the marketing vice- president is a primary factor that is considered in the deci- sion process. PRICE & IS CURRENT FORECAST PROMOTION HIGHER THAN PREVIOUS YES RECOMMENDATION FORECAST? y “KT MGR wAS PREVIOUS Each Product) FORE CA ST NE T ?> YES DECREASE WAS PREVIOUS PRESSURE ON FORECAST NET? SAIES DEPT. NO NO YES INCREASE PRES- SURE OI‘J SALES REQUEST DEPT. [PROMOTION V L INCREASE REQUESTED MORE RECENT L NAS PROMOTION INCREASE HAN PRICE CUT REQUEST ?> INCREASE T Figure 2-7. Flow Chart: PRESSURE ON SALES YES DEPT. _ g * REQUEST ('\L. PROMOTION uNCRMSE f Subroutine PRREC hi Subroutine SLSMEN Subroutine SLSMEN is the primary representation of the individual salesmen in the organization and their work per- formance. The behavior of the salesmen. as structured in this model. is primarily based on the theoretical framework established in Chapter 1. The particular decision that is considered is the prob- lem of allocating of sales effort among the different prod- ucts. . . . the objective of many sales force manage- ment activities is really to achieve a desired allocation of effort. . . . (Simon and Freimer, 1970. p. 173). The implementation of the sales effort allocation pro- cess is summarized by the flow diagram shown in Figure 2-8. A more detailed flow diagram is contained in Appendix I. The subroutine involves the following sequence of events:3 1. The marketing vice-president. on a monthly basis.' evaluates each product in the company line in order to determine the relative priority of each product. The relative priority. in turn. is used as an input in alloCating sales effort among the different prod- ucts. The basic information utilized by the marketing vice-president is the "index of pressure" that is an attribute of each product. Each product begins its 3The paragraph numbers in this section correspond to the box numbers in the flow diagram in Figure 2-8. #2 life cycle with a nominal pressure index of 1.0. but pressure may be adjusted due to the inventory posi- tion (subroutine INVENT), sales in relation to fore- cast (subroutine PRREC). or profitability in rela- tion to profit goals (subroutine EXEC). Pressure varies between a maximum value of 1.2 and a minimum value of 0.8. The output of the marketing vice president's decision process is a priority ranking of the firm's products. In effect. products are ranked according to relative pressure, with the product under highest pressure assigned the first rank. The information (priority ranking) is then transmitted to the sales force through the regional sales manager. Meanwhile. each salesman is undergoing a similar evaluation process regarding the relative behavior of the product line in his sales territory. The basis for the salesman's evaluation is his perception of the marginal response of the market to the marginal effort that he expands on each product. In effect. he attempts to evaluate how much sales have in- creased as a result of additional effort expended on "pushing" each product. The output of the salesman's evaluation process is a priority ranking. based on marginal response to marginal effort. He attempts to rank the products on the basis of those products which he expects to “3 EXT . v. P. SALESNAN 1 ‘RANK PRODUCT 2 RANK PROD. O. RANK PROD. A ON BASIS OP BASIS OF NAR. ON RASIS O° PRESSURE RESPONSE TO GROSS VOL. ' MAR. EFFORT - - REG. SLS. MGR. 51 RANKING 0F PRODUCTS DUE TO INFLUENCE OF SALESMAN ADJUST P IORITY ESTIMATE PERCEIVED DIFFERENCE BETWEEN MKT. V.P. RANKING & OWN RANKING JUS R S O RANK DUE TO INFORMATION FROM OTHER SALESMEN i I Figure 2—8 respond best to any additional effort. DIFFERENCE AFTER INFLUENCE 7 ESTABLISH MOTIVATING FORCE FOR EACH PROD. 1° ESTABLISH EFFORT TO BE EXPRNDED ON 'EACH PRODUCT 6 ESTIMATE PERCEIVED 1 ’i MAE FINAL PRIORITY RANKING ON BASIS OF RELATIVE VOLUME & SLS MGRS. PRIORITIES J . Sales Effort Allocation Process This evaluation and ranking process is the mechanism by which each salesman "learns" about the unique peculiarities of his sales territory. and how his territory will uniquely respond to sales effort. 3. After receiving the priority ranking from the market- ing vice president, each salesman compares this rank- ing to his own ranking regarding the marginal re- sponse of the market. The disparity in priority an rankings then becomes a base upon which the salesman can evaluate his ability to influence his super- visor, the regional sales manager. The salesman may be provided with the opportunity to communicate with other salesmen in his sales terri- tory. In doing so. the salesmen exchange information regarding the estimated response of the different prod- ucts. Each salesman then modifies his response ranking because of the new information. He considers the evaluations of the other salesmen and adjusts his ranking accordingly. The information regarding the estimate of poten- tial market response is then transmitted to the re- gional sales manager in an attempt to influence his final priority ranking. If the sales manager encourages and permits influence in decision-making, he then considers the information from the salesmen in establishing final priorities. The sales manager combines three sources of informa- tion in the decision process: the ranking from the marketing vice president. the unit contribution to margin of each product (". . .an approach to alloca- tion of effort based on contribution margin would seem fruitful" [:Simon and Freimer, 1970, p. 173:] ) and the information furnished by each salesman. His final priority ranking is then transmitted back to each salesman. #5 When the priority ranking is received from the sales manager. each salesman then compares the ranking with the suggested ranking (response rank) that he had initially transmitted to the sales manager. He is thus able to estimate the difference between his ranking and the sales manager's ranking after his attempt to influence. The salesman is then able to perceive if there has been any change in priorities received from manage- ment due to his (the salesman's) attempt to influence the decision-making. If a change in priorities has occurred. the salesman concludes that he has. in fact. exerted influence in the decision process. If the salesman has a high need for independence. he derives intrinsic satisfaction from the opportu- nity to participate in the priority setting process: The amount of influence of subordinates in decision making. . .affect(s) the speed and efficiency with which the decision is carried out . . . .it is clear that the effects of participation in decision-making are not confined to the nature of the deci- sion. but also extend to the probability that the decision will be effectively implemented (vroom. p. 227-228). In effect, a salesman's motivating force to carry out an assignment tends to be enhanced by the opportunity to exert influence in the decision pro- cess. In the model. the salesman is motivated by per- ceived participation with a negatively accelerating N6 relationship between effort and motivation.“ EFFORT MOngggéNG = f (perceived participation) It is also assumed that, for a salesman with a high need for independence. a lack of perceived participation will result in a negatively motivating force: REPORT “-.____ MOTIVATIHG = f erceived artici a- PORCE (Siam) p p' The motivating force to sell each product is also dependent on the index of the pressure exerted on each individual product. Since the salesman is paid by commission. he tends to emphasize those products that are high volume "movers." He ranks the products with the product of highest gross volume given the first ranking. The salesman's final priority ranking that governs ‘ his allocation of effort among the different products “See previous discussion Chapter 1 regarding relation- ship between motivation and performance. A7 is a combination of his own "volume" ranking and the priority that is received from the sales manager. In the model, the allocation process assumes that a given amount of total effort is to be dis- tributed among the different products. The product given the highest priority is assigned an allocation index number of 1.2. with lower ranking products assigned progressively lower index numbers. until the lowest ranking product is assigned the number 0.8. The mean allocation index number of all the products is 1.0. The salesman thus decides the rela- tive allocation of his time among the full range of products. 10. Finally, the effort that each salesman expends on each product is a combination of his final priority ranking (the allocation index number) and his motiva- ting force: EFFORT = MOTIVATING FORCE X ALLOCATION RANK Subroutine SAIESGEN During each time period, the demand for each product is defined by an exogenous function that calculates the orders as a function of the age of the product, price. promotion. ef- fort by the salesman. and a random component. 229 Product Life Cycle. The demand for a particular product can be expected to ‘change over time. For many products the variation in ue lifetime sales may reveal a typical pattern of development known in marketing literature as the'product life-cycle." According to Kotler (1967, p. 291), five stages in the growth and decline of sales of this kind of product can be distin- guished (Figure 1): Stage 1. Stage 2. Stage 3. Stage A. Stage 5. SALES VOLUME (Introduction.) The product is put on the market: awareness and acceptance are minimal. (Growth.) The product begins to make rapid sales gains because of the cumulative effect of introductory promotion, distribution, and word-of-mouth influence. (Maturity.) Sales growth continues but at a declining rate because of the diminishing number of potential customers who remain un- aware of the product or who have taken no action. (Saturation.) Sales reach and remain on a plateau marked by the level of replacement demand. (Decline.) Sales begin to diminish absolute- ly as the product is gradually edged out by better products or substitutes. Figure 2-9. Stages in the Product Life Cycle The validity of the product life-cycle concept is well established in marketing literature. Cox (1967), sampled 75h ethical drug products introduced in the United States in the #9 years 1955 to 1960 and concluded the following: "Product life cycles not only can be determined: they are particularly useful as marketing models." Polli and Cook. (1969) attempted to empirically verify the product life cycle as a descriptive model of sales behavior for 1&0 categories of non-durable consumer products. These product categories included health and personal care. food, and tobacco products. They found strong theoretical support for the concept in Rogers' (1962) theory of the diffusion and adoption of innovations: "Essentially. the concept implies that a product finds initial resistance of some new ways of be- having, and is purchased by only a limited segment of the buying population. later. as the product's performance and value are known and communicated, a larger segment of buyers adopts and sales begin to increase at a faster pace. Eventually, the rate of growth decreases as the proportion of adopters get closer and closer to a maximum, with most sales representirg repeat purchases. The rate of adop- tion remains constant throughout the maturity phase and diminishes in the declining phase. The link be- tween Rogers' theory and the life cycle concept be- comes obvious if one considers that the logistic curve usually employed to represent the life cycle is the cumulative equivalent of the normal density func- tion. which is precisely the shape of Rogers' adoption function." (Polli and Cook. 1969, p. 386) Based on empirical evidence, they further concluded: "It is clearly a good model of sales behavior in certain market situations - especially so in the case of different product forms competing for essentially the same market segment within a gen- eral Class of products." (Polli and Cook. 1969, p. #00) In the model. a hypothetical sequence of demand that approximates the life-cycle. is generated through the use of a.Weibull (1951) distribution: 50 DEN = KON*AA*ITINE*(8XP(-AA*ITIME2)) where: DEM Nominal Demand during time period t. KON,AA Constants IT ME (Age of product) Figure 2-10. Hypothetical Weibull Distribution For each time period. ITINE. a nominal demand, DEM. can be generated from the Weibull equation. Random fluctuations in sales (for each time period), can be simulated by sampling from a normal distribution with mean DEM, in order to obtain an adjusted value for sales. The nominal demand (DEN) thus reflects the age and the current life-cycle stage of the product. For each product. unique values of the constants of KON and AA are randomly generated by the initial condition program IC. Since the values of these parameters define the shape of the life-cycle curve. the underlying nominal demand is uniquely and exogenously defined over the life of the prod- uct. Nominal demand for each product is thus generated during each time period from a Weibull equation (with unique values of KON and AA) that includes the age of the product (ITIME(K)) as a parameter. 51 Price and Promotion The effect of pricing on the demand for each product is represented by the equation: uprice = (PRICE)PBEL/(20)PREL where: Mprice = Index for the level of sales due to the price elasticity factor PRICE = Price of the product FRET ~ Price elasticity (20)PREL = Normalizing factor. of representing initial condition of PRICE = 20. Price elasticity is randomly generated for each product by the subroutine GENPR. Thus. the response of the market to changes in price is assumed to be exogenously defined. Promotion is assumed to be an aggregate marketing expendi- ture that includes advertising. publicity. field promotion. etc.. but excludes personal selling effort. The response of the market to promotion expenditures is determined exogenously for each product. in a manner similar to the price response: MPROM = (PRON)P"EL/(1OOO)PMEL where : NPRQM Index for the level of sales due to the promotion elasticity factor. PROM = Promotion expenditure PMEL = Promotion elasticity (1OOO)PMEL = Normalizing factor. representing initial conditions of PROM = 1000. Promotion elasticity is also randomly generated for each product (by subroutine GENPM). Thus. the response of the market to changes in promotion expenditure is also exogenously 52 defined. Price and promotion policies. therefore. influence nominal demand according to the following relationship: (Adjusted Demand) = (Nominal Demand) x MPRICE *‘N PROM The logic underlying the use of elasticities to define an aggregate marketing response function is discussed by Urban (1969. Do “1) It can be expected that the consumer's willing- ness to buy a product at a given price will depend on his attitude toward the product's characteristics and appeals. This implies a marketing mix effect be- tween price and advertising. . .an aggregate sales response function will be postulated. The function should include three basic marketing variables: ad- vertising. price. and distribution. . . In unlogged form the formulation would be: - EPI -EAI EDI X31 ‘ 3P31 A 31 31 ' {31 is industry sales of Product j. a is scale constant. le is average price level of all brands in product group j. Ajl is total advertising of all brands in product group j. Djl is total distribution level for all brands in product group j EPl is industry price elasticity for Product j. EAl is industry advertising elasticity for Prod- uct j. EDl is industry distribution elasticity for Product j. This function captures marketing mix effects and allows nonlinearity in response to marketing variables. The nonlinearity is reflected in the parameters EPl. EAl. AND EDl. For example. if 0< EA1< 1. the marginal sales response to advertising would be constantly de- creasing as advertising increases. . . In general. EAl and EDI should be expected to fall between zero and plus one. The price parameter EPl should be negative because as price increases. sales should decrease. The para— meters EAl. EDI. and EPl are elasticities and reflect the proportionate changes in the product group's sales 53 resulting from a proportionate change inaone variable. Equation 1 reflects marketing mix effects since the sales response of one variable depends on other vari- ables as established. for example. by differentiating Equation 1 with respect to price. The marginal response to price changes (dXJ /dP1) depends on the level of advertising and distr but on. In this model. for purposes of simplification. advertising and distribution have been aggregated into the promotion ex- penditure. Also. interdependencies. (complementarity and substitutability) with competitors' products and other prod- ucts of the firm are assumed to be zero. Similar uses of elasticities as a means of defining mar- ket response have been employed in simulations by Bonini (1963). Kaczka (1966), and Kotler (1965). Random Deviation from Nominal Demand Any other factors which may have an effect on demand are assumed to occur randomly and are therefore not explicitly defined in the model. The aggregate effect of these factors is assumed to be contained in a random error adjustment to demand. A percentage deviation from nominal demand is gen- erated from subroutine RNDUX. This random adjustment to nominal demand is calculated by the equation: NOISE = SIGl * ERR x DEM where: NOISE = Random deviation (in units) from nominal demand. ERR = Percentage deviation (returned from subroutine RNDUM) from nominal demand. DEM = Nominal demand. SIGI = Scaling factor. 5a The scaling factor. SIGi. controls the stability of the demand function. For increasing values of $161. the effect of random deviation from nominal demand is increased. and the market becomes more volatile. The value of NOISE is additive to nominal demand. Effort of the Salesman. In the model. the response of the market to the effort of the salesman is approached in much the same way as pro- motion elasticity. Effort is represented by an index num- ber which is utilized to characterize the relative personal selling emphasis that a salesman places on a particular prod- uct. An effort index larger than one means that a salesman spends a larger proportional amount of time "pushing" that particular product. Conversely. an effort index of less than one means that a salesman spends a less than proportional amount of time on that product. The effect of a salesman's effort on the demand for each product is represented by the equation: MEFF = (EFF)EFEL/(1.O)EFEL where: “EFF = Index for the level of sales due to the effort elasticity factor. EFF = Index of relative effort expended by a particular salesman. EFEL = Effort elasticity. (1.0)EFEL = Normalizing factor representing normal conditions of EFF = 1.0. 55 An "elasticity of effort" concept is not new. Lambert and Kniffin (1970. p. 3.“. & 8) used this idea to model a market response to personal selling effort: The response of the market to alternative levels of selling effort can be measured and ex- pressed . . . Sales volume is probably the most easily utilized measure of market reaction . . . it may be very illuminating to examine the magni- tude by which sales volume changes in relation to varying increases and decreases in the level of selling effort . . . Conceptually any measure of market reaction can be coupled with each measure of selling effort for purposes of constructing response functions. . an exponential expression was found to provide the best fit. . . . The equation that approximated the behavior of medical x-ray film sales in relation to number of salesmen became: sv1 = a . PMlbl - Pi‘bz . sib3 where: 3V1 = Sales volume of medical x-ray film in Districti, a = Constant. PM1 = Product mix being sold in Districti. P1 = Prevailing selling price in Districti. $1 = Number of salesmen in Districti. b1. b2. b3 = Exponents. Final Sales The adjusted demand (orders) is compared with the amount available from inventory and production. If orders exceed the amount available. then sales equal the amount available (no backorders allowed). If orders are equal to or less than the amount available. then sales equal orders. 56 SALES GENERATION DEMAND FROM WEIBUL EQUATION ADJUST DUE TO _PRICE EFFECT ADJUS PROMOTION EFFECT ADJUST DUT TO SALES EFFORT EFFECT ' ’1 NO ‘00 ORDERS EXCEED [" AMOUNT AVAILABLE ASALES = ORDERS GENERATE NOMINAL ] L SALES = AMOUNT AVAILABLE Figure 2-11. Flow Chart: Subroutine SLSGEN Subroutine BOOK The purpose of subroutine BOOK is to calculate profits. collect summary statistics. and adjust model parameters if a product is dropped. Profit Calculation where: PRICE The basic profit calculation for each product is: PROF(K) = REV(K) - COSTS(K) where: PROF = Profit REV = Revenue COSTS : Total Costs K Product Number K. Revenue is calculated thus: REV(K) PRICE(K) * TSAL(K) Unit Price Sales in units TSAL 57 Costs for each product are composed of inventory holding costs. production cost. promotion expenditure. and fixed cost allocation. For simplification. all costs in the firm are assumed to be aggregated into one of these catagories. Inventory holding cost is calculated thus: INHCOS(K) INHCOS(K) where: INHCOS STDIC INV PREINV The production STDIC * (Average Inventory) STDIC * (INV(K) + PREINV(K))/2 Inventory holding cost Standard cost of holding one unit of inventory Current inventory level Inventory level in previous period cost for each product. in each period. is a function of the age of the product. It is assumed that a "learning curve" effect occurs. and. because of learning. the unit cost of producing one additional unit is less than the unit cost of producing the previous unit: The. . .learning curve (concept). . .has evolved from experience in airframe manufacture. which found that the number of manhours spent in building a plane de- clined at a rate over a wide range of production (Hirschmann. 196“). The functional form of the learning curve is assumed to be exponential (Andress. 195k): U(i) where: U(i) U(l) b 0(1) . 1" Cost to produce the ith unit Cost to produce the first unit Learning parameter 58 For each product. the total cumulative cost (of all units produced to date) can be calculated from the approximation (Teichroew. 196h. p. 30). T(n) = U(k)nb+1/(b+1) where: T(n) = Total cost of all units produced to date U(l) = Cost to produce the first unit n = Number of units produced to date b = Learning parameter This formula is used in the model to calculate direct production costs for each time period. The cumulative pro- duction cost is computed and then the cumulative production cost for the previous period is subtracted. Setup cost is then included in cost of production: COSTPR(K) = (CUMCST(K)t- CUMCST(K)t_1) + SETUP where: COSTPR = Total cost of production CUMCST = Cumulative direct product cost SETUP = Setup costs The average unit cost for each product in inventory is then computed:5 AVCOIN(K) = (COSTPR(K) +.AVCOIN(K) * PREINV(K))/ (ACTPRO(K) + PREINV(K)) 5The method of inventory valuation is average cost (Anthony. 1964). The value of each unit in inven- tory is equal to the value of the previous inventory. plus the value of production added to inventory. divided by the total number of units in inventory. 59 where: AVCOIN = Average unit cost of inventory COSTPR = Total cost of production PREINV = Previous inventory level ACTPRO = Actual production Fixed cost allocation is calculated on a percentage of total production basis: (Fixed cost allocation) = FCOS * (ACTPRO(K)/TPROD) where: FCOS = Total fixed costs ACTPRO = Actual production of product K (units) TPROD = Total production (in units) of all products Total cost of sales are then computed: COSTS(K) = (TSAL(K) * AVCOIN (K) + PROM(K) + INHCOS(K) + FCOS(ACTPRO(K)/TPROD) where: COSTS = Total cost of sales TSAL = Sales (units) AVCOIN = .Average unit cost on inventory PROM = Promotion expenditure INHCOS = Inventory holding costs Adjust Parameters and Summary Statistics If a product is scheduled to be withdrawn from the mar- ket. and all inventory of that product has been disposed. a housekeeping operation to adjust the parameters of each prod- uct is performed. Summary statistics are then accumulated for analysis at the end of the run. 60 Subroutine NEWPRO The purpose of subroutine NEWPRO is to provide a mech- anism in the model by which new products may be introduced into the market and obsolete products withdrawn from the mar- ket. Since the major focus of this experiment is not direct- ed towards the product planning process. the decision pro- cesses involved in this routine are simplified and merely fulfill the objective of defining a means of moving products into and out of the market. Just as new products should be introduced to infuse new blood into a product line. so should obsolete products (be- cause of fashion or technology) be withdrawn. The need for selective elimination of products is well recognized in the literature (Alexander. 196h: Kotler. 1965: Levitt. 1965). Profits can be enhanced by eliminating certain costs associated with products in the later stages of their life. . .the proper performance of the (withdrawal) program should result in increased profitability. . . since resources will be assigned to more productive uses . . . (Rothe. 1970. p. 45). Profits typically are reduced or disappear in the de- clining stages of a product's life cycle. At some point. it becomes marginally profitable to transfer resources from dying products to newer endeavors. After studying over 2000 companies. Booz. Allen. and Hamilton (1960) concluded: "Sooner or later every product is preempted by another or else degenerates into profitless price competition." Rothe (1970). in surveying 17“ companies found a number of factors to be important in the product elimination de- cision. Among them were: 61 1. A minimum sales volume 2. A minimum market share percentage 3. Comparison of market share with previous years b. Profitability These factors are not explicitly considered in the model because of the potential of confounding with the primary experimental variables. In the model. the product deletion decision occurs when a product's nominal demand has declined to a point of less than 50% of the past maximum nominal de- mand. "A possible policy with respect to timing may be to let one product almost complete its life cycle before taking on another." (Simon and Freimer. 1970. p. 93) Subsequent to the decision to drop a product. further production scheduled for that product is canceled. but sales continue until remaining inventory is depleted. In the model. when a new product is withdrawn from the market. a new product is introduced to replace the obsolete product. It is assumed that the full research and develop- ment process has transpired. less attractive potential prod- ucts have been discarded. and a steady supply of approved new products is readily available. Each new product has successfully completed the stages of screening. economic analysis. product development. market testing. commerciali- zation. risk analysis. etc. Each new product and the market to receive that product is therefore assumed to be developed and ready for introduction. The degree of acceptance of the product by the market is defined exogenously (and specified 62 by the generation of values for the variables KON and AA). Cross elasticity effects with existing products are assumed to be zero. Although not explicitly handled in this model. the sub- ject of new product decision process has been examined else- where (Freimer and Simon. 1970: Kotler. 1968: Pessemier. 1969) and would appear to be a fertile field for exploration with computer simulation. An experimental model designed to study the factors controlling the product deletion process would also appear to be a potentially fruitful line of investigation. Subroutine EXEC At quarterly intervals. the chief executive officer makes strategic decisions regarding pricing and promotion for each product. In general. pricing and promotion decisions are a response to a comparison of profits with profit goals. The logic in subroutine EXEC is adapted from the Cyert and March (1963). and Kaczka (1966) models. The profit goal is established as a lagged function of past profit. As profits rise or fall. the profit goal follows a similar trend. Thus. profit goals are continually chang- ing. reflecting the past history of actual profits. The firm learns. over time. which strategies are most successful in achieving profit goals. When a profit goal is achieved. greater emphasis is given to the strategy employed. Conversely. when a profit goal is not achieved. less 63 emphasis is given to the strategy employed and a search for a new strategy may be initiated. If the price for a product changed at the beginning of the previous quarter. the chief executive officer reviews the effect of the price change on profit. If profits have improved as a result of the change. the price change rule is altered to further emphasize the direction of the previous price change. Conversely. if profits have not improved as a result of the price change. then the price change rule is revised in order to decrease (or possibly) reverse the direc- tion of the previous price change. Future prospects are then revised. If the projected pro- fit exceeds the profit goal. then the requests for a price cut or promotion increase are evaluated and acted upon. If there are no requests for price or promotion changes. and price has not recently been changed. the price is revised in the best past direction. If projected profits do not meet the profit goal. then price and promotion requests are denied. and a search routine is entered. There are four possible actions which may be taken: 1. Increase sales pressure. 2. Increase sales promotion. 3. Decrease sales promotion. 0. Decrease profit goal. These strategies are employed. each in turn. until a suc- cessful strategy is found. (As a last resort. the profit goal 6a is decreased.) If projected profits still do not meet the pro- fit goal. and price was recently successfully raised (or un- successfully lowered) then price is increased. In summary. subroutine EXEC determines strategy for the future marketing of each product. based on the performance of that product in the past. At the end of each quarter. the performance history of each product is reviewed. and market- ing strategy is revised based on the degree of success of past strategies. Subroutine EXEC also computes a number of quarterly "housekeeping" statistics which are necessary for the opera- tion of the model. EXECUTIVE / DID PROFIT EXCEED COMMITTEE NO \Lg PROFIT GOAL YES ODUGT) [ RAISE PROFIT GOAL] LOWER PROFIT GOAL Se 4L (WAS STRATEGY SRARCR> wAS STRATEGY SEARCH) EMPLOYED? NO' * NO JYES WAS PROFIT GOAL ‘MOVE STRATEGY REDUCED TO BOTTOM OF LIST l DECREASE EMPHASIS ‘INCRRASE EMPR- ON RECENT FAILURL ASIS ON RECENT NO A we 0“— LwAS PROFIT > 9M0. PROFIT AVE.? Figure 2-12. Flow Chart: Subroutine EXEC 65 Figure 2-12 Cont'd. (::) EK§“PRTES\ YRS RAS'PRICE‘ INCREASED INCREASED NO INC‘E‘EASEWPRICR DECREASEPR E CHANGE GRADIENT CHANGE GRADIENT {2V 1 A LCOMPUTE PROJECTED PROFITS YES DO PROJECTED PROFITS MEET PROFIT GOA:)—- DENY PROMOTION i HAVE ALL STRATEGIES‘ & PRICE CUT REQUEST BEEN TRIED . , NO YES SEARCH FOR NEW ill. i.» STRATEGY: [REDUCE PROFIT GOAL 1. INCREASE PRESSURE ON SALES DEPT. ARE PROJ. PROFITS PRO IT GOA 2. INCREASE PROMOTION NO WAS PRICE RAISED SUCCESSFULLY OR VDECREASED UNSUCC. 3. DECREASE PROMOTION I An. —{RAISR PRICE ( IS THERE A REQUEST I \FOR A PRICE CUT IS T A REQUEST FOR A NO PROM. INCREASE YES YES wAS PRICE CHANGED IN LAST 6 MONTHS INCREASE ' N0 PROMOTION ' YES NO < HAS PRICE CHANGED . _ IN LAST 9 MONTHS DENY CUT REQUEST PRIC ' T BEST PAST DIRECTION L SE: A! 66 Utility78ubroutines FUNCTION RNDUM is a service routine for generating a random variate from a normal distribution with a mean of 0.0 and a standard deviation of 1.0 Let r1 and r2 be two uniformly distributed independent random variates defined on the (0.1) interval. then X = {-2 loge r1) 1/2 cos 21rr2 fis aJ'random variate from a standard normal distribu- tion. This method produces exact results . . . . (Mo Millan and Gonzalez. 1968. p. 260; Naylor et a1. 1966. p. 260) The pseudo-random numbers are generated by the standard multiplicative method (Rotenberg. 1960) which has been tested by Clark (unpublished). FUNCTION PRJPRO is a service routine that calculates the projected unit profit for a product. The routine is called from subroutine EXEC as a part of the pricing and promotion decision process. The projected profit is estimated on the basis of price less estimated unit inventory costs. estimated unit produc- tion,cost. estimated fixed cost allocation. and estimation promotion expense. SUBROUTINE RANK is a service routine that provides a mechanism for assigning a rank to any entity based on the value of some designated attribute. The routine assigns the first rank to the entity with the smallest attribute value. and then assigns progressively higher ranks to entities with progressively larger attribute values. 67 The routine ranks on the basis of smallest attribute value assigned rank number one. If it is necessary to rank on the basis of largest attribute value first. then the attribute values must be multiplied by minus one before the ranking routine is called. SW! This chapter has described. in a general way, the con- tent of the model. A more detailed description is provided in.Appendix I. the detailed flow chart, and Appendix II. the computer program. CHAPTER 3 EXPERIMENTAI DESIGN A simulation is basically used to compare the conse- quences of alternate types and levels of independent var- iables. While it would be interesting to examine the sen- sitivity of the model to each parameter value and decision process. the fulfillment of that objective is obviously impossible. While. in principle. simulation can be used to investi- gate the effects of any factors. conditions. procedures. and interactions of which the investigator can con- ceive. in practice this results in factorial experi- ments whose dimensions dwarf the most powerful com- puter and the most lavish budget. so that the experi- mental designs actually used are rather modest (Con- way. at al.. 1959. p. 104). The current experiment. therefore is limited to an investigation of the impact of a selected set of environmental. communication. and personality variables on selected depen- dent variables. The experiment was conducted by making changes in the mOGel and analyzing the effects of the changes on the per- fdrmance of the firm. As in any experiment. it is useful to focus on the terms factor and resEonse. The changes to be made to the model. the independent variables. are classified as factors. or as experimental variables. The performance criteria. the dependent variables. or the output to be generated by the model. are classified as the response 68 69 variable. The primary response variable that was selected for analysis is the profitability of the firm (see subroutine BOOK for details of profit calculation). As noted in the introduction. it is recognized that profit is not the single determinant of a firm's performance and effectiveness: nevertheless. in a free enterprise society. profit remains the primary factor upon which the performance of most business firms is judged. The particular statistic generated by the model is the mean profit per period for each run. Another response variable generated by the model is the mean index of pressure exerted on each product. for each time period. over the length of the run. Also generated is the standard deviation in profit for each run and the standard deviation in the mean index of pressure. These secondary criterion variables are useful in drawing conclusions regard- ing the effect of the experimental factors on the primary response variable. profit. Several techniques of analyzing the experimental data were considered. Cyert and March (1963). for example. used regression analysis to test the sensitivity of parameters in their general model of price and output behavior. Regression analysis is well suited to experiments where the factors are quantitative. rather than qualitative. In this experiment. most factors are qualitative. in that the "levels" represent alternative decision processes rather than different values of the same variable. 70 A factor is quantitative if its levels are numbers which are expected to have a meaningful relation- ship with the responses. Otherwise. a factor is qualitative . . . machines. operators. and days of the week are all considered qualitative factors. . weather conditions in the form of temperature and/or humidity would be a quantitative factor (Naylor. et al.. 1966. p. 32“). Another analysis technique receiving recent attention has been spectral analysis. This technique measures the behavior of the variance of a variable within a time series. . . .as one becomes more sophisticated in the analysis of computer simulation data. he may be interested in analyzing more than expected values and variances. "When one studies a stochastic process. he is interested in the average level of activity. deviations from this level. and how long these deviations last. once they occur." Spectral analysis provides us with this kind of informa- tion. "Spectral analysis studies the salient time properties of a process and presents them in an easily interpretable fashion for descriptive and comparative purposes" (Naylor. et al.. 1966. p. 331; quoting Fishman and Kiviat. 1965). In this experiment. the time series of the response variables is summarized by a single measure. Thus. since the focus of the present research is on a summary variable rather than on a detailed study of the time-dependent behavior patterns. spectral analysis does not appear to be necessary. If the purpose of the current experiment had been to make a quantitative. rather than qualitative. evaluation of the effects of the experimental variables. then further analysis using a multiple ranking (Beckhofer. et al.. 195h) or mul- tiple comparison (Tukey. 1959. or Dunette. 1955) technique would have been useful. In this experiment. however. the information provided by the analysis of variance appears sufficient to make qualitative judgments regarding the effects 71 of the factors. Through analysis of variance. it is possible to deter- mine whether differences between means arose due to chance or if they constitute real effects generated by factor variations. The objective of analysis of variance is to explain the re- lationship between the response variable and the controlled experimental variables. In this technique. the variations in the observations around their grand mean are segregated into those attributable to the main effects. the interaction effects. and unexplained variations. This technique is em- ployed to segregate the total variations in the data into components representing the experimental error. the con- trolled experimental variables and their combined actions. For test of significance. an F-ratio test relating the mean squared residual deviation to the mean squared deviations for each effect is used. In this investigation. a complete 26 factorial experi- mental design was selected in order to explore both main effects and interactions among six factors that were repre- sented at two levels. a "standard" level and an "alternate" level. The first category of factors selected for study in- volves two environmental characteristics (so-called because they are specified exogenously to the firm and thus beyond the firm's control). Factor A. stability of demand. controls the random fluctuations of demand in the model (see sub— routine SLSGEN for details regarding the structuring of 72 random fluctuations). At the standard level. demand is con- sidered to be stable (i.e.. the standard deviation of the NOISE generating distribution is 2.5% of nominal demand; SIGI = 0.025). At the alternate level. demand is considered to be volatile (i.e.. the standard deviation of the NOISE generating distribution is 10% of nominal demand; SIGI 0.10 ) . The selection of the degree of random variability in demand as an experimental variable provides an indirect test of stability in the model. This factor is a test to ascer- tain whether the model behaves reasonably under different conditions in the external environment of the firm. In Effe ct. the model should not be expected to "blow up" and PTOCeed to infinity due to relatively minor changes in the firm '3 external environment. The other environmental characteristic. factor B. market response to salesmen effort. controls the marginal resIbonse of demand to marginal changes in effort by the sale sman (see Subroutine SLSGEN). At the standard level. the response of all products to marginal effort is uniform. (i.e.. the value of effort elasticity is a uniform 1.0). At the alternate level the response of the market is different for each product. and the returns to marginal units of effort "111 be different for each product. (i.e.. the value of ef- fOI‘t elasticity is randomly generated and is not equal to 1'0) . This factor provides a mechanism for contrasting the effect of a "false" information input to the decision process 73 verses a "true" information input. At the standard level. when elasticity of effort is uniform between products. there is no potential differential response to marginal sales effort. {At the alternate level however. elasticity of effort is different for each product and each product displays a unique response characteristic to marginal sales effort. The second category of experimental variables involves the presence or absence of selected communication links. Factor C specifies the presence or absence of knowledge of the current inventory position by the marketing vice presi- dent. At the standard level. knowledge of current inventory is absent. At the alternate level. knowledge of the current inventory level is present and is used by the marketing vice president in setting priorities among different products. Factor D specifies the communication link between sales- men within a given sales territory. At the standard level. no communication between salesmen exists and each salesman must rely only upon his own perceptions regarding the poten- tial response of different products. At the alternate level. each salesman communicates with other salesmen in the same sales region and perceptions about market response are "pooled." The third category of experimental variables is con- cerned with personality characteristics of individuals in the firm. Factor E specifies the personality attribute of the regional sales manager. At the standard level. the sales manager is defined as authoritarian. He neither encourages 7h nor allows participation in the decision process regarding the setting of priorities among different products. At the alternative level. however. the sales manager is defined as equalitarian. He encourages participation in decision making and considers information received from salesmen in setting priorities among products (see Chapter 1 for development of the theoretical framework underlying the selection of this factor). Factor F specifies the personality type of the indivi- dual salesman. At the standard level. the salesman is de- fined as having low need for independence. He derives no satisfaction from an opportunity to influence superior de- cision making. At the alternate level. however. the sales- man is defined as having a high need for independence. He derives intrinsic satisfaction from the opportunity to in- fluence decision making and is highly motivated by the opportunity to participate (also see subroutine SLSMEN description for development of theoretical framework under- lying classification of salesmen personality types). It iS'recognized that personality types in a real firm will be distributed along a continuum. In the model. how- ever. personality types are classified as "polar" for pur- poses of controlled experimental variation. The selected experimental variables and concomitant specification of levels are summarized in Table 3-1. Table 3-1. Experimental Variables Alternative States Factor Standard Alternative A. Environmental Characteristics 1. Stability of Demand Stable Volatile 2. Mkt. Response Uniform Non-Uniform To Sales Effort Elasticity Elasticity 3. Communication Links 1. Inventory No Knowledge Knowledge Position By Sales By Sales 2. Regional Salesmen No Info. Information Intercommunication Exchange Exchange C. Personality 1. Sales Mgr. Authoritarian Equalitarian 2. Salesmen Low Need For Independence High Need For Independence 76 Factorial experiments involve the simultaneous in- vestigation of the effects of a number of different indepen- dent variables. Since, in this experiment. six factors (each varied at two levels) were selected for study. the experiment is described as a 26 factorial experiment. The total number of design points in the full factorial design is the product of the number of levels for each factor. Sixty-four runs were therefore required. each with a unique combination of the six factors. in order to make up one complete replicate of the study. The total experiment con- sisted of six replications. each of which contained sixty- four runs. Within a replication. the initial conditions for each run were idential. while between replications. a different set of starting conditions was employed. (See appendix IV for Program IC; random generation of initial conditions.) The starting conditions are controlled within each replication in order to increase precision and therefore may be regarded as "blocks." In experimental design. blocks are used to reduce the unexplained variation in the observations. [This practicefilsharpens the contrast between [bom- puter runs] by reducing residual variation. Differ- ences can be detected. their statistical signifi- cance tested. and their economic significance assessed with much smaller sample sizes than would otherwise be required (Conway. 1963. p. 53). In order to reduce the effect of any bias that may have been attributable to initial conditions. the first ten per- iods in each run were allowed to elapse before any response variables were measured. 77 Within each replication. the values of the experimental variables for each run were assigned by a zero-one generating algorithm (Gonzalez and McMillan. 1971. p. 161). CHAPTER u DESCRIPTION OF RESULTS This chapter describes the results of the experiment with the model of the firm. Results are reported in terms of the effects of the changes in the experimental variables on the response variables. The complete output of the analysis of variance program is shown in Appendix'r. Table h-i summarizes results of special interest. The results of changes to each factor will be presented separately. starting with the main effects and proceeding through the interaction effects. The description of results will focus on the primary response variable. profit. Table h-i. Summary of Analysis of Variance AVERAGE MEAN PROFIT FIRM PER PERIOD = 63.025 EFFECTS OF EXPERIMENTAL VARIABLES Factor Mean Increment In Profit (From Std. to Alt. Levels) A: Demand Stability -1.20h B: Market Response +3.321 *** C: Inv. Comm. Link -0.099 78 79 Table h-l (cont'd.) Factor Mean Increment in Profit (From Std. to Alt. Levels) D. Salesmen Comm. Link -0.152 E. Participation +5.2h6*** F. N. Independence +1.7juee B-E: Mkt. Response- Participation +2.053** E-F: Part. - N. Ind. +3.260*** **Slgnlficant at .05 Level *** Significant at .01 Level The behavior of the "average" firm is also summarized in Table h-l. The values listed in this table are the overall means of the response variables for all runs and all rep- lications. The values of the response variables for the "average" firm are the standard of comparison upon which the effects of the various changes in the model are evaluated. Conclusions regarding the effect of each experimental variable are de- rived from examination of the value of the "mean invrement" (from the average firm) and the approximate level of signifi- cant of the F statistic. The F-ratio tests for statistical significance were conducted to determine which of the null hypotheses could be accepted or rejected. 80 Results of the firms under different conditions varied considerably from the "average" firm. The object of the analysis was to determine which changes resulted in behavior that was significantly different from the average. The null hypotheses are that none of the changes to the experimental variables produces a change in the response variables. Re- jection of any or all of these null hypotheses implies that the specific factor (or combination of factors) exerts a significant effect to alter the response of the model.1 In terms of the notation. the null hypotheses were: Factor A: A(2):A(1) Factor B: B(2)=B(1) etc. Interaction AB: AB(22) + AB(11) =.AB(21) + AB(12) etc. where: A(2) = Mean of the response variable where factor A is at the alter- native level. A (1) = Mean of the response variable where fac- tor A is at the standard level. and: AB(22) = Mean of the response variable where both factors A and B are at the alternate level. AB(11) = Mean of the response variable where both factors A and B are at the standard level. AB(21) = Mean of the response variable where fac- tor A is at the alternate level and fac- tor B is at the standard level. AB(12) = Mean of the response variable where 1This conclusion is subject. of course. to the probabil- ity of making an incorrect inference. i.e. Type I and Type II errors (see Clarke and Schade. 1969). 81 factor A is at the standard level and factor B is at the alternate level. A numerical measure for each main effect was calculated by the analysis of variance program (A. 0. V.. 1966). An effect is measured as the average difference between the observations at the standard level and the observations at the alternate level. In mathematical terms. the main effect of factor.A is defined as: A = A(2) -A(1) where: .A = Main effect of factor A. In the analysis of variance output. the main effect is expressed as a "mean increment". or. the difference between the mean of the runs with the factor at one level and the overall means of all runs. . Interaction effects are defined mathematically as: AB (ab) 1/2 (AR (22) - AB (21) - AB(12) + AR(11)) where: AB(ab) 'Interaction effect of factors A and B. In the analysis of variance output. the interaction effect is also expressed as a "mean increment". or. the difference between the means with the factors at the same level and the overall mean of all runs. Following are the findings resulting from the analysis of variance. The limitations on the findings of this study. which are a caveat for this research. are explicitly under- lined in a later section of this chapter. 82 Factor A» Stability of Demand Factor.A controls the stability of nominal demand. At the standard level. demand is regarded as relatively stable. i.e. random fluctuations in the demand function are relatively minor. At the alternate level. however. demand is regarded as relatively volatile. i.e. random fluctuations in the de- mand function are substantially higher. The results indicate that profit was slightly reduced at the alternate level: the mean increment in profit when fluctuations in demand increased from the standard to the alternate level was -1.20’+.2 Revenue was also slightly de- creased at the alternate level. with a mean increment of 528. The results therefore indicate that profits and revenue fell slightly in the more volatile market. Main Effect -- Factor B: Market Response Factor B controls the market response to personal sales effort. .At the standard level. the market elasticity of effort is set at 1.0 for each product. At the alternate 'level. the market elasticity of effort is randomly generated and assumes a value between 0.0 and 2.0. This factor is a useful approach to specifying whether information is actually present or absent in the market. If a product has a large elasticity of effort attribute. it responds well to marginal increments of effort by the ZApprgximate significance probability of F statistic = 0.13 . 83 salesman. 0n the other hand. a product with a low elasti- city of effort will have a sluggish response to marginal effort. In general. it is much to the salesman's benefit to direct his marginal effort toward those products with a high elasticity of effort. and away from those products with low elasticity of effort. The salesman. however. has no way of knowing precise elasticity values. He is able to achieve a rough estimate. however. by continually "testing" the market on a trial and error basis. In effect. he is con- tinually undertaking a process of "search" to determine which products respond best to marginal sales effort. He is con- tinually attempting to rank the products in his line on the basis of marginal response to marginal effort. He seeks to find which products will produce the best "payoff" as a re- sult of his sales efforts. The validity of his estimate is of course disturbed by any other elements that may perturb the demand function. In this model. factor B is a means of specifying whether such information is present or not. In the standard case. where the value of elasticity is always 1.0. all of the products respond identically to marginal effort. There is no true basis for differentiating between the products on 'the basis of response even though the salesman may attempt to do so. Any evaluation of response by the salesman in such a case results in a false signal. attributable to random fac- tors alone. since no true differential response characteristic exists. When factor B is at the alternate level. however. BA and elasticity is not equal to 1.0. then true information is considered to be present and a relatively correct signal results. Pastor B. therefore. is a mechanism for specifying market response such that "false" information and a false signal result when factor B is at the standard level. and "true" information is present when factor B is at the alter- nate level. This factor is included for the purpose of evaluating the effect of valid vs. invalid information in the perfor- mance of the firm. The results show that when factor B was at the alternate level. a significant improvement in profits was produced. The mean increment in profit of factor B at the alternate level was an increase of 3.188. The results clearly demonstrate that when unique infor- mation about the market actually did exist. the firm was able to make use of the information for an improvement in profit performance. The particular mechanism by which the firm took advantage of the information is best shown through examination of the interaction effects. Interaction Effect -- BE: Market Response and Participation The interaction of factors B and E at the alternate level produced a significant improvement in profits. Recall that factor E controls participation in decision making. At the standard level. the regional sales manager is defined as authoritarian and no participation is allowed. At the alter- nate level. the regional sales manager is defined as 85 equalitarian and the participation of salesmen in the decision making process is encouraged. When factors B and E interact. the feedback of "true" information from the salesmen to the regional sales manager is accomplished. When valid information is considered by the sales manager in the allocation of effort decision. higher profits resulted. Conversely. if the information is "false". or when participation is not allowed, a lower level of profits resulted. Main Effect -- Factor 0: Inventory Link Factor C controls a communication link between the pro- duction function and the marketing function. .At the standard level. no communication exists and the marketing vice-presi- dent has no knowledge of the inventory position. At the alternate level. the communication link is present. and the marketing vice-president has knowledge of the inventory posi- tion. At the alternate level. if the remaining inventory at the end of a period is not within acceptable limits. the marketing vice-president makes the appropriate change in the index of pressure. which may result in a change in a product's rela- tive priority. The objective of the communication link. therefore. is to provide information regarding the inventory position to the marketing function so that marketing can hope- fully make improved priority decisions. resulting in higher profit. The results. however. showed an effect of factor C on 86 profitability that was somewhat unexpected: when the communication link was "open". profit was unchanged rather than being improved. The mean index of pressure was sig- nificantly increased at the alternate level. A tenative de- duction regarding the cause of this somewhat surprising re- sult is deferred until the discussion in Chapter 5. Main Effect -- Factor D: Salesmen Intercommunication Factor B controls the presence or absence of a com- munication link between salesmen within a region regarding response characteristics of the market. At the standard level. salesmen do not communicate. At the alternate level. salesmen exchange information regarding marginal response of the market to marginal sales effort. The results indicate that factor D had no significant effect on the firm's prof- itability. Main Effect -- Factor E: Participation in Decision Making Factor E controls whether salesmen participate in the decision process of allocating sales effort among the product line. At the standard level. the regional sales manager is defined as authoritarian and he refuses to accept partici- pation in decision making. At the alternate level. the regional sales manager is defined as equalitarian and he pro- vides the opportunity for salesmen to participate. The sales- men participate by transmitting their perceptions regarding the response characteristics of the market. 87 The results clearly demonstrate that factor E. at the alternate level. had a definite positive effect on profits. The mean increment in profit with factor B at the alternate level was 5.301h6. Interaction Effect-~Factors BE: Market Response and Partici- pation The interaction effect of factors B (response function) and E (participation) was significant. When "true" infor- mation was available. and the firm used that information in the effort allocation decision process. the mean increment of profits was an improvement of 2.139h9. Main Effect--Factor F: Need for Independence Factor F controls the definition of the salesmen per- sonality types. At the standard level. the salesman is de- fined as having low need for independence and deriving no intrinsic satisfaction from the opportunity to participate. He possesSes low instrumentality for the outcome of perceived influence in decision making. At the alternate level. the salesman is defined as having high need for independence and deriving satisfaction and motivation from perceived influence in decision making. Profit was significantly improved with salesmen with a high need for independence. The mean increment in profit with factor B at the alternate level was a positive 1.7“. For purposes of drawing conclusions. however. the most rele- vant aspect of factor E was its interaction effect. 88 Interaction Effect--Factors EF: Participation and Need for __m____.15'Volatile (Inventory Communication Link) No-Knowledge =7 Knowledge (Salesmen Communication Link) No-Link =7 Link (Participation) (Participation- Market Response) 000‘ Profit slightly decreased Revenue decreased Promotion decreased No change in profit Revenue decreased No change in profit Profit increased Non-Participation: Non-Valid Info=:> Participation: Valid Info (Participation- Need For Independence) Profit increased Low N.-Ind.: No Participation=©v High N-Ind.: Participation 90 The Problem of Realism in Simulation In discussions regarding the relative value of simula- tion. one dominant advantage is clearly evident: the capabil- ity to attain a high degree of experimental control at rel- atively low cost. For theory testing purposes. this ad- vantage is manifested by the potential ability to system- atically vary and control conditions which. in the real world. may be unwittingly confounded with other factors. The unique advantage of simulation is that not just closely identical but absolutely identical conditions can be maintained through repeated experimental runs. (McMillan and Gonzalez. 1968. p. R92) However. the advantage of control is not attained with- out cost. Concomitant with control is the persistent prob- lem of insuring that the simulation model is an acceptable representation of reality. Although a simulation is able to control potential confounding variables. it is impossible for the simulation to exactly replicate real life.3 Thus the problem that typically confronts any model builder is the question of validity: Is the model a realistic representa- tion of the real world? Cohen and Cyert (1961) have differentiated between models for synthesis and models for analysis. Models for analysis assume knowledge of the overall performance of the system. The objective of this type of model is to discover the manner in which the separate system elements function in 3Nor in most cases really desirable because of the need for parsimony. 91 order to generate the known overall performance. In con- trast. models for synthesis assume knowledge of how the basic elements operate. The objective of this type of simulation is to determine the effect of the elements on the performance of the whole system. The methods for validating these two types of models are different. Validation of models for synthesis takes place principally at the level of the model's basic elements. If essential characteristics of the elements are judged to be true representations of real life and if the re- sults of simulations are plausible. then one tends to consider the model valid. This, in turn. gene- rates greater confidence that the results of simulations can be used as reasonable predictions of what might happen in the analogous real life situation. . . . Models for analysis. on the other hand. are validated in reverse direction.’ Given information on how the total system performs. . . a model is constructed which hypothesizes how the elements of the system operate. If the model is plausible and able to predict. within acceptable limits. the per- formance of the whole system. . . then the model is accepted as a reasonable explanation of how the elements of the system operate. (Saltzman. 1965. pp. h-S.) One basic test for validity would therefore be to com- pare the output of the model with that of a real world organi- zation. This approach is patently impossible. however. in the present research. because the model simulates a hypo- thetical organization. No validity testing procedure has been suggested for surmounting this problem. Since this experiment would be classified as a model for synthesis. however. rather than a model for analysis. a different validation approach is necessary. The focus here _is on validation of the model elements. not on empirical validation. Rather than comparing a simulated time series of 92 output behavior against an actual time series of behavior. validation must rely instead on the judicious selection of the internal components of the model. The validation approach in this experiment will be similar to the technique employed by Bonini (1963. p. 22): The first question to be asked about our model would properly be. "Does the model correspond to the real world?" In other words. "Do the information and decision systems reasonably represent real-world situations?" We would not expect. of course. the model to be an exact replication of the real world--all models are simplifications to some degree. . . . We do be- lieve. however. that the model is a reasonable re- presentation of real-world behavior. We cannot. of course. completely validate this belief. but what we can and will do is set forth the major ingredients of our decision rules for separate examination. We will attempt to justify these rules by relating them to existing theory in the scientific literature of economics. accounting. or the behavioral sciences. or to the literature on business practice. In many cases. we have recourse to "rule-of- thumb" decision rules. . . .validity here essentially depends upon the correspondence of these rules to some aspects of business behavior. . . . Other decision rules rest upon accumulated know— ledge of economics or the behavioral sciences. We shall not attempt to justify these concepts but will merely relate our constructs to the appropriate literature. Validation therefore depends on the care that is exer- cised to clearly indicate the structural variables and re- lationships in the model. and most importantly. their theoret- ical and empirical foundations. It should be clearly recognized that this approach to validation does not categorically insure that the model is isomorphic with reality. This test. in fact. may be inter- ‘preted as a null test. If a model failed. it would be sus- pect: on the other hand. one that has valid elements and 93 produces reasonable output cannot be categorically guaranteed as representing reality. In view of the limited objective of this experiment. and because the organization is hypo- thetical. an effort to justify the elements of the model appears to be the only reasonable method of attempting to incorporate realism. The value of this experiment. therefore. will lie in its ability to contribute to the understanding of the structure and internal consistency of behavioral theory. rather than its ability to reproduce and predict the actual behavior of a real firm. In this approach to validation the relation between Simulation methods and survey methods becomes extremely im- portant: . . .the computer model compliments field research and experimental techniques. it does not replace them. The working hypotheses contained in the model must still be developed and verified empirically and deductively. and the more familiar research techniques will continue to be important here. (Meinhart. 1966. p. 30h.) The limitations of this experiment should be considered when the findings. implications. and conclusions are examined. While the model is complex. certainly not all aspects of a business firm are considered. The emphasis has been on the marketing sector. and only on selected decisions within the marketing function. Behavioral characteristics of individ- uals have been defined to only a limited extent. The sensi- tivity of the model to many parameters has not been tested and interaction effects with variables that have not been included in the experiment could possibly be significant. 9h It is believed. however. that despite the limitation cited above. computer simulation can lead to further insight regarding behavioral theory and may also provide additional insight into the world of reality. CHAPTER 5 _S_U_MMARY AND CONCLUSIONS This chapter attempts to derive conclusions and impli- cations regarding the effects of the experimental variables on the profitability of the firm. Recall from Chapter 3 that the experimental variables were separated into three cate- gories: environmental factors. communication factors. and personality factors. The following discussion will analyze the reasons why changes in the experimental variables pro- duced the results described in Chapter A. and will attempt to relate the findings to the empirical and theoretical concepts upon which the model was founded. Effects of Demand Stability When factor A was at the alternate level. indicating a volatile demand function. overall revenue and profits were slightly reduced. This result was somewhat in conflict with Bonini. who found that a volatile demand function led to an increased level of profits. He explained his result as follows: . . .1n a highly variable environment an occasional crisis is caused within the firm by chance alone (that is. by factors outside the control of the firm). Such crises cause the firm to initiate search for better alternatives. generally ending in finding higher sales or lower costs. (Bonini. 1963. p. 136) In this model. however. an effect was present which tended to counteract the mechanism deduced by Bonini. 95 96 Profits in this model were sensitive to the capability of the salesmen to estimate the marginal response of the market to marginal changes in selling effort. Any factor which tended ‘to diminish the validity of the salesmen's estimate of marginal response therefore also tended to reduce the firm's profitability. If the salesman's perception of marginal response was impaired. then the company's ability to make a better sales allocation decision was also impaired. and decreased profits resulted. Factor A. at the alternate level. represents a stronger influence of demand factors exogenous to the firm's sphere of control. The stronger influence of such extraneous fac- tors tended to interfere with each salesman's capability of estimating marginal response. and profits were accordingly reduced. The strength of this effect completely overshadows the contrary effect found by Bonini. No direct evidence is present that the effect found by Bonini was also found in this model. One result. however. leads indirectly to the conclusion that the "Bonini effect" was indeed present. In a limited post experimental run. the average level of promotion was found to be higher when fac- tor A.was at the alternate level (see Table 5-1). indirectly indicating that "search" was more frequently implemented. .Although promotion was higher at the alternate level. revenue and profits were lower. indicating that the interference to the process of estimating marginal response was the pre- 'dominate effect. 97 Table 5-1. Variation of Promotion Expenditures with Factor A Level of Factor A, Mean Promotion Expenditure 1 63u5 2 6361* *significant at 0.15 level In summary. the results present indirect. albeit incon- clusive. evidence that the effect found by Bonini may be pre- sent. but strong evidence of the effect of the interference of a volatile market with the process of predicting marginal response to marginal sales effort. Effect of Inventory Communication Link Since the results of factor C. the inventory communica- tion link. were counter to prior expectations. a post- experiment investigation was conducted in an attempt to dis- cover the structural mechanism that would explain why profits were not improved when the inventory communication link was open. Recall from Chapter 2 (Subroutine INVENT). that upper and lower limits on inventory levels are established as a lagged function of total sales. In the event these limits are "breached"1 by unexpected demand for a particular product. an adaptive mechanism is initiated. First. the pressure on 1In effect. a "breech" means the end-of-month inventory is outside the previously established upper or lower limit. 98 that product is adjusted (if factor C is at the alternate level). Second. the limits on inventory levels are also adjusted as a self-correcting mechanism. The scheduled level of inventory is also adjusted. A subsequent trial run was made for the purpose of dis- covering if subroutine INVEHT behaved differently at the different levels of factor C. One of the first observations was concerned with the timing of the "breeches" of the inventory limits. Preliminary indications in the trial investigation demonstrated that the breeches of the lower limit typically occurred at the beginning and the growth stages of the prod- uct's life cycle. while breeches of the upper limit typi- cally occurred at the maturity and declining phases of the life cycle. It was also noted that the number of breeches of the lower limit were typically more numerous when factor C was at the standard level than when factor C was at the alternate level. This behavior led to the tentative explana- tion for higher profits at the standard level as shown in Figure 5-1. With factor C at the standard level. pressure is not lowered in the event of a lower limit breech. thus. sales tend to be higher. which. in turn causes more lower limit breeches. Conversely. when Factor C is at the alternate level. pressure is lowered in the event of a lower limit breech. which tends to limit sales because of a lower index ‘of pressure. In general. a higher level of sales of a product 99 tends to increase profits because of a decreasing ratio of fixed cost to total sales and also because of a slight "learning effect". which tends to slightly decrease unit variable costs as cumulative sales volume rises. The question should be considered: If lower limit breeches cause increased sales at the growing stages of the product life cycle. should not the upper limit breeches at the declining stage of the life cycle cause a counter and offsetting effect? In the maturity and declining stages. the pattern is for upper limit breeches to occur. which would cause an increase in sales when C was at the alternate level. This phenomenon does. in fact. occur. With C at the standard level. sales tend to be lower in the declining stages. The loss in sales. however. is not sufficient to overcome the prior increase in sales that was experienced in the growth stage. The net effect is that. over the total life of the product. sales are higher when C is at the stan- dard level. causing a concomitant increase in profits. 100 c=1 c=2 (wo- INK) (LINK) LOWER LIMIT BREECH PRESSURE LOWERFD LOWER LIMIT BREECH PRE S SURE UN C HANGED WHICH CAUSES MORE LOWER LIMIT RREECHES LIMITS RAISED MORE SCHED. INVENTORY RAISED MORE \ SCHED. INVENTORY RAISED INVENTORY MORE ABLE TO HANDLE SALES HIGHER PROFITS Figure 5-1. Effect of Inventory Communication Link As shown in Table 5-2. REV-1 is the mean revenue for the first 10 periods of a product's life. Note that revenue is higher when C=1. Higher revenue (at C=1) continues through the growth stages of the product's life. REV-2 is the mean revenue for a product's declining stage only. Note that in the declining stage. when C=1. the mean revenue is lower. TOTAL REVENUE. the mean revenue over the total product life. is higher at the standard level. These statistics can be interpreted as follows: When C=1. revenues tend to be higher in the growth stage. but lower in the declining stage. When C=2. revenues tend to be lower in the growth stage, but higher in the declining stage. The net total difference in revenues between the standard and 101 alternate levels. however. is such that total revenues are higher at the standard level. The conclusion. therefore. is that the differences in revenues at the growth stage is the predominant effect. and has sufficient influence on profits. In summary. two counter-balancing effects2 occur as a result of factor C. On the one hand. the inventory communi- cation link does cause lower inventory costs. On the other hand. the communication link has an inhibiting effect (be- cause of decreased pressure) on sales during the crucial growth stages of the product's life cycle. The unexpected net result of these counter-current effects is that factor C has no significant effects on the profitability of the firm. Effect of~§alesmen Intercommunication Somewhat surprisingly. factor D. the communication link between salesmen within a region. failed to have a significant effect on profits. It was expected that the presence of the communication link would significantly improve the quality of information. A reexamination of the communication link mechanism as conceptualized in the model is necessary in order to explain this result. 2It is worth noting that Bonini also found counter- balancing effects: With a "loose" Industrial Engineer- ing Dept.. profits were expected to be higher because of higher costs. Costs were. in fact. found to be higher. but sales rose to counterbalance the increase in costs. 102 Table 5-2. Effects of Standard and Alternate Levels of Factor C REV-1 REV-2 TOTAL INV. REV. HOLDING COST 1 (Standard) 14069 71975 116h17 ##31 C: 2 (Alternate) 13688 73118 115955 ##06 Mean Increment -190 571 -230 -12 Each salesman undertakes a "search" process for the purpose of evaluating how each product will respond to marginal effort. The result of his "search" is a ranking of the products on the basis of high marginal response to marginal effort. His search and ranking. of course. are subject to error from other extraneous factors that might affect demand. It was expected that communication with other salesmen would tend to reduce this error. but the re- sults generally did not support this expectation. It is proposed that the reason the communication link did not reduce error is that the difference between the sales- men's rankings was negligible in the first place. Therefore. communication regarding the relative rankings generally served no useful purpose. In general. the communication was an attempt to "pool" information when. in fact. there really »was no substantive difference between the information possessed 103 by each salesman. There was. however. an exception. In the cases where factors A and B were at the alternate level. profit was slightly improved when the communication link was present (see Table 5-3). Table 5-3. Profit Under Different Conditions of Factor D FACTOR A B D PROFIT 2 2 1 59.086 2 2 2 59.351 This result. although not statistically significant. was in keeping with expectation. With factor A at the alternate level. the market was subject to a higher degree of random changes in demand. With factor B at the alternate level. the information to be transmitted was truly valid.3 Therefore. the pooling of information between the salesmen appeared to result in a reduction of random error. improving the quality of information. The differences between the salesmen's error. however. were not sufficiently large to produce a statistically significant improvement in profits. It should be noted that. for reasons of simplicity. the motivational effects of group intercommunication are not 3See previous discussion under factor B. 10h explicitly modeled in this experiment. Likert (1967) pro- poses that face-to-face group communication. aside from the informational transfer effects. also results in favorable attitudes and higher motivation to produce. It would-seem reasonable to assume that the effects of group intercommuni- cation. like participation in decision-making. can be segre- gated into an information transfer component and a socio- psychological (ego-involvement) component. but not the socio- psychological component. was explicitly considered. The failure of the factor to produce a significant improvement in performance. therefore. may be explained by the absence of motivational effects in this model. and also by the fact that an improvement in the transfer of informa- tion is not possible under the particular conditions of information structure proscribed by this model. Effects of Participation The effect of participation. factor F. cannot be ade- quately discussed except in conjunction with factor 8. mar- ket response. and factor F. need for independence. The interaction effects of these variables provide an excellent means of demonstrating the partialling of the effects of participation into a result due to "decision quality" and a result due to "ego-involvement." Ihen acting in conjunction with factor B. factor F pro- duces an improvement in profits due to "decision quality." .because of the improvement in the quality of information that 105 is utilized as an input to the effort allocation decision. When factor B is at the alternate level. the elasticity of effort varies between products. The potential exists for an improvement in performance if more effort is directed toward those products with a high elasticity of effort. and it is the responsibility of the salesman to "search" for those products which will give a high marginal response to marginal effort. The results clearly show that when valid information actually does exist. and when the regional sales manager does consider the information in making the alloca- tion decision. profits significantly improve because the sales manager is in possession of information needed to make a higher "quality" decision. With the interaction of factors 3 and E. the improvement in decisionemaking is due entirely to an improvement in the quality of information that is re- quired for the decision. The interaction of factors E and F. however. deals with the motivational aspects of the process of participa- tion. In the model. a salesman with low need for independ- ence attaches no instrumentality to the opportunity to participate. A salesman with high need for independence. however. attaches high instrumentality to the opportunity to influence decision making. In the model the results clearly demonstrate that salesmen with high need for independence, when allowed to participate in decision making. are motivated to higher levels of effort which significantly improve the ‘profit performance of the firm. The salesman becomes "ego 106 involved" in the decision. and exerts significant supportive effort to assure that the decision is correct. The results also show that this effect may occur even though no concomi- tant improvement in information and decision quality is pres- ent. The value of the conclusions derived from these results lies in the fact that the model clearly demonstrates that. from a structural standpoint. participation in decision making can be partialled into a decision quality effect and an ego involvement effect. (Fig. 5-2) With the decision quality effect. the improvement in performance occurs be- cause the decision-maker has improved information upon which he can base his decision. With the ego-involvement effect. the improvement comes about because the participant becomes involved in the decision making process. and strives to in- sure desired results from his recommendation. This simula- tion has modeled structural mechanism which segregates the two effects into independent components: the decision qual- ity effect can exist independently of the ego-involvement effect. and vice versa. In other words. the two effects can act concurrently to improve performance. In summary. the experiment has demonstrated structural mechanisms by which participation in decision making can act independently to improve performance. The conclusions derived from the experiment serve to demonstrate the consistency and sufficiency of Vroom's conceptualization of the effects of 'participation and lead to further confidence in his theory. 107 INFORMATION EFFECT (DECISION QUALITY) PARTICIPATION IN DECISION- MAKING A MOTIVATIONAL EFFECT (EGO- INVOLVEMENT) Figure 5-2. The Effects of Participation Future Research The model suggests a number of directions in which future research projects could be developed. In the marketing area. the model could serve as a framework upon which experi- ments might be constructed that focus on the product intro- duction process. the product deletion process. or the behav- ioral process of allocation of sales effort. An experiment that developed the application of smoothing-type forecasting techniques (especially adaptive smoothing) to life-cycle type products might also be appropriate. Another area of future interest might deal with a more formal examination and delineation of the counterbalancing effects found in this experiment. Without a doubt. this experiment has been distinctly limited in scope--dealing with only a few of the myriad of variables that might form the basis of interest for such an experimental approach. The work of Joan Woodward (1965), for example, regarding the juxtaposition of technical and 108 behavioral aspects of the work environment might provide useful focus for a computer simulation experiment. Alternate ways of developing attributes of individuals might be con- sidered. especially various aspects of need-achievement. need- affiliation. and need-power. Finally. the original objective of the research was to provide a step toward the melding of the "macro" and "micro" simulation approaches into one model. This model has served to provide a framework (i.e.. the "macro" portion) for such an integration. but requires considerable in-depth develop- ment of the micro decision processes if such an objective is to be completely accomplished. The challenge of the micro-' process development--perhaps the integration of a HONUNCULUS type of decision system--will hopefully provide stimulation for fruitful future research. Summary This chapter has described the major mechanism which produced the experimental results. Two factors. stability of demand and the inventory communication link. were found to have little effect on profitability. primarily because of counterbalancing mechanisms that tended to result in opposing effects. The intercommunication of salesmen was also found to be insignificant because of the lack of substan- tive information transfer. The strongest effect on profit- ability was caused by participation in the sales-effort . allocation decision. especially when considered in conjunc- tion with valid market information and salesmen characterized 109 by a high need for independence. The model has demonstrated a structure which explicated the partialling of the effects of participation into an information transfer component and an ego involvement com- ponent, each independent of the other. The experiment has provided an initial step toward the integration of "macro" and "micro" type simulations. as well as demonstrated the viability of simulation as a means of explicating the structural mechanism under lying behavioral theory. .Above all. the experiment has provided a demonstra- tion of the usefulness of computer simulation as a vehicle for the exploration of behavioral theory. It is believed that the conclusions drawn from the experiment. when tempered by the limitations cited in Chapter h. are useful in demon- strating the tractability of the dynamic modeling of behavior in organizations. BIBLIOGRAPHY BIBLIOGRAPHY Ableson, Robert P. and Bernstein. "A Computer Simulation Model of Community Referendum Controversies," Public Opinion Quarterly, Vol. 27, No. 1, (1963), 93-122. Alexander, R.S. "The Death and Burial of Sick Products," Journal of Marketin , (April, 1964), 1-7. Amstutz, A.F. Computer Simulation of Competitive Market Response. Cambridge, Mass.: M.I.T. Press, 1967. Andress, Frank J. "The Learning Curve as a Production Tool," Harvard Business Review, (January-February, 1954). Anthony, R.N. Management Accounting. Homewood, 111.: Richard D. Irwin, Inc., 1964. ------- . 59V: Stat Description No. 14. East Lansing, Mich.: Michigan State University, Agricultural Experiment Station, 1966. Barton, Richard E. A.Primer on Simulation and Gamigg. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1970. Bechhofer, R.E., Dunnett, C.W., and Sobel, M. "A Two-Sample Multiple Decision Procedure for Ranking Means of Normal Populations with Common Unknown Variance," Biometrika, (1954), 170-176. Bellman, James Ross. "Behavioral Simulation Models in Marketing Systems," Unpublished Ph.D. dissertation, Yale University, 1969. Benson, Verel Willard. "Interregional Competition in the Broiler Industry," Unpublished Ph.D. dissertation, University of Mary- land, 1969. Bettman, J.R. "Behavioral Simulation Models in Marketing Systems," Unpublished Ph.D. dissertation, Yale University, 1969. Bonini, Charles. Simulation of Information and Decision Systems in the Firm. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1963. Booze, Allen, Hamilton, Management Consultants. Management of New Pro- ducts. 3rd. Ed., New York: John Wiley and Sons, Inc., 1960. 110 111 Brightman, H.J. "Individual Behavior and the Small Work Group; A Simu- lation Study," Unpublished Ph.D. dissertation, university of Massachusetts, 1970. Brockhoff, Klaus. "A Test for the Product Life Cycle," Econometrica, 35, (July-October, 1967), 472-484. Brown, Robert Goodall. Smoothing, Forecasting, and Prediction of Dis- crete Time Series. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1963. Campbell, Donald T. and Stanley, Julian C. Experimental and ngsi- Egperimental Designs for Research. Chicago: Rand MCNally and Co., 1963. Carroll, Tom W. SINDI 2: Simulation of Innovation Diffusion in;e;Rural Community of Brazil. East Lansing, Mich.: Computer Institute for Social Science Research, 1969. Chervany, N.L. "A Simulation of Analysis of Cash Flow Patterns Within A.Manufacturing Organization," Unpublished D.B.A. dissertation, Indiana University, 1968. Clark, Charles T. and Schkade, Lawrence L. Statistical Methods for Business Decisions. Cincinnati, Ohio: South-Western Publishing, 1969. Clark, N., "Testing of Pseudo Random Number Generators," AMD Tech. Memo No. 89, Applied Mathematics Division, Argonne National Laboratory, Argonne, Illinois. (unpublished). Clarkson, G.P.E. Portfolio Selection: A Simulation of Investment Trust. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1962. Cleveland, Fredrick Willis, Jr. "Resource Allocation in the Firm: The Study of Tangible and Intangible Investments by Computer Simu— 1ation of Management Heuristics," Unpublished Ph.D. disserta- tion, Columbia University, 1967. Cohen, K.J. and Cyert, R.M. "Computer Models in Dynamic Economics," The Quarterly Journal of Economics, Vol. 75, (February, 1961), Conway, R.W. "Some Tactical Problems in Digital Simulation," Management Science, (October, 1963), 47-61. Conway, R.W., Johnson, B.M., and Maxwell, W.L. "Some Problems of Digi- tal Systems Simulation," Management Science, (October, 1959), Cox, William E., Jr. "Product Life Cycles as Marketing Mbdels," The Journgl:of Business, (October, 1967), 375-384. 112 Crow, Wayman J. The Role of Simulation-Model Construction in Social Research on Post-Nuclear Attack Events, La Jolla, California: Western Behavioral Sciences Institute, 1967. Cyert, Richard M. and March, James G. A Behavioral Theory of the Firm. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1963. Desjardins, Robert B. "An Investigation of Decision Systems in a Com- plex Management Simulation," Unpublished Ph.D. dissertation, University of North Carolina, 1964. Dunnett, C.W. "A Multiple Comparison Procedure for Comparing Several Treatments with a Control," Journal of the American Statistical ‘égsociation, (1955), 1096-1121. Feigenbaum, Edward A. "Computer Simulation of Human Behavior," Midwest Human Factors Society Symposium, Milwaukee, Wisconsin, (August, 1963). Fishman, George S. and Kiviat, Philip J. "Spectral Analysis of Time Series Generated by Simulation Models," The RAND Corporation, RMFA393-PR, (February, 1965). Fondren, William M., Jr. "An Industrial Dynamics Study of the Effect of Market and Cash Flow Forecasting on the Growth of a Small Tech- nical Enterprise," unpublished M.S. thesis, Massachusetts Institute of Technology, 1963. Freimer, Marshall and Simon, Lonard S. "The Evaluation of New Product Alternatives," in.épplications of Management Science in Market- igg, David Montgomery and Glen Urban (eds.), Englewood Cliffs, N.J.: Prentice-Hall, 1970. Frijda, Nico H. "Problems of Computer Simulation," Behavioral Science, (January, 1967), 59-67. Gensch, D. "A Computer Simulation Model for Media Selection," Unpub- lished Ph.D. dissertation, Northwestern university, 1967. Gonzalez, Richard, R. and McMillan, Claude. Machine Computation: An Algorithmic Approach. Homewood, I11.: Richard D. Irwin, Inc., 1971. Green, Jay R. "A Computer Simulation of a Marketing Organization," Unpublished Ph.D. dissertation, Ohio State University, 1960. Guetzkow, Harold (Ed.) Simulation in the Social Sciences. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1962. Guetzkow, Alger, Brody, Noel, and Snyder. Simulation in International Relations. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1963. 113 Gullahorn, John and Jeanne. "The Computer as a Tool for Theory Develop- ment," Santa Monica, California: System Development Corp., SP-817, (June 5, 1962). Gullahorn, John and Jeanne. "Computer Simulation of Interaction in Small Groups," American Fed. Information Processigg Societies Conference Proceedings, (1964), 103-173. Gullahorn, John and Jeanne. Social and Cultural System Simulations. East Lansing, Mich.: Computer Institute for Social Science Research, 1969. Gullahorn, John. "Testing Organization Theories by Computer Simulation,‘ East Lansing, Mich.: Unpublished application for research grant, 1970. Hershauer, James Cliff. "An Empirically-Derived Simulation to Explore Sequencing Decisions," Unpublished D.B.A. dissertation, Indiana University, 1969. Hirschmann, Winfred H. "Profit From the Learning Curve," Harvard Busi- ness Review, (January-February, 1964). Homans, George Casper. Social Behavior: Its Elementary Forms. New York: Harcourt, Brace, and World, Inc., 1961. Hunt, Arch William. "Statistical Analysis and Verification of Digital Simulation Models Through Spectral Analysis," Unpublished Ph.D. dissertation, University of Texas at Austin, 1970. Hutchinson, George K. "The Design and Simulation of a Management In- formation and Control System," Unpublished Ph.D. dissertation, Stanford University, 1964. Kaczka, Eugene Edwin. "The Impact of Some Dimensions of Management Climate on the Performance of Industrial Organizations," Un- published Ph.D. dissertation, Rensselaer Polytechnic Institute, 1966. Kotler, Philip. "Phasing Out Weak Products," Harvard Business Review, (March-April, 1965), 107-118. Kotler, Philip. "Competitive Strategies for New Product Marketing Over the Life Cycle," Management Science, XII, No. 4 (December, 1965), B-104-ll9. Kotler, Philip. Marketing Management. Englewood Cliffs, N.J.: Pren- tice-Hall, Inc., 1967. Kotler, Philip. "Computer Simulation in the Analysis of New-Product Decisions," in Applications of the Sciences in Marketing Manage- ment. Frank Bass, et. al. (eds.), New York: John Wiley and Sons, Inc., 1968. 114 Lambert, Farrel V. and Kniffin, Fred W. "Response Functions and Their Application in Sales Force Management," Southern Journal of Busi— ness, (January, 1970), l-ll. Levitt, Theodore. "Exploit the Product Life Cycle," Harvard Business Review, (November-December, 1965), 81-94. Likert, Rensis. The Human Organization: Its Management and Value. New York: McGraw-Hill Book Co., 1967. March, James G. and Simon, Herbert A. Organizations. New York: John Wiley and Sons, Inc., 1958. McMillan, Claude and Gonzalez, Richard F. Systems Analysis: A Computer Approach to Decision Models. (revised edition) Homewood, I11.: Richard D. Irwin, Inc., 1968. McPhee, W.N. "Note on a Campaign Simulator," Public Opinion Quarterly, Vol. 25, (July, 1961), 184-193. Meinhart, Wayne A. "Artificial Intelligence, Computer Simulation of Human Cognitive and Social Processes, and Management Thought." Academy of Management Journal, (December, 1966), 294-307. Miller, Richard F. "Product Growth in Competitive Markets," Unpublished SM.Thesis, School of Industrial Management, Massachusetts Insti- tute of Technology, 1962. Naylor, T.H., Balintfy, J.L., Burdick, D.S., and Chu, K. Computer Simulation Techniqpes. New York: John Wiley and Sons, Inc., 1966. Norek, Bernard Jean-Marie, "Simulation Models of New Product Introduc- tion and Market Evolution," unpublished Ph.D. dissertation, University of Pennsylvania, 1970. Parker, Alan J. "Optimizing a Complex System: A Simulation Study of a Commodity Distribution System," Unpublished Ph.D. dissertation, Columbia University, 1969. Pessemier, Edgar A. "Models for New-Product Decisions," Paper No. 247, Institute for Research in the Behavioral, Economic,and Management Sciences, Lafayette, Indiana: Krannert Graduate School of In- dustrial Administration, Purdue University, 1968. Polli, Rolando and Cook, Victor. "Validity of the Product Life Cycle," The Journal of Business, (October, 1969), 385-400. Raser, John R. Simulation and Society, Boston, Mass.: Allyn and Bacon, Inc., 1969. Rogers, Everett. The Diffusion of Innovations. Glenco, 111.: Free Press, 1962. 115 Rotenberg, A. "A New Pseudo-Random Number Generator," American Association of Computing Machines, Vol. 7, (1960), 75-77. Rothe, James T. "The Product Elimination Decision," MSU Business Topics, (Autumn, 1970), 45-52. Saltzman, S. "Survey of Principle Types of Simulation Models of the Firm," unpublished paper, Cornell University, Computing Center and Department of Industrial Engineering, (June, 1965). Schelling, Thomas C. "Experimental Games and Bargaining Theory," World Politics, Vol. 14, No. 1, (1961), 57. Siemens, Nicolai. "Analysis of Selected Industrial Decision Rules In a Dynamic Environment by Means of Computer Simulation," Unpublished Ph.D. dissertation, University of Oregon, 1967. Simon, Leonard S., and Freimer, Marshall. Analytical Marketing. New York: Harcourt, Brace and World, Inc., 1970. Smith, Robert D. "The Heuristic Simulation of Psychological Decision Processes," Journal of Applied Paychology, Vol. 52, No. 4, (August, 1968), 325-330. Srinivas, K.M. "A Computer Simulation Model of Newcomb's Consistency Theory: A Case in Theory Development," Unpublished Ph.D. dis- sertation, University of California, Los Angeles, 1970. .Stallings, Billy Gene. "A Simulation Model for Budgeting and Cost Con- trol of a Medium-size Hospital Intensive Care Unit," Unpublished Ph.D. dissertation, Mississippi State University, 1970. Summit, Roger K. "Simulation of a Psychological Decision Process Uti- lizing a Computer Model of the Aerospace Industry," Unpublished Ph.D. dissertation, Stanford University, 1965. Tarter, Jim 1. "A Study of Computer Simulation," unpublished paper, Sociology Dept., Michigan State University, 1971. Teichroew, Daniel. An Introduction To Management Science: Determinis- tic Models. New York: John Wiley and Sons, Inc., 1964. Tomkins, Silvan S. and Messick, Samuel (Eds.). Comppter Simulation of Personality. New York: John Wiley and Sons, Inc., 1963. Tuason, R.V. "Experimental Simulation on a Pre-Determined Market Mflx Strategy," Unpublished Ph.D. dissertation, Northwestern Univer- sity, 1965. Tukey, J.W. "The Problem of Multiple Comparisons," unpublished manu- script, Princeton University, 1953. Also in Scheffe', H. The Analysis of Variance. New York: John Wiley and Sons, Inc., 1959. 116 Uhr, Ernest Bernhard. "Management of the Marketing Mix for a Retail Food Chain: A Simulation Study," Unpublished Ph.D. dissertation, Rensselaer Polytechnic Institute, 1969. Urban, Glen L. "A Mathematical Modeling Approach to Product Line Deci- sions," Journal of Marketing Research, (February, 1969), 40-47. Verba, Sidney. "Simulation, Reality, and Theory in International Re- lations," World Politics, XVI, No. 3, (1964). Vroom, Victor H. Work and Motivation. New York: John Wiley and Sons, Inc., 1964. Idallace, Fraser G. "A Computer Simulation of Selected Hypotheses Describing Behavior of a Firm within a Given Economic Environ- ment," Unpublished Ph.D. dissertation, University of California at Los Angeles, 1961. Weber, W.L. "Toward an Integrated Model For Manpower Planning," Un- published Ph.D. dissertation, Carnegie-Mellon University, 1970. ----- , Webster's Seventh New Collagiate Dictionagy. Springfield, Mass.: G. & C. Merriam Co., 1967. Weibull, W. "A Statistical Distribution Function of Wide Applicability," Journal of Applied Mechanics, (September, 1951), 293-297. Wilson, G.T. "Behavioral Theory of Production Planning in the Firm." Unpublished Ph.D. dissertation, Graduate School of Industrial Administration, Carnegie-Mellon University, 1969. Woodward, Joan. Industrial Organization: Theory and Practice. London: Oxford University Press, 1965. APPENDIX I Detailed Flow Chart 117 PROGRAM PAJAMA .___._)@B 2020 x - 1,10) mm ZOOI, ("m“), an - mg] 2020 cmmw: [new 2003, map, steal CALI. unscflsuo) < 00 3003 a - 1.30) I [new 300:, 1204,“), xx - 0,8] 1 Ilwm - 6 mums .- mm: I ALGGITM FOR CENERATIM ZERO-ONE VALUES 0' EXPERIMENTAL VARIABLES Lsncm mum. VALUES no manners: wants] DO IOOO KLOCK II I,LENGTN CALL SLSIEN LOOP FOR EXISTING PROMICTS ASSICN ATTRIIITES OF EXISTINC PROOL‘TS EID LNP REAO REPLICATION NIHBER AID RAIDOI MMBER SEED SET INITIAL RAID!” NIHIER LOOP FOR FUTIRE PRODUCTS READ ATTRIIUTES OF FUTIRE PRWCTS SPEC IFY ”OER OF EKPER INENTAL VARIABLES AID NUNIER OF RUNS MAIN TINEREEPIM LOOP Fm EACH RIII CALL FORECAST ROUTINE CALL INVENTMV ROUTINE CALL PROOIIT ION ROUTINE CALL PRICE 5 PROHOT ION RECWEIOATION ROUTINE CALL SALES EFFMT OECISIOI RWTINE \ PROGRAM P AMA (CONT.) CALL SLSGEN” I! [ CALL BOOKJ ITIQ I ITIQ + I r [us ITIQ.LT.—3\. vts ND L—)LL:ALL newafl CALL EXEC —w J h—EDOOiCONTIqu" [CALL summfij Es REPLICAT non :mPLET£\ YES NO [BEG m ANOTHER Run] E3 118 CALL SALES GENERATING ROUTINE CALL BOOKKEEPING ROUTINE ADVAICE TIME- IN-QIARTER COUNTER DOES QUARTER PAVE TIME LEFT CALL NEW-PRODUCT RWTINE CALL EXECUTIVE DECISION RNTINE no MAIN TIMEKEEPIKI LOOP CALL 00er ROUTINE 119 SIBRWTINE FCAST I TSLFOR - 0. I - 1:30 out TOTAL SALES nuns: I! ______)(00 M9 1: - um» - LOOP son chm manners Lproncm - trmcmj - mu nevuous Foams: [ rmc . 0 I - zno our roams: rm uncut: x A 00 “.8 L - I,NSALS - LOOP to: cum: SALES!“ as": - SAVEIK,L): sAVEmL) - Am. - rust“! SALES 000E050: L) 9 “autumn“ mums: - SAVE(K,L - PSAVE -. POELOR(K,L) - new: (ML): 0£Loa(x,L) - Amew-ucc o (I-ALPI) IDELGHIQL): FORCIK,L) - SAVE(K,L) o (ALPI.0£Lm(x,L)/(I - ALPI) I 3L [noncm - woacm o mango] . Accunuuu roams: Iron meow: I: j, M8 | counuufl - no sALtsctn L00? .2. | TSLFm . reron . "new | . ACCUMULATE TOTAL SALES roams: I M9 comm: - no newt: LOOP aim 120 SIMOUTINE INVEN'T —X00 I99 I: - human) _Jr me(x) - ALP7tVINL(R) + (I-ALP7)A TSAL(K)AO.3 amux) - ALPhRINHR) O (I-ALPHATSALIKhOJ [Is INV(K).LE.VINL(R)\ NO YES SCINLIK) - mvm + 0.5t(VINL(K) - mvm ] . [Is INVFAC.EQ.O\ no yes nesm - ncsm o our nesm - mu(n£s(x).l.2 1, mem - .sswmux) nmux) . .ssanmLm /IS M(x).0£.AmL(T)‘€——-—- O VES ,scmLIK) - INV(K) . 0.5::(nmux) - mm )] ‘ vuuth) - 1.05-me(:0 F nmux) - Los-Alum M £0 YES mes“) - mask) - on? PRESIK) - ux(mes(x).o.81 ... a.” H:...- — m--- -. , ...-- . - . _. _. ._ LNP ran CVCLIIG PRMIETS ADJUST INVENTMV LIMITS IS INVENTMV WITHIN UPPER LIMIT? UPPER LIMIT EXCEEDED--SCNEDULE DECREASED LEVEL IS INVENTMY COIMUNICATIDN FACTG AT STANDARD LEVEL? IKREASE PRESSIRE DN SALES DEPT. LOVER LIMITS III INVENTMV IS mvcutonv UITMIN LOVER LIMIT? LOVER LIMIT DREECIIED--SCNEDULE IICREASED INVENTORY LEVEL IS INVENTORY FACTM AT STANDARD LEVEL? DECREASE PRESSURE GI QLES DEPT. RAISE LIMITS OI IINENTMV __ _. .-__-—...— I" SJHIIITINE IINENT (COIL) I ’ Ecmux) - mvm O a.mamva— . AgggzaLmI‘muégcgsnuscnmuu susm 1199 L commt] - cm noouct LOOP sumwnut GENPRO ' ZERO WT TOTAL PRQUCTIGI TPROD I O 00 29 ROI, NPROD I ACTPRO(K) - O I ° ZERO OUT ACTUAL PRODUCTIM Z IS DRUNK) :- I \ YES ' MS PRDUCT R SEEN DROPPEO ' LOOP Fm CYCLII‘ PRWUCTS no rscuto - vrmcm - mm 9 sum (:03 ° smut ncoucnou LACIPROIK) - scum. (I o .OZSMUIDUMIx) )J - GENERATE MIDI): nan AID ACTUAL PRODUCTION ITPROO . TPRI'DOACTPRIHK) I - ACCIHULATE TOTAL PRQI‘TION l 249 CONTINUE~ ‘ . ‘ EID OF LOOP 122 SUDRWT INE PRREC 00 324 x - human) IS tranc(x).L£.rrmc(x)\ YES / IS TSAL(K).GE.Pch(K)\ us no I nzsm - mcsm . 0ch 1 .2 PRESIK) - nun (PRESIR), I I Is RPCIK).EQ.I.OR API(K).EQ.I\ YES ' NO [us ITRPIIK).LT. mwc(x\v:s ITRPC II KLmK ITRPI I NLGK vn£s(x) - PRESIR) . DELP mzsm - names“). 0.a)‘ NO [mm - xLoch J! mes“) - mtsm o OELP Pnesm - MINIPRESIK), I.z) \ll 421224] CONTIIEFI 113E us 151M051. "MW“ LOOP FOR PRWCTS IS PREVIOUS FORECAST NIGMER MN CIRRENT FORECAST? CURRENT FORECAST DINNER-4‘5 PREVIOUS FQECAST MET? IICREASE PRESSURE GI SALES DEPT, MS A REQJEST rm PROIOTIUI IKREASE 0 PRICE DECREASE SEEN MADE TNIS MRTER? \AS PROIOTION lKREASE REQJESTED NME RECENT THAN PRICE CUT REQJEST? REQIEST PRICE CUT STGE TIME N PRICE CUT NEWEST REQUEST PROIOTION IIIREASE STQE TIME 6 PROIOTION REQIEST DECREASE PRESSIRE GI SALES DEPT. \IS PREVIWS FMECAST MET? REQIEST PROIOTION IICREASE STORE TIME N PROIOTIM REwEST IKREASE PRESSURE OI “LES DEPT. ED PRWCT LOOP 00 501 1: - 1.1191100) 501 I BASISIK) - -PRES(K)I CALI. RANK (umoo, BASIS, 1110111110] X00539 L - I,NSALS> r'——X 00 5111 x - 1.1191100) I 000110 - (m05115(1<,L) -PREORD(K,L)) ~PDELORIK,L) I L—-I 51. I I oarr - (EFFIK,L) - PEFF(K.L))I I BASISIK) - macho/DEFEI I DVDE(K,L) - 0151500 I 5’4 commas J CALLRANKOIPROO, BASIS, 1111101 I 1 FX 00 525 x - 1.1114100) Incmnxuc, .—J 525 L) - RNKIK)I 4k..— CONTINUE I PERDIIL) - o. I DO 530 K - 1.0111100 SUlIOUTINE SLS‘IEN o LOOP FOR PRODUCTS ASSIGN PRESSURE AS RANKIN BASIS RANK PRWUCTS ON THE BASIS IN: PRESSURE LOOP r011 SALESI EN LOOP m1 PROOIETS ESTIMATE MARGIML RESPONSE N ORDERS ESTIMATE MARGINAL EFFORT ESTIMATE MARGIML RESPONSE TO MARGIML EFFORT STME MARGINAL RESPONSE Em PRODLCT LOOP RANK PRODUCTS ON THE BASIS OF MARGINAL RESPONSE LOOP FOR PRODUCTS‘ STmE RESPONSE RANK EID PRODINIT LOOP ZERO OUT FIRST PERCEIVED DIFFERENCE IN RANKING LOOP FOR PRODIKITS .t I C? 1111001111: sum: Icon.) 53o PERDIIL) - 9:001“) . 11050110111111“) - RESRNK(K,L)) 539 I c01111110_£] \ IF SLSCOLEO. O I V IJJ I NSALS/Z .141 - .11 . 1] ———X00 5550 1: - harm» E—Xoo 55411 L - 1E 33“] our. .4511 . Noumm I I YES L DO 5551} L I JJI,N9LS 55413 I 01111 - 0011 o 0VOE(K,L)] 00 55% L - 4.11.1151“: I 551111 0110:1114) - 00111.14 ~I sssoI 0011111105] CALCULATE FIRST PERCE IVED DIFFEREICE IN RANKIIflS EDD SALEQIEN LOOP IS THE SALESIEN CD‘IMUNICATION FACT“! AT THE STANDARD LEVEL? SET UP SALES REGION "DEX “RIDERS LOOP r011 PRmUC TS ZERO OUT DIHMY VAR IAOLE LOOP FM REGION I SALESIEN ACCUMULATE MARGINAL RESPONSE VALUES LOOP FM REGION I SALESIEN ESTIMATE MEAN MARGIML RESPOISE ZERO GIT DUMMY VARIABLE LOO PM REGION 2 SALESIEN ACCIHUIATE MARGINAL RESPOISE VALUES LOOP FOR REGIUI 2 SALESIEN ESTIMATE MEAN MARGIML RESPNSE EID "00111:: LOOP 125 (c) ”INTINE SLSIEN (CONTJ I 5551L001111110:¥——— 1—————-X 00 1000 L - 1.1151053 ° LOOP 11011 mum on 5110 11 - 1.1111100 ) . LOOP 11011 1111000015 J 910 11151500 - (1101c:00 . £51111": 1101:1111“ 111100111“ communal - ucsrme))tovo:01.L) RANK ON OASIS N POTENTIAL MRGIML CONTRIBUTION ICALL 10111110111100. 011515. RNKU IS PART.NE.I YES 0 IS PARTICIPATIN DleURAGEDT DO SAI K I I,NPROO> LOOP FOR PRODUCTS I 5111 I 111151501) - 11111101) . 1111111111100] wICALL 11111111 (11111100, 01515. 111111“ 00 558 11 . 1.111111%———1 55: Lawn) - 1111111111071] - mu: 011 011515 01' 1111011111 mu: run ,I/ wanna v.0. L4 559 I comma] 00 $60 11 - 1.1101109 IKLIDE II‘OIMATION FRO! SALE’IEN SALES MANAGER'S RANKIN USIIB NEU IIN’MMATION LOOP m1 PRNUCTS ZERO OUT 2101 PERCEIVED DIFFEREICE IN RANKII‘ CALCULATE 2ND PERCEIVED DIFFEREI‘E IN RANKIIG 0:110: - 0:110: + 1105(911111 91111111(11.L) - 11111101) 11 (K,L) - 11:5111111(11.L) ) 126 5110110111111: 5L51:11 (00111.) /IS ANEEDI.EQ.O \ YES 0 IS NEED-INDEPEIDEICE FACTM AT STAIDARD LEVEL? ID CALCULATE CIRNGE IN PRCEIVED DIFFEREI‘E 00:110 - 0:1101 - 0:002] ’IS DPERD.GT.D\ YES MS TMERE BEEN A CMANGE IN PERCEIVED DIFFEREICE? EFPARTIL) - “0111110) - DELPAR] 11:01:: :00:c15 00 PARTICIPATION IEFPART(L) I EFPART(L) 0 IDELPARA IKREASE EFFECTS OF PARTICIPATION 0.5100010 tun - masurmnu) 0011 - 1.07 CALCULATE 201:" PARAMETER - (0.025110 - EXPI-AIDUMIH I H 0111 - 0.05:1! - EXH-MUPARHLH) + 1.0] - CALCULATE EFFECT PARAMETER l 00 S79 11 . 1.1101100 0 LOOP :01 011000015 579 I 11011-01100 - 0111.011316] - CALCULATE 1101111111111: 11011:: u! 00 582 11 - 1.1101100 LOOP 0011 0110011015 582 01151500 - - 0110:1504“ 1 5:: 0110:115 A5 1111515 IcALL 11111111010A00. 0A515. vouufl J. 00 587 11 -1.11011® RANK ON THE OASIS N VOLIPIE LON PM PRQUCTS 5071 01151501) - 511111111(11.L) . v0u1111100] - 5:1 0A515 0011 011m 101111111: 127 91010111111: 5191:11 (00111.) [01111. 11011111111100. 1111515, swung - 111111: 91:91111115 :111111. 1111111111: [0111111 - 1111110072] . 5:1 111 01111111 :01 1111111411011: 00 592» 11 - 1.1111100 . 10011 1011 111001101: Ion-1 - 51911111111) - 011111 - 0.5 01110111111: 11111111410101 ANIG’AC I I.0 - (0.0“1MHIJIH |1:r:(11,1.) - :r:(K.1)| 1 110131 1119111115 :rrmt h---[ssz] 91111.1.) - 110110111111."qu . 01110111111: 11:11 :110111 L1000 0011111111: . :10 91:91:11 1001 1:351:21 128 5110110111111: SLSGEN 0 ZERO WT TOTAL SALES N ALL PRMIITS TTSAL I O. LOOP :01 CICLIIG PROOIKTS ,00 7119 11 - 1.111100 1 [11111:(11) . 11111:(11) 1 II I v 01111 - -11AA(11)a((11111E(11) 12)::12) 0:11 - KON(K)*AA(K)¢(IIIHE(K) 12111811110111) AOVAIC E PRWIIT AGE COUNTER ZERO WT TOTAL MOERS CENERATE N04 INAL ODIAID [0011 - 0:11 11111001111) - 0:11] I 5111111111(11) - 1111x1511111111111).0:11)] 51011: NOI IML OENAI» CALCULATE MAX PAST NOIIML OEHAIO I v _ 0:11 - (001111110111111111101111)n1111:1111)))l] - 11011151 0011110 011: 10 1111:: A10 ((1111 ICE(K)nPREL(K))1PHFD(K)) 111101011011 new” 00 710 1 -1.11s11.s In:m0(11.1) - mous111.1.)| [710I 0011111111: [.11 - 115111512 1.11 - 1,1 13 00 730 1. - 1.11 N LOOP FOR SALESIEN "DEX PREVIOUS ORDERS EDD SALEQIEN LOOP SET UP "DEX IMOERS FOR SALES REGIONS SALESIEN LOOP PM RECIN I”. I J; [11015: - 5101100111110011191111] - 11:11:11": 1110111 0:1111111011 111111 11011101 \gl 0011110 Ce) Luv- 210111-1:::111.1.) - :r:111.1)n:1::1111.1))] - 0:11:11111: :110111 «51011:: I \1 [ORDERS(K.L) - 0011:1111“) -0:v 11015:] - 0:11:11": 0110:” I G) C'P I 69 5111110111111: 5:50:11 (00111.) 730 counm: 129 00 731 1 - .111. 1511.5 Eels: - 510110111111100111:11)| 0:11 - 2100110111114) - r:rr(11.1)n:r::1(11.1) ) I 010:15(11.L) - 0:111:FF(K.L) 1 11:11 1 NOISEI 731 CONTINUEJ 11-1.1151115 00 711 _ 711’ 1010111) - 10110111) 17 010:1; (11.1.) [01111 . 11:111001) 1» 111v(11)l Z15 1010(11).1.:.ou11\ 1:5 1511.111) - 01111 111111.111) - 1.0511111111111) 11111.01) - 1.05111111111) L 1511.111) - 1010111) 1.. 115111 - 115111. 1 1511.111) E51111) - 05111111) 1 151nm II 7‘19 I CONTIEEI 0 END REGION I LMP SALEQIEN LOW :01 REGION 2 GENERATE RAIDOI DEV IATION FROI NMIML DEMAND GENERATE EFFORT RESPINISE GENERATE MDERS EIO REGION 1 LOW SALES‘IEN LOOP ACCUNULATE ORDERS :01 PRMUCT R CALCULATE MOUNT N R AVAILADLE IS ENOUGH AVAILABLE TO MEET MDERS SET TOTAL SALES EQJAL TO 1111011111 AVAILABLE RAISE INVENTMV L II ITS SET TOTAL SALES EQIAL TO TOTAL MDERS ACCUIULATE SALES U ALL PROUCTS ACCUNULATE QIMTERLV SALES EID PRWI‘T LOOP 130 SJDRWTINE DOOR 00 799 11 - 1.1111100 111111111111) - 111v(11) J1 [1101111) - 11111110111) 1 1111111v111) - 191111)] ' 11111105111) - (111111111) 4 P111:11111(11))Iz)115101c #4; [71111111010 - common) 1 10111110110] I 0111 - 0111151111) L cuncsHK) - (ICOSTtCUHPIMKhO (”RAM 9 LIN/(OPRAH o I.) Eosnnm - (cuncsux) - 0011) o suuMKj [ 11151111111) . (cuncsflx) - own/11111110110] _ momm - «6511111111 1 11vco111111) oPRElW(K))/IACTPROIK) 1 1111111111110) c0575111) - 11511111111111110111111) 1 111101111) 1 11111105111) 1 1105111511110/715111) [111v - 1111cs11111115111111) I 11, 1111019111) :2191101111) 1 .8101101'9111) um0n111) - .210111101'111) o .8101110r9111) 111101111) - 11111 - 10515111) 11111101111) - 11101110115111“) ZERO OUT TOTAL PROP IT LOOP m CYCLIIG 11110011015 IIOEX PREVIOUS IINENTGV CALCULATE INVENTMV’M-MU CALCULATE INVENTGV IIOLDIIG COST CALCULATE NEH CIIIULATIVE TOTAL PRQUCTION STME OLD CIIIULATIVE PRODUCTION COST CALCULATE NEH CLIIULATIVE PROOUCTIN COST CALCULATE PRWUCTION COST 1011 THIS PER IOO CALCUIATE UNIT COST N PROOUCTIM CALCULATE AVERAGE COST G INVENTMV CALCULATE TOTAL COSTS FM PRWIIT R CALCUIATE REVEIIJE CALCULATE PAST 9 MONTH AVERAGE PROFIT CALCULATE PR6 IT 131 Q? 5110110111111: 0001 110111.) TPROF II TPROF 4- PRNUQ ACCIHUIATE TOTAL PRW IT 7 111013111) - 5111101111) 1 3111013111) . 0111121111111: comma 341011111 "511110: 11110111 011013111) - 510111101111) + 5111911013111) JLHS I 1011111111: I 9793 K - K o I EDD LOOP INITIALIZE LON INDEX IICREHENT LOOP "DEX [15 0101(1).u:.1.011.111v(11).111.1\ no 115 H 11111100 - 11111100 - fl [go 51 1111001111 1113111111113 1‘ yes IS K.LT.NPROO ' SNOULD PROOWT IE CONTIIRIEDT REDUCE IRIHDER OP PRMIITS SNIIILD PRWIST LOOP CONTIIRIET MVE LESS TMN ELEVEN TINE PERIWS ELAPSED‘I Accuuu U111: 5111mm STATISTICS] “- 132 SUDROUT INE NEWRO [1105111 - 1051111 . 111091 1113110115 0111111115 511:5 f—TTEILsMT-E—j . 11cc111101111£ 0111110 101111. 0111111111 511115 W511. 1 0511111)] W 11“.ng - 10011 101 01011111: 1110011c1s v IS NMINAL DEMAND LESS THAN 502 OF [15 1101091111).11. 0.5015111111111111“ ves 111111 PAST 5111.15 | 110 .1 870]:co111111u£ - 10011 END [880 [ 0101111) - fl( . 01011 111qu1 111100 - 111100 + 17 - .1100 11:11 PRODUCT 5k 5 511 11111 11001101 111111111 111111115 “(TI PARAMETERS l RETURN 00 999 11 - 1.1111100) [1s 11111£(x).1£. 6\ 115 no rnon ’9’ EB [15 011013(x).c1.1noa no 1:5 BETAI(K) - BETAI(K)t(I-ALP3) I 1110001411) - (1 - 011nm) 0PR060L(K) + 011A2(x)au11013(x) PROGOL(K) - n1x(o.o.110001(xn I15“) - IS(K) + II / 7 1s (IS(K) - h) T‘\. 133 SUBROUTINE REC - LOOP FM PRMUCTS ° BYPASS PRICE S PROIOTION DECISIOIIS IF PRODUCT IS NEH ° DID PRNIT EXCEED PROFIT GML ° REDUCE EHPMSIS ON RECENT SUCCESS ° REDUCE PROFIT CML - INDEX SEARCII CWITER - HAVE ALL ALTERMTIVES DEEII TRIED? . o . 157(1$(1<) - 11X 0 o O ' US SEARCH DIPLOYEDT ’ SET INDEX OF POSSIBLE PROFIT CML REDIITIOII nun - 150(x.1) 15011.1) - 15011.2) 15001.2) - 15001.3) 150(x.3) - nun 928l 1s(1<) - 1F. ° RESET SEARCH CNITER - ALTER MDER U SEARCII 134 G) I 5110101111111 EXEC (10111.) 930 1111111111) - 1113 - I I .(1-111111(11)) 1 01111102] ' 1110101111) - (1 - 3111111)). T 11110001111) 1 011111001011101301) ‘ 11101101“) . 1111x(0.0.1110001(11))| I 15 (15(1)) \ 15(15(K).01.lo YES NO [1111112110 - 011mm.“ - “'3“ 3111201) - 11111201) 1 1111310 - 111112111) 11L 15(1101(x.1)) g -1 o] l F§(u11013(x).01.u11019(x)) \ 110 m] 11cc(11.1) - 1111(11 - I) -j IS (BETA3(K) - I) .3 . ED |11cc(11.1) -11cc(11.1)11}-1 @Cé IIIREASE DIPMSIS OII RECENT SICCESS RAISE PROFIT GOAL HS STRATEGY SEARCII DIPLDYED‘I HAS PROFIT 110111 REDUCED RESET SEARCH IND ICATM HS PRICE CHANGED TO IKREASE PROFIT MS PRUIT GREATER TMR 9 IIOIITII AVERAGE PRUIT MS PRICE IICREASEDT 1.355 (9 51101001111 EXEC (CONTJ L1s(a11113(11) . 1\ 111111301) - HIN(BETA3(R),DETA 01111301) - 01111301) + A1111 0 111001.” - 11cc(1<.1) - 3 01111301) - 11111301) - ALIA 111111301) - 111111(011113(11). 011111) 1111301.” - 111001.01 2 0011 - 11111100111101). 5101c. ucs111101). 110111. 1511011. 11.11101). 1101101)) v 15 004.51.110001“) YES 110 111101) - 0 $ 15 1101(1).10.1 115 NO 011111 - 150(11L1) [ 15 (01.1111 - 2) l— O o 1115(1) - 1115(1) + 0111] [953 I 11101101) - (1 - 11191111011“) @565 MS PR ICE IICRflSED II€REASE PRICE CHAIRSE NULTIPLIER DECREASE PRICE CNAIIIE NULTIPLIER COIPUTE PROJECTED PROFITS DO PROJECTED PROFITS NEET PROFIT GMLT DENY PROIOTION AND PRICE REQIESTS IIAVE ALL STRATEGIES OEEN TRIED SEARCH FOR NEH STRATEGIES IKREASE PRESSIRE ON SALES DEPT. IMREASE PRO‘IOTIOI Q SUIROUTINE EXEC (CONT.) 136 (011 TFORC (K) I I .D3tTFORC (K) ] . I 1101101) - (1 - 1119-1101101) r1 1111c101). 1101101) 01111 - 11J110(111v(11). 5101c. 01511101). 1101101) 151101. a IS DLH.GT.PROGDL(R)\ YES _1s(1111:(11.1)) 1- I DETANK) 0 HA! (I .OI.BETA3(K)) 1111101) - 0111300111 11101) 11000101) - (1 - 1111201))1 110130101) 1 0111200101101301) .. 1101:0101) - 111x(0.o. 11.000100) 0111 - 11111011) 1 11cc111.2) 1 1111101.!) “33‘ 15 11c01).1:1.0 v15 1s 11101).01.o 115 NO IS DW.GT.O YES Cy @565 REVISE FMECAST DECREASE PROIOTIN REVISE PROJECTED PROFIT REDUCE PRNIT GOAL IS PROJECTED PROFIT GREATER THAN PRU IT GOAL? HS PRICE RAISED SUCCESSFULLY G DECREASED UNSICCESSFULLY RAISE PRICE CALCULATE PRICE CI-IIfiE INDEX IS THERE A REQIEST 101 A PRICE CUT IS THERE A REWEST 101 A PROIOTION IIIREASE 137 SUMOUTINE EXEC (CONT.) BETA3(K) - 111110. 01, BETA3(K)) BETAT3(K)- MAX(. 99, BETA3(K)I 111cE(K) - BETA3(K)31111cE(K) i I110M(K) - (1 + ALP5111011(K)) I? IS DUM.GT.O CHANGE PRICE IN BEST PAST DIRECTION IICREASE PROHOT ION HAS PRICE CHANGED IN LAST 9 MONTHS N0 BETA3(K) = MIN( .99. 8ETA3(K)) ' CUT PRICE PRICE(K) - BETA3(K)~:.-PRICE(K) 995 DUM = - m— IPcc(K,3) - 11cn(K,2) . UPDATE PRICE CHANGE RECORD IPCG(K 2) - 11cc(K I) IPCG(K 1) - 01111 111:- ISAL3Q - 0,5;1TQSAL + 0.55ALfl DO 885 K - 1,111100 ISHARE(K) - 0.1510511qu] - UPDATE PRICE CHANGE moex (PRICE 1115 BEEN CHANGED) 0 UPDATE PRICE CHANGE INDEX (PRICE HAS NOT BEEN CMNGED) END OF PRODUCT LOOP CALCULATE 3-QJARTER SALES AVERAGE _ o LOOP FOR CYCLING PRODLRZTS CALCULATE 3- QJARTER AVERAGE SHARE ITQSAL) + 0. 55101500 01' TOTAL SALES 6) 1L 138 51111101111111 one: (man) 1110:) - 11:01) . o - ZERO our PRICE AID "011011011 REQIESTS I OSAL(K) I O I - ZERO OUT QUARTER SALES ‘— 385 COIITIMJE ' EID PRMIIT LOOP I ITIQ I O I . ZERO WT TIHE-IN-QIARTER I RETURN I FUICTION PRJPRO IDlflI I INVtSTD IC I . EST IHATE TOTAL INVENTMV HOLDIIG COST E00112 I UCSTétTFMC 6 SETUP I ESTIMATE PRNIITIGI COST I 00113 - fCOSdTFO'C/TSLFORI I . ESTIMATE "no cost ALLmAtIou * 11.1101 - 111cc - ( (00110001301113.1101) - EmmuE UNIT 110111 Irrmc) [mm 139 SUDR OUT INE RANK ‘ C-RANR SIALLEST VALUE OF OASIS FIRST-- LARGEST VALUE U MSIS LAST m . 1001 101 CVCLIUC 11001115 I0U111 - 1. 01112 - M515(IL| - 5E1 11111111 VALUEs 01 011111111 UA1IAILE5 ___3( 00 10 1 - 2,11)' - 1011111501 1001 10111101 fls 0112:1E10151501) YES - 1011mm Lows! 111110E1 111111 NEXT VALUE - 01 IA515 110 011111 . 1 00112 - 0151501)] ~ LE1 00112 EQIAL 11E11 LovEst VALUE 101mm: -‘ E10 01 1001 I 1111011111) . JI - DEFINE 11111 or 11EU 1011151 VALUE [35151011111 . 10.117] . A551011 A10111A11u 111011 VALUE 10 111515 110 connmE - E10 01 1001 FUKTION RNDIII x - 5011 (-2.oAALoc(1A11E(-I.0)) ) 11105 . GENERATE A 11110111 VARIABLE 11m A (6.283I853071RANF(-2.0) ) 1011111 01511101111011 111111 11EAII or 11:10 7 1110 STANDARD DEVIATION or 1.0 m: APPENDIX II Glossary of Variable Names 140 $030022 «0 «0.000 001.000“ $0300.05. 00 00>: 0000000 .000. 000.00000h .5000. 00000 000 0030000. 0030:000l 6000.00 00 =03 .33.... 2: .0 00.0000 0 602.050.... 00 00 .0 0001000 I .2. 10000050.. 000500 1000— 000000 0000050 0_ 00050000 000050 _030< 000.000.0000 30000:. 030,—. 33>Z—Um 332.: 2.0.00.0 a. OUT—OW Our—Omn— OSOMAPU< GOmmh Ozmzuo 05.06.33 .000..000 .«0 X005 000 003.000 0u0aso 40000000000 .03. 00 005.30 .00 1000— .30: 00200000000000 0000001003 0000000 03309, 000005.000xm $0300.00. .00 00.100 0030000m 000000 00 000 0. 00>: >00000>0_ .00.. 300002 40100000300100. 00 :00: .0030: 0:003— ..0090: >00000>00 0.00000. 5 0000.000 affooEm .~0>0~ >00000>0_ 00 2.00: 00003 .00.. 00 0.00000h .5000. 0100 00 000000 0000000 5000 000—000 0.. 000.00 03.000000 mea 3:00 00 o<0>z_ .03250 s:>z_ 3340 00500.05 .4 03H>00nm H> ho fifléOa—U 141 00.0.0000. .0 9.3000 00.00.00 ~000h 60030.. 00 3.000 00 00300.0 .0000. 00303.02 030.00.. 0000000 4000.000 000000 00.000.00.000nm 00300003000 00000 000.000 00.00.000.00 0 00:00:00. .0000— .0020032000 00 000 .u0i000 0. 00000030 00300000 0. 000000 000000.20 00300000 00000m 60000000 .0 M03000 3.00.00 ..00m000E.0~0m 400000000 -00; u0..00._00:. 0.0.... M00000. 3.00.0.0 0.00 :0: 030.0 000.000 .000. 00300000. 00.. .000500 .0000. 409.000.000.500 00500—0: 300000 030.00.. ~000050000xfl 000.000 0.00m0000. .0—0. 000 0.00.0000 0.005.000. 0003.00 0000000000 00300000 0.....— .00030 000.000... 0.. 0000000 0 .0 0.000.00 000.00 0:. .0 .0030 .0000M000 05:00 00 1000 00.000000. 0.03000 .0 004.00 >0000 00 00.: 030.00., 00.0.3— .0300000. 00000.00 000.000: .0000. 000.30 30.00:. .0030..— .0000. 000.00 0.005.0—0m 3:02.03. 6:020..O> 3:00.000: 2:0 .mSnma—MQ AAVH¢00m 000000 300000. .00_>00n~ ..00000 35002 000000 0. 000000 «003000. 00,—. 5000000 0000 .0 :00. 0:09.00 0 00.00.00. 00000000000 .0... 00 000.300 .00 00000 .0005. .0 000.00.. 0 00.03 00 0.0000000 .00 0030., .03.: 00. 00.: 630.00.. >550Q 0000 3 0.003000 3 000000000 3 00.000000 00000 3:02.050. Cavmuml 359w000 00 .000. 00 0.00000..— ..0_0. 00.000000h 600.00.. 0.00000 7.0000 00000000000. n00.00000u 00.0 00:00:..— ZNSWJO 0:. 03009.5 $0048 .5000... 30:000.. 0:000 0:000 0400 000.000 .0050... 00.00000 00.00000. 00000 2.083... 3329900 . 142 .300 0020000.:— 35003 600.59:— 2..3 an»: ac .000 73:: £305.22.— o>u=u 35.....qu £03093 2.: H0 3: on» 095 3.00 00300002... ~30. 03:30:30 £03093 of «0 3: 05 no.3 0003.000 03:: no .3050: 0333530 $330023. no :5. W00 3530: you :00 2.0.—Sm .3000 9:30: T3303.— .noio 35:02 £030.60...“ 330< 40>: >uouco>5 2.3.:an 43.3 >u3¢0>5 0.5005 u:0_>oum 3.0.3 35:0: .0013 300.0000 .034. 35:05 ~30k. .25: >u0§o>5 2030: 30:3“ .35: 53:02:. 0233 40>: >u0u00>5 «60:00 . 0030000:— ~430< .oucoaoou «new. 0. 0:1 .coEuu:_v< .008: no 33302” 4.3:» 5.5.» 3.3?— H80 H80— z<¢n~ n 7.:th230 E63230 0.0.5 0:000:2— 3:339 3:053”: c:>z. 3:5.me koala. £00m 05:63.5 3:4 z. 3:93 90¢ >m0 3 iiuhum 3 .vzhhu .033u¢> 75.3: 10323030 unsauocun now .320ch .cozlmhfl> EO‘GQH .BmuluaaUmUU a: 10.3 «GCaIEOU icon-oi 5 00:07.2, EOE—am .0365 59.9 ..:0.uou no.0. no“ 0.3052. .80:— .000050 0.5005 .00.:an 6.2.1.3 35:05 .=0_>ou.m 03.230 .333 0030500“ 5:33. 20 :02050..& .ouszvcoaxo Baton—cum 3.0.3:. 00.03 our—om £30532— :03230 ~3033 600590 3580: :00 535.3: .9553. $5502 40000.2. 00.. 000501 35502 (3.0532. .3323» 2.533 .032...¢> >859 .nuovuo 35:02 .3 «0:00.:— uo and ask .3259:— a no 00—: 35:05 ~30P $030: 0.005003. :0: 0. one uct.: 2:32:93 88 "mm—oz 3:095 z... .2. 3 ifizmnxo 3 Jan—momma £6020 3:48: 3:29... sang: 6:20: Q§§3a 5:05 002:. .3309; ‘0 0m< £0000...“ 9.30000 08 .805 6:05:00 35502 6:09:00 1550: 0000 508332 .0300 300.0000 .0300 0.0000000 E00050 ~30“. .00—00 0.000.300 00030.3 .30.“. £30000 0:00.30 3 5:05 0:00.30 15.305... 0.020 3— van Ozn0 an: 2030— 00 auavoul ~> 23m run—dam >sz Nana—23m Gnu Xian .039, 000000.:— ouau0>< .0030» x005. 0.500000 «0 :020550@ NSQADM. .0500:- 0501500 no 50.0. 00:05 0:00.30 60.50.; on 0. 0_ 00:00.3 0 3 ugh-00:05 x00:— 56000 3:: unauo>0 5:05 002:. .0200: ouuuo.‘ 5:05 003:. .059:— 3.09 .52.. 2.5 .50.:— 35. 0u000>0 5:05 052 .33.:— 35:0: .060...— 0u000>0 5:05 052 . 00.0% . 02:03.“ 6030300.:— 35005 ‘30.“. .0300 35:05 «30,—. .3000 005.. 35:03 00035003.» 53350:" .3000 ~33 35:02 $03002: 00 0000 ouduo>< .0000 030m . .000 0020:0000 35:05 3:: «0100 0000 30200. 5.30 0.003. 00:95 0:30:33 0:29.: 00:00 0:00:00 0.30000: 0:093 0:533 0:002: >0: 00:00 3.0:. 0000 0:295 0.6.500 0:250>< 03.3000 33‘.“— F8: 144 63: 3:93.? .33 mo chasm do... 350:2... 7.9—. .omnuu.‘ .07. hounds—7002:. .03.... n3: 3:: n J X! .0323 3.1:. :2. «o .89: .3332: .3: 35:02 .cofioEOLQ 9.3523 .3. 53339 94:20 .ouszvcoaxo scion—3.3 J39... 0.23.0.5 you E3339 93.2.0 .fiofiuuamov no?» so cusauoul .26 out; a .8 3260‘ 6040.55. coZoEOHQ a .3. 3350.5 $30.3 6030.3:— 323350 3 van: :ozucah £029.23: omcazu our... :0 :E: .333 £333.35 cucusu aura :0 SE: yuan: £33335 omcusu our:— uOH Egucou owcasu £33525 omccsu our—m . 3.3.3 3:: ounuv.‘ 5205 05.2 Quins—v 32 mo «5.5:.qu an voucanu :02. um: our—n .: 33.3955 x32: .oucovouuua guns...- uo .320 .6030:qu 10% “:02. 033.0.- uo .89: scum—$3 .262. 034;... 93m 02m $2 .509: 3:35.: mmq< 3320.: name 3:35. 3:9: E :m m . . . .5535 Az. mmzoammc pug:«x >aouw»wa ram aux; z«m: m34<> m34<> uzo_p«_)wn ....... -.:-s;e‘1;;,‘ - a¢.az¢»m‘ z=:.x.x zsx~z_x nu¢.:un Lo :3» . 44.¢m>o . »_Lomx;iln.oa m 4 a < b .x m. m4c<_z<> pzmozmawa .2o.mmo mm».o.4¢m¢,x»_x J.H¢0»00. 0".“ III II In so. u" l'7() -( ) N N N N N . N N N N N M M ”amazon fan... -33“; £133.23, NNNN N.NNN N NN..N N. .N NNNN.NN.N N N N N N N N : mono..NNNNNV NnNaN.N. NNNNN... . . u n u M N N N NNNN.oNNnNN NNNN¢.N NN.NN.NN “..;M"MM.MMMM N N N N N N N o..N.mNoNNN omNoN.N mNNNm.NN o. NNNN..oNn N N N N NNN.LN.NNNN omNoN.N. .NNNN... . NNN . N N N N N N . . .1- I m V.. . NNNN N N N N NNNN NNNNNN NNNNN N NNmNN.Nm N. N.NN.N.NN N N N N N N NNNN..NNNNN NmNoN.No NNNNN.Nm N. NNNN.NNNN N N N m H N pig. NNNN.N.NNNN NNNNN.N.- NNNN..NN N. N.NN.NNNN N N N.NN.NNNNNN NNNNN.N NNNNN.NN N.. .NN..NNNN N N N N N N ,‘,«-- -_- - N NNNN.NNNNNN . NqN N - -. - N N oN cNNoN N0 N n u N N N N mmNm.NNNmoN N.oNN.N. oNNNN.No u" ”www.mmnm N N N m N N N NNNN.NNN.NN; N.NNN.N. .N.NN.NN N. NNNN.N.NN N N N N N N N NNNNNNNNNNN N.NNN.N. NNNNN.NN --N. .NNNN..NNN,I;111 N N N N N N N .NNN.NNN.NN N.NNN.N. NNNNN.NN N. ..N..NNNN N N N N N N N N.NNWNNNNNNV N.NNN.PI mmNN...N N. .NNN..oNn N N N N M n N NNNNH.NNNNN N.oNN.N NNNNN..NV N. -NN.N.NNNN N NNN. NNN.NN N.NNN.N. NNN.N..N N. NNNN.NNNN N o N N N o N ‘ .1]; i lit it: . : Til: NNNN.NNNNN. NNNNN.N. NN.m . N . M u M M N N N NN.N.NNNNN. nNoNN.N Nocom.mo “M ”www.mmuw N N N N H N N NNNNHNNNNN. NNNNN.N NNNoo.NN-l-:.-NN--NNNN.N.NN N N N..N NNNNN. NNoNN.N. NA.NN.NN No NN.N.NANN o N N N N N II I. .. I £3.11. A! A , . lli. N NNNN.N..NNN NNNoN.N . . N . mnooo .N N. NNN N u u M- N N N NNNNNNNNNNN: -. NNRNNNNH ‘NmmNm..N N. N.NM.MHMM N N N N N N.NN.NNNNNN NNNNN.N. Nmmmm.Nm N..IM N . N N N N . . N N NNNN N N N N N NNNN mNNNNN NNNoN N NN.NN.om N. NNNN.N.NN N N N .N N N Nail .NNN.NNN.NN-- NNANN.N. NmmNN.NN N. N.NN.NNNN N N N N N N N NNNN..NANNN NNNNN.N NNNNN.NN .‘N. NNNN.NNNN - N N N N u M N NNN.UNNN.NN NNNNN.N NNNNN.NN N. NNNN.N¢NN N NANNINNNNPHIIIIIINNNNNnpualununns.NowwmsnuuslbhluuNNN..NNNIIIII: N N N N N N N . -.|;-|I. Tn- NNNN.N..N...‘-:.-.NNoN.N N.NNN.NN . u u n M u N N NNNN.NNNNm. .NNNN.N.T:$1:,NNNNN.NN x.:‘ u” “MM“.MWMM---: N N N N N M M NNNNHNNNNNN .NNNN.N. monmn.om No NNNN.Non NNsN‘.NNNNn:IIII..NNNN.N NN.NN.NNeIIIIINN;INNNN.NNNmulruuu N N N N N N N . NNN. NNNNNN -12: NNNNN.N N.NN .NN . u n u M N N N NN.N.NNNNNN NNNNN.N...::I.NNN.M.NN less.un "an”.mmnm.;lsa- N N N N u N N No.NHN.NNN. NonNN.N. NonN.n. NN N.NN.NNNN N N .MNN:NNNNNNIIIIIINNNNNHPIIIIIIINNmpmnwo mp:lw.NN.NoNN O O c N . i I.—-‘ I- illllll. A ill|p|ll .li ‘ N N N N u u " NNNN.NNNNNN NNNNNHN. NNNNN.NN NNN NNNN.NNNNNT;-;n. N.NN NNNNNN NNNNN N NNNNN.NN NNN N..N.NNNNN o o N N N N . V N N N.N. NNN.NN oNNNN.N NN.NN.NN N. No . N N N N N N N NNNN..NNNNN NNNNN.N. .NNNN... N. NN»M.HNNM “h“ C -J- k 3”" 171 N N N N N N- -N:|I.NNNN..NNNNN -;- NNN.N.N. NN.NN.NN N. NNNN.NNNN L. N N N N N N N NNNN.NNN.NN NNN.N.N 311:. NNNNN.NN.II N. N.NN.NNNN N N N N N N Nil: NNNN.NN.NNN NNNNN.N -NNNNN.NN.::I_ N. N.NN.NNNN . N N N N N N N NNNN.NNNNNN NNNNN.N. NNNNN.NN N. NNNN..NNN N N N N N N. NIII‘NNNN.NNNNN .--:I NNNNN.N. s NNNNN.NN N. NNNN.NNNN N N N N N N N NNNN.NNNNN NNNNN.N 1:. NNNNN.NN N. NNNN.NNNN N N N N N N N N.NN.NNNN.N NNNNN.N. NNNNN.NN N. NNNN.NNNN N N N N N N N NNNN.NNNNNN NNNNN.N NNNNN.NN N. NN.N.NNNN N N . N N N N N NNN¢.NN.NNN NNNNNNN wNoNN.NN N.- NNNN.NNNN 1111:. N N N N N N N NNNN..NNNNN NNNNN.N. .NNNN.NN N. NNNN.NNNN N N N N N N N N.NN.NNNNN. NNNNN.N. NNNNN.NN NN NNNN..NNN N N N N N N N N..N.NNNNNN NNNNN.N NNNNN.NN NN NN.N.N.NN N N N N N N N NNNN.N.NNNN.. NNNNN.N;1|I|I.NNNN..NNIIIII:NN NNNN.NNNN unini. N N N N N N N NNNN.NNNN.N NNNNN.N. NNNNN.NN NN N.NN.N.NN N N .I N N N N N NNN..NNNNNN NNNN..N. NNNN..NN N. NNNN.NN.N ,N N N, N N N N NNNN.NNNNNN NNNN..N NN.NN.NN N. NNNN.NNNN N N N N N N N .NNN.NNNNNN .NNNN..N IIIII.NNNNN.NN.IIII N. NNNN.NNNN --1 z N N N N N N . N NNNN.NNNNNN NNNN..N. .NNNN.NN N. N.NN.NNNN N - N IrN N N ;N --N.19.NN...NNNNNN NNNN..N NNNNN.NN N. .N.N.NNNN N N N.t N N Nt N NNNN.NNN.NN :11: NNNN..N..-.. NNNNN.NN‘::-- N. NNNN.NNNN .---: N N N N N N N NNNN.NN.NNN NNNN..N. .NNN..NN N. N.N....NN N N N N N N N N.NN.NNNNNN NNNNN.N NNNNN.NN-II::;N.; NNNN.NNNN-,1:1: N N N N N -qN-nsusxu:N‘:;;NNNN.NNN..N _NNNNN.N N.N.N.NN NN NNNN.NNNN N N N N N N N NNNN.NNNNNN NNNNN.N.aII|::NNN.N.NN‘,.u NN NN.N.NNNN ;-; s N N N N N N N NNNN..NNNN. NNNNN.N. NNNNN.NN NN NN.N.NNNN N N N N N yN_ N;u‘-NNNN.NNNNNN-;|;!‘NNNNN.N;::III.NNNNN.NN..nn‘-NN NN.N.NN.N; z.. -N -I-N N .N ulunlaIN,;In---:N ‘ni:-:-N-; -NNNN.NNN.N. NNN.N.N. NNNNN.NN NN N.NN.NN.N N N N N N N N NNNN.NNNNNN NNN.N.N -zxi- NNNNN.NN ;.:: NN NNNN....N { N N N N N N N NN.N.NNNNN. NNN.N.N NNNNN.NN No NNNN.NNNN N N N N N N‘ N N..NNNNNNNN‘ NNN.NHN. NNN.NNNN -NN..NN.N.N.NN; N! --N N N ,.xt. N .: I N . -: N .- NNN..NNNN.N NNN.N.N . NNNNN.NN NNN NNNN.NNNNN N N N N N N N NNNN.NNNNNN NNN.N.N. NNNNN.NN NNN NNNN.NNNNN N N N N N N N NNNNJNNNNN N.NNN.N NNN.N.NN .N 33.23 N N N N N_s.11.- N ...... - NII- NNNN.NNNNNN N.NNN.N. NNNNN..N .N NNNN.N.mN N N N N N N N NNNN.NNNNNN N.NNN.N. NNNN...N .N NNNN.N.NN N N N N N N N NNNN.NNNNNN N.NNN.N NNN.N..N .N NNNN.N.NN N N N N N N N NNNNNNNNNN NNNNm.m. NNNNN.NN .N NNNN.NN.N N N N N N N N NNNN.NNNNN N.NNN.N NNNNN.NN N~,:NNNN.NN.N N N N N N N N NNN..NNNNN N.NNN.N mNNnn.Nm .N NNNN.NN.N I. ---xel I11 N .Ie- - N -2 N -;! N . - N - N - N NNNN.NNNNN N.NNN.N. .NNN..NN .N NNNN.NN.N N N N N N N N NN.N..NNNNN N.NNN.N. NN.No.NN .N NNNN.NNNN . N N N N . N N N NNNN.N.NNNN N.NNN.N NNNNN.NN .N NNNN.NNNN N N N N N N N N..N.NNNNNN N.NNN.N NNNNN.NN .N NNNN.N.NN N N N N N N N NNNNNNNNNNN N.NNN.N. nNNqumN .N NNNN.NNN« N N N N N N N NNNN.NNNNN N.NNN.N NNNNN.NN .N NNNN.NN.N N N N N N N N NNNN.NNN.N N.NNN.N. .an&NN.NN .N NNNN.NN.N ‘0.1.w¢. “ 9. \ar 1' 1'72? ( ) v~-_ov~o.«nnu .~ .a....omfl.z cw ~n-.m..¢ .~.-amm~.nnma «w no.0.nnve- on nnvn.sovu c~ ,mowm.oonw I'll-ll -wui ”Hua.s.mu om ”a.m.ssnfi ‘oo, xvno4mooo o. um~o.~mmn aw:.~m.o.«mm~ a. va»..ncmn a. suum.sum~ c.--oonn.n«- a. "a.m.nnon a. usoo.oc- o. amna.omu~ o. ann«.~m~n a. Nohv.m-~ o. «.mn.~onn o. -m~.~nm~ a. ocho.amnn o. n»oo.~«m~ co um-.memo c. mum..oomm co nou«.n~mm v~ namm.-~« .u cmo«.monn cu mvnn.mo~« v~ namo.smna cw o~pn.anmu om avm~.o~nu cw onmn.nova v~ annm.~nna .«2...mn.~cmyx. cw ammo.occu cu oomn.onnu v~ a~¢o.«vcu vn owmn.oonu v~ ammo.«onu .u n~n«.mnma v~ n‘uo.vcnu a. “ann.m~mn o. oomo.onu~ .. na~n..vnn a. \wmowamnou no onwo.ooen a. .mmmHfionn . '1'!" o o u a o a e 9 ~ ~ hnvoaonkou omno.o . o o u m u w « coco.do~d.« mwunaqo ..mmwwm.mws o o a a o w « mno~.oun~o somno.o. ocmo,.=o o o u u o a « nuvq.~monva somno.o. .mooow.on o 9 ~ u a -: m- a ~.o~.~oa«o “can“... cohvh.nw - o o a a o a a amo~.nnnoo somno.o. onoon.~o . o u a o M « mnoflw¢ofima nomn°.c mnrmw.km . o a u a a M canonm,ao°« souno.o some..mo.|||| Ru , nag» wanna somno.o. coma..sn a o N a 9 4w .1 a o a u a noa~.oo~o- sownu.o . o o w w u a a cod..«h~osul smmm°.o. mmumm.mm a o a a o N a ooo«.cnsq- hm~n9.uu mucac.ns o o a a o m o aam..uaoo.« summo.o masoc.cm a o « ~ o o o°-.~msooa somm°.°. omsn~.~o . . a o a a o M e omo~.onnnom nommc.ms o'cnw.om-u a. 06:0 ~.a~ o o a a o a o «omnuflchON sommn.g mahno.no ‘:-::-I.a --.«num «unwed ..u-‘sommo.c..:,-z-avfimc.om ! ( o o a u o o o o N .d~a~.nnom- . m . . e e m m u 9 ~ ~oo~.°nonca MHMMM.Mn “WNW“.NM ov‘ a“ o.~c~n o o n u o o m nocn.w~mo- nomaa.u. .oauwa.no o a n u, oi..:||- ” -.ac-:.~:.a,o°ow.oamoofi o:.- nvmm,.o ~c~nm.on o a a w a a oonm..»°~.~ a.mofi.°.=-, :.~.¢oh.oo “- o o N a o n M oosvumnm~sn nemod.o Ncoofl.on . o a a a 9 « oanuunawmm nvmmd.u .aucn..°~ ,M coo. .~cnun nqnod.¢. ..¢~w“mm -::-. o o m w ,glnle--.;;. o o u o o o muov.a~o~om-a :1. .. . t.1 . a o n m u o o ~nmo.noc~cn ”MWMM.W MMMMM.MM a o a a a “ “aneusmaonoonaow anomn.o-|ll ougoh.mo mono nononn os‘mn.p. -wovnm.sm .III. a Q ~. g :; .q- ..:I.sgr‘;-j ‘ o o d ... ~ ... 22.33316..- 28.. ii! .‘ 1-- o o N u m a N “cm“.snmmo ”mfiao.w. mmmmm.wm . a a a a u w 'hn~.eon-n- oawoo.c. Nocmw.«n . o u a N ~ ~ ~nn“..,¢no omflon.o. ~=o~0.5m : a o u 9 ~ M ~ oooo.onnoo omfloc.o. ~¢~am.~o o o w as a - ~-::-nnon..~m- - u-{ oofiau.o om¢m~.mm o o w a a a w Keno.nasoo oowoo.o ..-::.«..-.~o |.|¢- a o w o ~ g a ~mno.nnao~ ooaoo.o. -n~m.mm o o a 9 ~ m « o°~*.ooo°.m-- .oofioo.u. cacou.ms . o n a a a « .nuo.n~¢~o oowoa.o hnm~0.oo . a a a i a a .css.moooc« oofiao.o oaodm.on o o u e a a « noo...:o~oA u!!! omaoo.oo ~nmmo.eo . o u . a w « an.”.o'ooc« omfiao.o - -,: mv¢mm.~o . o w a a u « onno.ua¢.¢ conco.o. oooo\.~n a o u a a a H nnmmmonsvmu oofioo.ou ~mrom.no ”an“ ocean oofloo.a o-oe.nn ‘:z‘ a o. m a , o o a u u a «a.m.ouoosun. movo.o. -- ‘. .van!. a o u n m a o aov~.cocssa “movo.o mmmum.mm a a a a a a a unno.noon- «no.0.9 nnavu.ns o o u o a a noaoqmaosnu anooo.ou vam.d . . a a . oucc. . . ...n a o u a ooooon sauce 9 "mean no a ooon.nuouou unevo.o. osnuv.on 'IJQI' '- 1'73 C ) . 1". NN NN NN NN NN- an «N «N «N «N «N «N «N «N «N «N «N «N «N III? N N N .N N N N N N N N N,- N quu.NNNN.NNNNN . . . N N d N N ..... NNN”: N N N m N N N NNNN.NNNNN NN.NN.N «NNNNHNN_ N N N N N N N NNNN. NNNN« NN«NN. N. NNNNN.NN N N N NNNN. NNNNm NN«NN. N. -NNNNN.NN NNNN. NNNN« NN«NN. N NNNNN NN. ” u N N N ..nl: r.aa. «NNN«.NN N N N N N N N «NNN. no ‘;|:- . N NNNN . u N w H N N ” N«NN. NNNNN ““muu.u. NN«NNHNN -sN -.N N N N N N NNNN. NNNNNN NNNNN. N. NNNNN.NN N N N . N 1.; N -11. N N NNNN.NNNNN NNNNN. N NN«NN.NN N N . N N N - N -- N NNNN.«NNNNN NNNNN.N. «NNNN.NN N N N N N N N NNNN NNNNN NNNNN.N NNNNN.NN N N N N N N N NNNN. NNNNNN NNNNN.N NNn«N.NN N N N N N N N iNNNN. .NNNNN NNNoN. N.i NNNNN.«N N N N N N N iNN«N. NNNNN NNNNN. N.l NNNNN.NN N N N N N N N «NNN. NNNNN NNNNN. N N««NN No N N N N N.. N11. NNNN. NNNNNN NNNNN. N NNNNN.NN N N N M N N N NNNN. NNNNN III, NNNNN. N. :11. «NNNN.NN N N N N N N N NNnN. «NNN. NNNNN. N NNNNNUNN N N N N N N N «NNN.N.NNN- NNNNN. N. NNNNN NN - N N N «NN« NNN«NN NNNNN. N. i. NN«NN. NN N - NNNN. NNNNN NNNNN. N NNNNN. No N u N N N - ul-azl- s;-- NNNNN. NN N N N N N N N NNNN.NN N N«NN . u N m M N N M N««N. NNNNN HMWMM.M. NNNNNHNN N- N N N N N N «NNN. NNNNNN NN.NN.N. NN«NN.NN N qu. N : N -nunu N N NNNN.NNNNN NNNNN.N NNNNN.NN " N N N N t N -zezzN -2 NNNN.«NNNNN NNNNN. N. NNNN« NN IIIN N N N N N «NNN «NNNN 11a NNNNN. N -..: NNNNN.NN N N N N N N N NN«N.NNN«NN NNNNN. N NNNN..NN N N N N N N Na: NNN«. NNNNN NNNNN. N. NNNNN.NN ..I1:|u .: “.;:r. N ;- N N N N N NNNN. NNNNNN NNNNN. .N..uIII-NNNNN.NN N It; N,l:-i: N .:a N - N N NNNN.N«NNN NNNNN. N N«NNN.NN N N N N N , N NNN« NNN«NN NNNoN. N NNNNN.NN N N N M N N N NNNN.NNNNN - NNNNN. N. .- NNNNN.NN N N N N N. N N N««N.«NN«NN NNNNN. N NN«N¢.NN N N N N N N NNNN.NNNNN iNNNoN. N. ««NNN.NN -.n|.al -:.:I N N N NNNN. NNNNNN- iNNNNN. N. «NN««. NN In. :||: -- ---- n- - N NNNN.«NNNN NNNNN. N N«NNN. NN “ n N N N a «NNNN NN N N N N N N N NNN«.N N NNNNN . 1- N N . N N N N «NNN.NNNNNN NNNNN. N. NNNNu. NNi ” N I:-N;:= ,N -.2I-e.N !:;-- N N NNNouNNoNNN NNNNN. N NNN«N. No} N N N N N ::u,a: N-u NNNN.NNNNNN NNNNN.N. NNnNN.NN N N N N N N N NNNN.««NNNN NnNNN. N NNNNN No N N N NNNN NNNN«N NNNNN. N NNNNNUNN - N «NNN.N««NNNIIIIIINNNNN. N. «NNNN No In: -Ilu::|||. N Isl ;. ”a ..111 N N N ;‘ NNNN«.Nn N- ..... N N N N . N ..:.-: N -. NNNN.NNNNNN . N N M M N m M NNNN.NNNNN -, Hummu.u. . NNNNNHNN u \N N N N N N NNNN.NNNNNN NNNNN. N NNNN«.NN N N N N N N N NNNN NNNNN NNNNN. N. NNNN«.NN N N N N N N NNNN. NN«NN NNNNN.N NNN«m Nn. N N N MflMMNMMNoNN NNNNN. N. ”"MMM.NM NNN . N -NNNNN N. -NNNNN. NNi «N «N «N «N «N «N «N «N «N «N «N «N «N «N «N «N «N «N 0. N«. xv ov av cv av IIIIIINN «N «N «N «N «N IIIIIwNun ,.«N :NN»«.NNNN NNNN.NNN NNNN.NNN NNN«.NNN NNNN.«NN NNNN.NNN NNNN.NNN NNNN.NNNN NNNN.NN«N NNNN.NNNN NNNN.NN«N NN«N.NNNN NNNN.NN«N N«NN.NNNN NNNN.NN«N NN«N.NN«N NNNN.NNNN NNNN.NNNN «NNN.NNNN NNNN.«N«N «NNN. onnN N«NN. «NNN ««NN. NNNN . Illul'l vii NNNN.NNNN “MNNHNNNN ..m.mn NNN . N NNNN NNNN.NNNN NNNN.«NNN NNNN.NNNN NNNN.NNNN N««N.NN«N NNNN.NNNN «NNN.NN«N ”Mm«HNNNN nnun.o~v« co mood N«oN.NNNN mmNNUN«NN N N NNN”.MMMM «oun.NNNN NNNN.NNNN N. NN. Non .NNNN N nNnn :muao. omsmt. N««N.NNNN MflNnnNN«N NNMM.Mmm“ NNNN. .. Nil] NNN .me« a on»« 1.741 ( ) o N o o N o a ann.o,Nonv oNcoo.¢ NNNNN.¢o on ooNv.NNNo c N o o N o o «Noe.omNoNn mNmom.ou NNNNN.No co ooNN.oonn . N o o N o a «NNNNNNNNN. cN.oo.c. onmoN..o om:uw.mn.oNNo o N N a N N e ooNn.occoon oNcoo.o Nommc.No co Nons.ooan o N a o o N N ovco.NcNNNN ooNvo.o ocomN.co o. .nmv.vaNn . N a a o N N NNN°.oNNooNIIIII moNvon. onwONJNo NanlnmNN.NNon o N o a o N N NONN.mNNaaN ooNcc.o. oNNNN.°o a. Nnon.o.oN o N N a o N N coom.NNoooN ooNco.o ocoNc.Nn a. NoNo.ooNN p N a a o N N mmNm.No.NmN |:I. oaNqa.o. .. .‘NnNoa.oN- - -.. .Noom.N.Nn [III] e N a N a N N NNNN.NNnoNN ooNvo.o mvNoc.mo o. .NNm.onNn o N. a a o N N ncno.NooooN ooNv°.° NNmoo.No o. Nooo.oooN . N a a p N N NnNansNNNFHIIIIIINNvahbw onNNoummullllluNnénmNoNanNN o N o a o .N 9:11 mooN.noNNo. 2:1- NNNNN.N oooNN.No ..... o. Nonn.Nomo o N o a o N o onen.noQNN. noNNm.o. NonmN.qo oo N0NN.NNN. o N e o o N o noao.cNNNon moNNn.o. oNoNN.No o. onmc.oonn . N n a a N o “NNN...nN.n|IIII moNNanezlnuu|n°ooNHamill i No: NNN...omu N N o N o o N ov¢n.Nooan-.lxua.oN°N.a.-e:-:-NNO...No -III e. NN.N.N°°. ,ztn: . N N o o o N OONC.NaNoon .omoN.o nonoN.No a. Nano.NNNn o N a a a o N o.n...n.o.. vomoN.e NNNN°.oo oo no.0.N.No N N N o o o N NNNNNNNNNNN -.. noapNuo. |:.|.NNNoNnN.-...I ..-.N.NN.oo.N.:ucnl a N a a a a o oo.¢.Nmnooo voan.N anomN.qo NoN nnNo.nn.NN . N o N o o o .ooN.oNNooN .°.NN.N. N.N.N.No NNN onoo.NoNNN o o N N N N N 0No°.oNNmo oNqno.o. NoONn.NN NN NNNN.N¢N a o .N .N N. N until. N.a- NNNc.Nonn¢ oN.no.o NNONN.Nn NNN No.N.nNN .:::|: o o N N N N N oNoN.Nano oNqn°.o NNNNN.NN NN o.m~.nwo . a N N N N N aNn...N.N. o~.n°.o. namNN.Nm NN no.0.No. c e N N N N N cho.oano: oN.no.o ‘chNm.ON -NN NNNo.o.o‘ - N o N N N N N Non..o.aq. oN.nN.o. mooma.om NN NNN°.No. a o N N N N N NNNm.oncvo NNNn=.o. NN.NN.NN NN NNN..om. ‘9 o N N ,xln-- N -xxlts N 11. N -- N.N¢...NN. oNon°.N ovamm.Nn‘. :; NN Nmoo.ooo 1:! a o N N N N N oooo.mnmo. oN.n°.o NNNon.No NN ooov.o.N . o N N N N N =NNN.Nocon 9N.no.o. onmoN.mm NN No.m.noo o o N N N N N Nomo.vNoo. oN.ao.o. monc.No NN-.N.NN.NNN c a N N . N N N nNoN.N.oon oN.no.o NoNNN.mm NN NNNo.N¢o c o N N N N N oce...ognc oN.no.o. No.o..No NN No.m.oNN o a N N N N: mnuu.NNoo..oNon,||. ..N.no.o NNNNN.mn :1 NN oooN.noo .:¢-1; a N N N N N N oNnN.NoNon oN.qo.N Nuccc.no NN .NNN..NN . o N N N N N «NNN.c.uon oN.no.o. oomNN.mm NN Noon.¢oo o o N N N N N NnoN.mNcoo oNNno.o NnoN°.cN NN NNno.ooo o o N N N N N NNN..oonn. oN.n°.o. NmoNN.Nn IIIIINN;iNomN.oNN -.ll. . N o N N N N N NnNo.NNNNN oN.NN.N. NNoNN..N NN NNN..NN. a .9 Ni N N VN N‘: no.0.onso. oN.nN.° onoNn.eo NN NoNo.NNN . a N N . N N N NNNN.NNNNN NNNNN.o. .Nmnn.oN NN .oNo.oNo o o N N N N N .Nno.oN9N. oNcn°.a ccoow.aa NN NoNo.NNN . e N N N N N Noon.NNONN oN.n°.o ”vooo.oN NN NNnn.mmo . N N N N N N onno.n.aa. oanc.o. ”NooN.oN NN NNNNNNNN o a N N N N N n.an.nNnoc oNqno.o. NoNoo.No NN NNno.oNN . . N N N N N «Nan...»Nc oN.na.. NocNu.Nn Nms:.NNNNnoma. ‘.'> "O D“- In '.'*I 1 753 .III:-|a - N ....... - N ;- N N N . N -;- N.-; NNNN.NNNNNN NNNNN.N. NNN.N..N N N N N N N N NNN¢.NNNN.N NNNNN.N. -- N.NNN.NN N N N N N N N NNNN.N.NNNN NNNNN.N .NNNN..N N N N N N n.N. ..Nl. NNN..NNNNNN 11a:.NNNNN.N. - NNNNN.NN N N N N N N N NNNN.NNNNNN NNNNN.N 'nu: NN«.N.NN N N N N N N N N.NN.NNN.NN NNNNN.N NNNNN.NN N N N N N -;I-, N -:.:1:.N.!: NN.N.N.NNNN . NNNNN.N. -.- .NNNN.NN N N N N N, N.. NilanNN..NNNNNNIIIaIsNNNNN.N1 - NNNNN.NN N N . N N N N N .NNN.NNNNNN NNNNN.N NNNNN.NN N N N N N N N NNNN.NNNNNN NNNNN.N NN.NN.NN III at--- N.-:--, N - N,:I:- N N N N NNNN.NNNNNN NNNNN.N. NNNNN.N. N N N N N N N NNNN.NNNNNN NNNNN.N NNNNN... N N. N N N N N NNNN.NNNNNN NNNNN.N. NN.NN.NN N N N N N. N N :I-NNNN.mmwNNN- - NNNNN.N11inau-NNN.N.NN N N N N N N N NN.N.NN.NNN NNNNN.N NNNNN.NN. N N N N N N N N.NN.NNNNN. - NNNNN.N. NNNNN... N N N N N N N NNNN.NNNNNN NNNNN.N NNNNN.NN N N N N N N N u.sN.NN.NNNNN.- NNNNNNN..xII:.N.NNN..N N N N N N N N NNNN.NNN.NN NNNNN.N. NNNNN.NN N N N N N N N NNNN.NNN.NN N.NNN.N. .NNNN... N N N N N N N .NNN.NNNNNN N..NN.N NNNNN.NN N N N N N N- mu.: NNNN.NNN.NN N.NNN.N, NNNNN.NN N N N N N N N NNNN..NNNN N.NNN.N. NNNNN.NN N N N N N N N NN.N.NNNNN N.NnN.N NmNNN.NN b N .N. N- N -:.N.:ealni-N1- N.NN.NNNNN:uznl-,N.NNN.N.n::u. .NNN..NN N N N N N N N NNNN..NNNN N.NNN.N. NNNNN.NN N N N N N N N N.NN..NNNN N.NNN.N NNNNN.NN N N N N N N N .NNNN.NNNNNN N.NNN.N .NNNN.NN N N N N N N N NNNN.NNNNNN N.NNN.N. NN.NN.NN N N N N N N N NNNN.NNNNNN N.NNN.N. .NNNN.NN N. ‘N. N I.N.e --N-,x-;,| N ....,;.N- NNNN.NNNNNN:;.:u.N.NNN.N .‘In.-NNNNN.NN N N N N N N N NNNN.NNNNN N.NNN.N. NNN.N.NN N N N N N N N N.NN.NNNNN N.NNN.N NNNNN.NN N N N N N N ..N ..NNN.NN.NN,-::|1.N.NNNNN- NNNNNNNN N N N N N N N NNNN.NNNNN N..NN.N. NNNNN.NN N N N N N N N NNNN..NNNNN NNN.N.N. .NNNN.NN N N N N N N N NNNN.N.NNNN NNN.N.N NNNNN... N N N N N N N NNNN.NNNN.N1:||:.NNN.N.N .NNN...NN N N N N N N N NNNN.NNNNNN NNN.N.N. .NNNN..N N N N N N N N NNNN.NNNNNN NNN.N.N .NNNN.NN N.;::u--;N.. N.- N --N..I:::, N .z.--» N- s-NNNN.NNNNNN NNN.N.N. NNNNN.NN N N N N N N N NNNN.NN.NNN NNN.N.N. NNN.N.NN N N N N N N N N.N...NNNNN NNN.N.N NNNNN.NN N N N N N N N NNNN.NNNNNN NNNNN.N. NNNNN.NN s..;1§:y.:-:z:- N :::u N i-.:- -N -11.: N-:n¢.::.N .:.t-: N -.- N-- NNNN.NNN.NN NNNNN.N N...N.NN N N N N N N N N.NN..NNNNN NNNNN.N NNNNN.NN N N N N N N N NNNN.NNN.NN NNNNN.N. N.N.N.NN N N N N N N N NNNN.NNNNNN NNNNN.N NNNNN.NN N N N N N N N NNNNNNNNNNN NNNNNNN. NNNNN.NN N N N N N N N .N.N.NNNNNN NNNNN.N. NNNNN.NN N N N N N N N NNN..NNNNNN NNNNN.N .NNNN.N. N. NNNN.NNNN N. NNNN.NNNN N. N.NN.N.NN N. .NNN.NNNN N" “MNNHNNNW N N. .NN N. NN.N.NNNN . N. NNNN.NNNN N. N.NN.NNNN N. NNNN.NNNN N. N.NN.NNNN N. NNNN.NNNN N. NNNN.NNNN :EII o. no....NNn N. NNNN.NNNN NN NNNN.NNNN NN .NNN..NNN -;!stN NNNN.NNNN No NNNN.NNNN . .N NNNN.NNNN .N NNN..NNMN .lru .N NNNN.NNNN .N .NNN.NN.N .N NNN..N..N .--...N NNNN.NNNN .N NNNN.NN.N .N NNNN.NNNN .N NNNN.NNNN- .N noNo..NNN .N n.NN..cNN ---..N NNNN.NNNN .N .NNN..N.N .N NNNN.NN.N .N N.NN.NN.N .N .NNN.NN.N N. NNNN.NNNN N. NNNN.NNNN a.N.1.NNNN.NNNN N. NNNN.NNNN N. NNNN.NNNN N. NNNN.NNNN N. NNNN..NNN N. NNNN.NNNN N. NNNN.NNNN N. .NNN.NNNN N. NNNN.NNNN N. NNNN.NNNN N. sNNNN.NNNN N. NNnN.NNNN :Nm 1NNN..NNNn ..- NNNN.NNNne ’0 3" l ( ) 1.763 "-«'OIO‘" N N M u w H N N NNNN. NNNNNN NNN.N. N N.N.m. .. .N NN.N. N.NN N N N N N N NNNN NNNNNN NNN.N. N N.NNN. N. .N .N.NN. NNNN N N N N N N .NNN.NN..NN NNN.N. N. NNNNN... .N NN. N. NNnN N N N N N N N NNNN.NNNNN NNN.N. N. NNNNN. N. .N NNNN. .NN.N N N N m N N N NNNNNNNNNN NNN.3 N NN..N. NN .N NN .N .NNNN N N N N N N N NNNN.NNNNN NNN.N. N NNNN- .NN .N NNNN. NN.N N N N N N N NN.N.NNNNN NNN.N. N. NNNNN.NN .N NNNN. NNNN N N N M N N N N.NN..NN.N NNN.N .N NNooN.No .N--NNNN. on.N N N N N N N N N.NN. NNNNN NNN.N. N. NNN.N.NN .N NNN.. NNNN N N N N H M N N.N.. NNNNN1111111NNN.N. .b1111111wnmom. mm .N iNN.N.NN.N N NNNN. NNNNN NNN.N. N NN.NN NN .N NNNN. NN.N111111 N N N N N N N NNNN..NN 11111 -1-!. 1111--- . -1111- NNN .NNNN. N N NN . N . n M u m N . N N NNNN..NNNN .oxNN. N. NMNNu.mo um “www.mmm“ N N N N N N N NNNN.NN¢NNN .ome. N. NNNN..NN .N NNN..NNNN N N N N N N NNNN. .NNNN .NNNN. N 1111 NNNNN.NN.11111.N;-NNN“.N.NN N N N N N N N NNNN. NNNNNN .NNNN. N. .N.NN.NN .N NNN..NNNN N N N m N N .N111-NNNN.NNNNN - .NNNN. N .NNNN.NN1.N NNNN.NNNN N N N N N N N NNNN.NNN.NN .NNNN.N NNNNN. N. 1.. .N NNNN.NNNN -- : N N N N N N NNNN.NN.NN .NNNN. N. NNNNN. NN .N NNNN....N N N N N N N -NNNN.NNNNNN- -1- .NNNN.N. NNNNN.NN .N NN-N.NNNN N N N N N N N .NNN.N.NNN .NNNN.N. 1NNNN..NN ..N NNMN..N.N .111: N N N N N N N NNNN.NNN.NN .NNNN.N NNN.N.NN .N N..N. NN N N N N N .N11 -N111-NNNN.N.NNN 11:1 .NNNN.N. NNNNN.NN .N NN.N.NN.N N N N N N N N NNNN.NNNNNN .NNNN.N .-.:-NNNNN.NN .. .N NNNN.MNNN -. N N u N N N N NNNN.NNNNNN .N.NN.N. NNNNN.NN .N N. N. mm- N N N M N N N NNNN.NNNNNN NNNNN.mu NNN.N. N. .N nnm.. NNN” N N N NNNN.NNNNNN .NNNN.N NNN.o NN .wm1.N.NN. NNNN N N N N. N .N -N.11-NNNN. 1111. . . 11 ;;1. -.,- NN.NNN . N . .. . u M u m N N N NNNN.N.NNNN “MnMN.M wwwww.nu u" "mm“.mmmm N N N N N N N .NNN. .NNNNNN 1NN..N. N NNNNN... N. NNN». NNNN N N N N N N N.NN. NNNNNN NN.N3 N. NNNNN.NN .1111N. ..NNN. NNNN 1111 N N N m N N N NNNN. NNNNNN NNNN3 N NNNNN..N N. NNNN. NNNN a N N; N m -N -N111 mmNm. ooNNoNl1- oneno. N. Nm..m .NN N. Nona. ..NN N N N N N N N nomN NnoNNNl on.nN. N.:-l NNNmo. .N N. .NN.N. NNN» N N N .NNN. NNNNNN NN.N3 N NNNNK. N. N. NNNN. NNNN N N N N N N N n . - . . : ..... 11 .--111- 11 NNN N.N.NN .111 NNN.N N 111 Nnoo. N . n w n m N N N NN.N.NNNNNN NNN.N.N. 1 MNNNN.MN-111 um “MM“.MmM“ 1111 N N N N N No.0.oNNNNN NNN.N.N. NNNNN.NN .N NNNN.NNNN N N u N N N N NNNN.NNNNN NNN.N.N .NNN..NN .N NN1N.NN.N N N N N N N N NNN..NNNNN NNN.N.N. NNNNN.NN1 .N..NNNN.NN.N N N N w N N N NNNN. NNN.N NNN.N.N NNNNN.NN .N NNNN.NNNN N N N N N N- N .NNN. .NNN.N .1.! NNN.3 N 11 NNNNN.NN .N NNNN.NN.N N N N N N N NNN.. NNNNN NNN.N. N. -..NNNNN.NN .-.1 .N NNNN.NNNN --1 1 N N N N N N N NNN». Nno.NN NNN.N. N. NmNNo.oo .N NNNN.NNNN N N N N N N NNNN.NNNNNN NNN.N. N ..Nmo.mo .N NNNN.NNNN N N N N N N N NNNN.NNNNNN NNN.N. N NNNNN.NN .N11NNNN..NNN N N N N N N NNNN.NNNNNN NNN.3 N. N.N.N.NN .N NNNN.NNNN N m N N N. N..- -N111 NN.N.NNNNN NNN.N.N NNNoo.NN .N NNNN.NN.N N N u M N N N ..N...N.NN . -. NN..N.N.. -. N.NN¢.NN . .N NNNN.NN.N -- N N N N N N N NNNN.NNNNN NNN.N.N. NNNN..NN .N NNN..NN.N N N N NNNN.NNNNN- NNN.N.N NNNNN.NN .N NNNN.NN.N N N .NNN..NNNNN NNNNN.N NNNN..N. N. NN.N.NNNN 11" '11111ll 1 7'7 111111 111-1- N 1 N 1:111N. -1 N --11 N N N N.NN.NNNNNN NNNN..N. NNNNN.NN N. .NNN.NNNN N N N 1111 N N N N N.NN.NNNNNN NNNN..N NNNNN.NN N. NNNN.NNNN N N N N N N N NNNN.NNN.NN NNNN..N NNNNN.NN N. NNNN...NN N N N N N N N NNNN.NNNNNN NNNN..N. NN.NN.NN 1 N. NNNN.NNNN 1 1 N N N N N ..11- N 111. N NNNN.NNNNNN NNNNN.N NNNNN.NN N. NNNN.N..N N N N N N N -11 N-1. NNNN.N.NNNN 1 NNNNN.N. - NNNNN.N N. NNNN.NNNN N N N N N N N NNNN.NNN.NN NNNNN.N. NNNNN.NN N. NNNN.N.NN N N N N N N N N.NN.NNNNNN NNNNN.N .NNNN.NN N. NN.N.NNNN N N . N N N N N NNNN.NNNNNN NNNNN.Nn111111N.NNN..N111111N. NNNN.NNNN 1 N N N N N N N NNNN.N.NNNN NNNNN.N NNNNN..N N. NNNN.NNNN 1 1 N.1111-1 N 1 -N11111. N - N 1 N N NNNN.NNNNNN NNNNN.N NNN...NN N. NNN..NNNN N N N N N N N NNNN.NNN.NN NNNNN.N. NNN.N.NN N. NNNN.NNNN N N N 1N N N N NNN..NNNN.N! -NNNNN.m1. 111-NNNNN.NN11111-NN NNNN.NNNN 11:11 N N N N N N N NNNN. NNNNN. NNNNN.N. NNNNN.NN NN N..N..NNN N 111 1. N N N-1-.1.- N 111.- N -- .1. N 1 NNNN.NNNNNN 11111 NNNNN.N. -1 NNNNN.NN NN NNNN.NN.N N N N N N N N NNNN.NNNNNN NNNNN.N N.NNN.NN NN NNNN.NNNN N N N N N N N NNNN.NNNNN NNNNN. N.;1111-NNNNN.NN -1.. NN NN.N.NNN .11111 N N N N N N N .NNN.NNNNN NNNNN.N NNNNN.NN NN .NN..NNN N 1 N 11N N N N - N111 NNNN.NNNNN -1111 NNNNN. N NNNN..NN NN NNNN.NNN N N N N N N N N.NN. NNNN. NNNNN. N. NNNNN.NN 1 -1 NN NNNN.NNN -- 1 N N N N N N N NNN..N..NN NNNNN. N NN.NN.NN NN NNNN.NNN N N N N N N N111.NNNN. NNNN. NNNNN.N. NNNNN.NN NN N.N..N.N N N N N N N N NNNN.NNNNN NNNNN.NN .NNNN.NN -111.NN NN.N.NNN -1111. N N N N N N N N.NN. NNNN. NNNNN.N NNNN..NN NN NNNN.N.N N N N ,N N 11.N.11 N111 NNNN..NNN. NNNNN. N NNNNN.NN NN NNNN.NNN N N N N N N N ooNN..noN. 1INNNNN. N.-INNNN..Nn NN NNNN.NNN 11 -1 N N N N N N N NN.N. NNNN. NNNNN. N. NNNN..NN NN NNNN.NNN N N N N N N N.11.N.NN.NNNN.1 NNNNN. N NNN...NN NN NNNN.NNN N N N N N N N N.NN..NN.. NNNNN. N. NN.NN.NN- . 1 NN NNNN.NNN-- .1 N N N N N N N NNNN.NNNN. NNNoN. N N.NNN.NN NN NNNN.NNN .N - N. -N. 1 N N 11.N - N.11.NNNN.NN.N. 1INNNNN. N NNNNN.NN NN NNNN.NNN N N N N N N N NNNN.NNNN. INNNNN. N. 1111.NNNNN.NN .11. NN NN.N.NNN - .- N N N N N N N ..N..NNNNN NNNNN. N NNNNN.NN NN NNNN.NNN N N N N N N M1111N.NN.NNNNN NNNNN. N. NNNN.. .N NN NNNN.NNN N N N N N N N N.NN.NNN.N NNnNN. N. NNNN..NN| !-NN- ..NN.N.N N N N N N N N NNNN.NNNNN NNNNN. N NNNNN. NN NN NNNN.NNN N -11. 1 N N - . N 1111.- N .11111 N 11111 N.11 NNNN.NNNNN -1. NNnNN.N. 1 NNN...NN NN NNNN.N.N N N N N N N N NN.N.NNNNN NNNNN.N NN.NN.NN NN NNNN.NNN N N N N N N N NNNN. NN.NN NNNNN. N N.NNN... NN NNN..NNN N N N N N N N NNNN. NNNNN NNNNN. N. NNNNN. moi NN NNNN. NNN N N N N N N N NNNN.NN.N. NNNNN. N. N.NNN. NNl1NN1,~NNN. NNN N N N N N N N NNNN. NNNN. NNNNN.N NNrN.. Na NN .NNN. NNN 111111111N -11-- N N.11 N 11.1 1 N. -1.1 N .:1.1 N 1 NN.N.N.NNN NNNNN.N NNNNN.NN NN .NN..N.N N N N N N N N NNNN.NNN.. NNNNN.N. - N.NNN.NN NN ..NN.NNN N N N N N N N NNNN. NNNN. NNNNN.N NNNNN.NN NN NNNN. N.N N N N N N N N NNNN. .NNN.. NNNNN.N. NNNNN.NN NN NNNN. .NN- N N N N N N N NN.N. NNNN. NNNNN. N. NNNNN.NN111111NN--NNNN. N.N- N N N N N N N NNNN.NNNN. NNNNN. N NNNN..NN NN NNNN.NNN N N N N N N N .NNN.NNNNNN NNN.N.N .NNNN.NN .N N.NN.NNNN N N N N N N N NNNN. NNNNNN NNN.N. N. NNN.N..N .N NNNN. NNNN N N N N N N N N.NmNNNNNNN NNdeqN. ..Nm..NN .M1 N.NN. N.N«111111 N N N N N N N NNNN.NNN.NN. NNN.N. N NNNNN. .N .N NNNN. N.NN N N N N N .m N N.NN.NNNNNN NNN.N.N. .NNNNN. NN -.N-1NNNN.N.NN ‘.-| I 101l1llv|l|11 15-1“! M Dc v-n-ouu. own-c4- 4 I|IIIQI'I'I v"! . \ii' fifth-11., LI: .. ‘1...” -.ll. uV!I ~u l'7é5 -( ) N N N N N N N NNNN.NNNNNN NNNNN.N NNN.N.NN NN NNNN.NNNN‘ . N N N N N N N NNNNNNNNNN NNNNN.N. NNomn.¢m .N .NNo.NN.N N N N N N N N NNNN.NNNNN NNNNN.N. conNN.mm .N NNNN.NNNN N N N N N N N NNNN..N..N NNNNN.N|. NNNNN..N .N NNnN.NNNN N N N N N N N NN.¢.N.N.NN NNNNN.N. = NNNNN.NN_::I! .N.-.NaN.NNNNnIIIII N N N N N N N NN.N.NN.NN NNNNN.N NNNNN.NN .N NNNN.NN.N N N N N N M N NNNN..NNNN NNNNN.N -NNNoN.mm rwN‘ N.NN.NNNN- N N N N N N N N.NN.NNNNN NNNNN.N. oNNmN.Nm .N NNNN.NNNN N N N N N N N NN.. NNNNNN NN..N.N. NNNNN NN .N NNNN .NNN N N N N N N N N.NN.NN«No mN..N.N Nann..No .N NNN..oo.N N N N o N N‘ N-lanNN.mNNNN|ln!mN..N.N ooNoNHmm .Iall,vN _NN«N.NNNN£II. N N N N N N N NN.N.NNNNN NN..N.N. .N.NN mm .N NNNN.NNNN N N, N N N N N NNNN.NNNNNN mN..N.N NNcoN.NN .N NNNN.NNNN N N N N N N willeNom.mn.ooN mN..N.N. NNNNN.No .N. .NNo.NN.N N N N N N N N 2.....2: rNooNNN. NNNmN..m .N NTRJNNN N N N N N N N 32.22: mN..N.N 3.36m .N NNNN.NN.N N N N N N. N N :a-oNNN.N..NNN [.11-NN..N.N.n:;:1|mNNNN..N -Iui .N N.NN..NNN 1:21 N N N N N N N 23.23: mN..N.N. 33...: .N .NNo.NNmN N N N o N o N 2:52; mN..N.Nu oooN..Nm 2.. 33.2: N N N o N ..N N nNoNJNéNoilll‘mN..NJolll!ooN.oN..No llvN‘--MNNN....N N N N N N N N NNN...NNN.N NN..N.N. No.oN.mN .N NNNN.NNNN N N N N N N N NNNN.N.NNNN NN..N.N mo.N..mo .N NNNN.NNNN N N N N N? il-l- N .m-;--.NoN.noN.N -::Ii mN..N.N i-;:s‘NmNo..Nm.,i£: .N NNNN.NNNN ,,iI: N N N N N N N NNNN.NNNNN NN..N.N. NNNNN.NN .N NNNN.N..N N N N N N N N NNNN.NNNNNN NNNNN.N. NNNNNJNN N. NNNN.NNNN N N N N N N N ..No.NNNNNN NNoNN.N .NN.N.NN N. NNNN.NNNN N N xm N N. N P.II.NNNN.N.NNNN ....... NNoNN.N 3:.1- Nom.n.om,-:-- N. NNNN..NNN ;I;;. N N N N N N N NNNN.NNNNNN NNNNN.N. N.NNN.NN N. NN.N.~.NN N N N N N N N NNN...NNNNN NNoNN.N NNNN..NN N. NNNN..NNN N N N N N N NIII.NN.N.NNNNNN :NNcNN.NN NNNNN.NNV -N. NNNNN.N.NN- N N N N N N N NNN..NNNNNN NNNNN.N. NNNoN.Nm N. NN.N.NNNN N N N N N N N NNNN..NNNNN NNoNN.N NNNNN.NN N. .NNN.NNNN N N N N N N N NN.N.NNNN.N .NNNN.N NNN.N..N .N NNNN.NNNN N N N N N N N ..NN.NNNNNN -..NNNN.N. NNNNN.NN : NN-.NNN..NNN~ N N N N N N N NNNN.NNNNN .NNNN.N. NNNNN.NN .N NNNN.NNNN N N N N N N N NNNN.NNNNN .NNNN.N NN.NN.NN .N NNNN.NN.N N N N N -N I .N N.|I:NNNN.NNNNNN.zxul..NNNN.NN,.II: NNNN..NN -;:'..N NNNN..NNN.IIII N N N N N N N 23:33 .NNNNJ .3373 .N NnNmJoNN N N N N N N N .NNN.NNN.N .NNNN.N NNm.N.Nm .N NNoN.NNNN N N N N N N N NNNN.N.NNNK- .NNNN.NJ .NNNO.NN .N NNN..NNNN. - N N N N N N N NN.N.NNN.NN .NNNN.N. NNNmm.NN .N NNNN.NNNN N N N N N N N NNNN.NNNNNN .NNNN.N NNNNN.NN .N NNNN.NNNN N _:N N N N N N NNNN.NNNNN !.: .NNNN.N _:a:.NNNNom.NN .N NNNN.NN.N - -- N N N N N N N N.NN.NNNNN .NNNN.N. NNN.N.NN .N NNNN.NN.N N N N N N N N .mom.NoNoNN .NNNN.N .NN.N.NN .N NNNN.NNNN N N N N N N N NNNN.NNNNN .NNNN.N. .NNNN.NN .N NNNN.oN.N N N N N N N N onn..mNNN .NNNN.Nm ooc~n.em «N.:.noo.NmNN N N N N N N N NNN..NNNNN .NNmN.N N.NNN.NN .N NNN»..N.N N N N N N N N NNNN. NNNNN NoNN..N NNNc..oN .«|;NNNN.NNNN N N N N N N N NNNN NNNNNN NNNN..N. NNNNN4MN N. .NNN.NNNN N N N N N N N NNNN.NNNNNN NoNN..N. N.NN..Nm N. NNNN.onnN N N N N N N N NNNN.NNNNNN N.NN..N NnoNo.No N. NNN..NNNN : . . ‘Ir pL‘.D>bDI‘§ V. xi. I ,Frvw aritrt 3,-4.I..lrsnlsv‘\ru ,N .- N] N {links in r .. a 1'1‘) ( ) N N N u m H M "I! ”.1 -H NMNNQBNOVOH NHCfiDQD o N N N N N N NNNN.NNNNN IIIII NNN N. I-III.NmNNN vo.II. NN NNN . N N a a a N m 9 vv N . : s onn e N - N o NNN vNoNo NNan.N r N ca oN NNNN. NN N N N N N ,N -N..- noNn.vcnch NNNNN.N. N.NNN.NN NN NNNN.MN N N N M M N N M ““uNHNNNNNN:.II..NNNNN.N.-.II: "w“MNHNN - .N NNNN.NmuN ..IIII- --- . NNNN N M N N N N I, n N. «N NNNN.NNNN N N . N N N .wIII.m.NN.NNNnoN . N N N . , .NNNNN N. . N N N NNN. NNNNNN . . NNNNN NN N . u .w M N I . -- .. ” “macs: Mam.“ mmnm II". mu.m3---1 N N N - N - III. N .. N NNNNNN NNNNN.N. N. .NNN..NNN N N N NNN..NN.NNN NNmNN.Nn N - m. N N N u N NNN..NNNNNN munmmun NNNNN.NN u" "mnmumuum N N N-.I.I:I N..:.II .u - NNNNHNNNNNN I. NNNNN.NH NNNNNHNN N. NNNN.N.NN , .I -- NNN» NNNNNN I. NNNNN. -- NNNN..NN N. NNNN.N N N -I:III-. N NNNNN on N N N M N 111111 N ---- . N ..I-- . N. .....NNNN - N I .- . . N N N N N N N NNNN.NNNNN - NNN . N . u N H u N N M ”MWMUMNMNN .N.MN.N. mwmmm.uo N“ NNNN.NNN - N N N N . N NN .N.NN.N. . N NN.N.NNN I I IIIIF I!“ O N N 00° . 'O'Nu O NO“ N [I'll .. (n 080 o N m NanN . o mN No NN . u M H u ,N M- 1.- ..M! REE... 1.! "am.” ”mm?“ NN “a“.m” N N N N N» NNNN. .N.NN.N IIII . m -.. NN NNNN.NNN N N N N N N NNNr.chmv NovNN. ooroN.on NN NN. . .-- I- o N N o N N NNNNN¢NNmn .cu mNcmc.on NN NN.vso fl fl N o . lllllllVOde on «a N anhv "#8 N N N N NNN. N..NN - . -IIIII.N N. No NN . N N N . NN.NN N . .zIII. NNNN NNN N N N ..11. N N NNN. ...NN . NN... NN NN . ...I:.. N .1 ......1111 I- N N wavwn .- ac N Coos Nos a u o u d 1.!!! I..- NNNV 026. we NNN v ow ..m Nd now N .... till}. . a. a 0.0 .N N N N N N N NNNN.NNNNN N. .N I... Nocmr cm NN . . - N N . NN.N NNN N N N- N N N NNNN.NN.N v NN NNNNN No N . . - N . .N N N v NNNNN.N. . N NN.N NNN c N o N N .. III vvoNNeNmon o . mecca cm NN mNN . N N N N N N NNNN.NN.N..I.:II-N .NN.N..I.II.NNNNN.NN N N.NNN .---I. .II N N . NNNNN N . .II N NNNN ..N N N -I .N .11. . N N NNNN NNNNN . NNNNm N» N .r -.III N -1: - --1. N N voewN on . N vnoN coo N N N N - - - NNNN.N.NNN . N.NNN NN NN . N N N . ---I..NNNNN N . NNNN NNN o o N “NNN want. . - csomN on N . . . N . .. . N . NNN.“ M. "NM... N N N N N NIII-NNNN.NNNNN N . . NNomN.NN NN NN.N.N N . . N . . N M 555.11."...“115551 .. 5."... o -N a I u N onoNNrovnv '0 N o vomoo as Inn a .N 1 Il|.!.. 1!!! NH e0 N vmm «no N N N N N N ....... N ..... - NNNN.NNNN. N. . NNNNN Nu NN NN . u m M n N M “ “muonNnNMm "Mum“.nu “"mmwnnm I - MN “www.muu -I- N N N N v oNomcIIIIII.voqNN.oo .c NN anno.s o N N N N «NNN.NNNNN I - -IIIIII:NnNoN on N . o a 1“ NI.I. .o -I, N H N ocNN.Nncom "“"MMHM “coma. w 1111-MNIIMMMN.NNN-- o u d o .11|I d --.! d ..... !- d1... homoNhnno' COVNuNOO omQOONCO an "ham.““M N N N N M NNNN.NNNN. NNNNN.N NNNNNHNN NN N.NN.NNN “nDO-dfiflv' COQNGNOI “chmofim Nd fihmHoOBO ” M N o N .-- ILIIIOII Ill!!! B on u“ “0“..“9” . -. N o 1.1. N N N- : N- -N NNNN.NNN.NN N . . N N N N N- m- --N--. NNNN.NNNNNN Nnmmu.N. NNNNN.NN .N NN... N N M u N N ” Nonn.vcmco I NNNNN." -- “NmNNHNo .N NNN..MM““ N N N NNNN.NNNNN . N on Na NN N . N G N a on N NNNNO 0. 000° 9 b no MOM” N N NN-NNN N N No N a n N N m N NNNN.NNmm«N “MMMWHN NNNNN.NN um "MMMHMNNM N N N NNNN.NNNN¢ .N. NoamN.NNIIIIIIuNII.N .. NNIIIIII N O Nnbhonfid'. NKGNO O. NNORNohfi 8 C” ””0 -.. NNNNNNN NNNNN.NN .N NNNN.NNNN - . CNI.o.oun.nnv.u|.!1I ." \a. .. “DO 1E3C) -( ) NN N«NN.«NN -IIIIINN--NN.N.NNN NN «NNN.NNN NN NNNN.NNN INN.N. NNN NN NNNN. «oN NN nomN. .coo NN- NNNN. NNN NN oomv. cos NN.ImNmN.NNN N noon. o«o NN omoo. omo NN o««o.ooo NN -NNNN.NNN NN NN«m.m«o NN.,NNNN.NNN NN NNNN.«NN NN NoNo.N«o NN NNNN.NNN NN NNNN.«NN NN onNn.noo NN- oopv.:«m NN NN«N.NNN NN NNun.N«N NN NNN«.NNN NN “NNN.NNN NN nomN.Nmo NN NNNN.NNN NN NNNN.NNN NN NoNn.NNN NN N«NN.NNN NN NNNN.oNN NN-.NNNN.NNNw «N NNoN.nnoN «N «NNN.NNNN «N m«No.«.«N «N N«N«.NNNN «N NNNN.NNNN «N NNom.Nm«N «N ««No.NNmN «N NNNN.NNNN «N N«NN.NNNN «N NNNN.NNNN «N NnNN.NN«N «N NNNN.NNNN- «N NNNN.NNNN «N No«o.no«N «N NNNN.NNNN «N NNNN.NN«N «N NNNN.NNNN «N NNNn.no«N «N NNNN.NN«N N N N N N N N NNNN.NNNNN NNNNN. N NNNNN.NN N N N N N N N NNNN.NNN«N NmoNN. N. NNNNN.NN N N N N N N N NNNN.NNNN« NmoNN. N. NNNmN.Nm N N N N N N N NN«N. .NNNm« NNNNN. N NNNNN.mm N N N N N N N NNNN. mNn«N NmonN. N. N«NN«.NN N N N N N N N NNNN. NN«NN NNNN1 N NNNNN.NN o N N N N N N NNNo.«m«on NmoNN. N wococ.mm N N N IN N N N NN«N.«NNN. NNNN1 N. NNNNN.NN N N N N N N N NN«N. NNNNN NNNNN. N. NNNNN.NN N N N N N N N NNNNJNNNN« Nnon1 N NMNNN1NN N N N N N N N NNNN.NNNNN NmoNN. N oNNoN mm o N N N N N N noo«.NnNN« NnonN. N. Nm«No.on N N N N N N N NNNN.NNNNN NmNNN.N -N«NN«.NN N N N N N N N NNNN N««N« Nmon1 N. NNNNN.NN N N. N N N N N NN«N. NNNNN Nncn1 N. NNNNN.NN N N N N N N N NNNN1«NNNN NNNN1 N N«NN1 Nm N N N N N N N NNNN.NNNNN NmNNN. N. NNNNN. No N N N N N N N «NNN.NNNNN NNNNN.N «NNNN. NN N N N N N N N NNN«.ono«« NmonN. N - nnom«.om: III N N N N N N N N«NN.NN«N« NmonN.N. «NNNN N N N N N N N NNNN.NoN«N chnN. N NNNNN.NN N N IN N N N N NNNNNNNNNNIIIIIIINNNNN.N.IIIII.NN.NN.NN IIIII N N N N N N N NNNN.NNNN« NmonN.N. NNNNN.Nn N N N N N N N NNN«.NNNN« NmNNN.N NmNmN.No N N N N N N I.NI Nm«N. NNNNN- .IINNNNN.N-IIIII.NNNNN.NN.:III N N N N N N N NNNN. oNnn« NmoNN.N. onmN.Nm N N N N N N N mNN«.NNoN« NmonN.N. «mNoN.om o N N N N N N NNNN.«NN«« NmoNN. N NNo«o.cm III; N N N N N N N NNNN.«NNNN NmonN. N.I NNNNN.NN N N N N N N N NNNN.N«NN« NNNN1 N Nowom.om N N N N N .N. .INIII «NNN. NNN,» .IIIIINNNNN. N IIII.-«NNNN.NN--III N N N N N N N NNNN. NNN«« NnonN. N. «N.No.cn N N N N N N N NNNN. NNNNNN NNNNN.N. N.NNN NN N N N N N N N NNNN. NN«NNN NNNNN.N NNNNN.NN N N N N N N N-II.NNNN.NNNNN-IIIII.NmNNN.N I NN«NN.Nm. .III. N N N N N N N N«NN.NNNNN omNNN.N. NNNNN No. N N N N N N N omN«.«NNomN omNNN.N NNNNN. NN N N N N N N N Nmnn.NnN«NN omNNN. N. ««oN .No N N N N N N N n«N«. «NNNN emu NN N. NNNNN. on N N N N N N N NNNN. NNN«N onmNN. N mNNNN. No N N N N N N -IIIII-N-I. NNN«.NNNNNN oNNNN.N...II- NN.NN.NN -I- N N N N N N N N«NN.N«NNN NNNNN.N. NNNNN.NN N N N N N N N NNNN.NNNNN NmNNN.N. NNNNN.mm N N N N p. -N N NNNN. NNNNN I. . NNNNN.N I-.II.NNNNN.NN N N N N N N N «NNN.N««NNN NNNNN.N. unmoN.NN N N N N N N N NNNN.«NNNN NNNNN. N NNN«N.Nn N N N N N N: N NNNN.NNNNN--I;II,omNNN. N - -- NNm«N.«n I N N N N N N N NNNN.NNNNN NNNNN. N. NNNNN.NN N N N N N N N NN«1 N.NNNN NNcmN N NNNNN.NN N N N N N N N NNNN.NNNNN NNNNN. N. NNNNN.NN N N N N N N N «NNN. NNNNN NNeNN.N.-IIII-NNNNN.Nn :I,I N N N N N N N NNNN.NNNNN NNONN.N «NNNN.NN N N N N N N N ««NN.NNNN»N NNNNN.N. moc««.NN N N N N N N N «NN«.NNNNN NNmnN.N NNNNN.MN - . N N N NI N N N N«NN.NNNNN NNcmN.N N«N«m.«NI_. . «NIINNNN.NNnN N N N N N N N NN«N.N«NNN NNNNN. N. NNNN«.NN N N N N N N N «NN«.«NNNNN NNNNN. N. NNNN«.NN illfll‘ II -..!I II «m--mnao.mmnw 11111«N. mmcq.sonms ..NIINNNNiNqI «N NNNN.NNNN 181 ( ) u. « N N N I -II; N II. IIIN II NNNN.NNNNN NNNNN.N NNNNN.NN N N N N N N «NNN .NNNN«IIIII-.NNNNN.N.I.III NNNNN.NN N N N N N N N NNNN.NnNon NoNNN.N. N«moN.nm N .m N N N IN- IIII IN - NNNN.«NNN« NNNNN.N NNNNN.NN N N w M M N N .I NNNN.NNNNN IIIII NNNNN.N..III NNNNN.NN N N N N N M “ NNNNNNNNN« NNNNNHN NNNNN.NN N N N .N N II: .N --IIIIINIII NNNN.N«MNN.I :II NNNNN.N .-. N«N«N.«N NNNN NN.N« NNNNN N. NNNNN.NN N I ..... . I:IIIIIII . II, N m w m N N N «NNN.NNN«N NNNNN.N NNNNN.NN N N N N m N N NNNNHNN«NN NNNNN.N. N«NNN.NN IIIIII N.. _N I N - N III . II- N I , N . N«NN NNNNN NNNNN.N. NNNNN.NN N N N N N NN«N.NNNN« NNNNN.N NNNNN.NN N N N N N N N NNNN.NNNNN NNNNN.N. NNNNN.NN N N N N N N N.III NNNN. mmno« NNNNN.N NNNNN.NN N N N N N N N «NN1 NNNNN NNNNN.N IIIII «NNNN.«N N N N N N N N NNNN. NNNN« NNNNN.N. NN«NN.NN IIIIII IIII- N III. - N- I.N :IIII N III N -I N N - NNNN.«NNNN NNNNN.N. «NNNN.NN N N N N N N N NNNN. NNNN« NNNNN.N NN«N«.NN N N M N N N N NNNN. NNNNN NNNNN.N «NNNN.«N N N N m N NIII NNNN. NNN«« NNNNN. N. NNNN«.NN N N N N N N NNNN.NNNNN- NNNNN. N -IIIIINNNNN.NN. I N N N N N N N «NN«.NNNN« NNNNN.N. m«NNN.No N I M I.“ M H N IINII IN«NN.NNNNN I NNNNN.N.- NNNNN.«N N N N N N N INN«N. NoNn« NNNNN.N INNNNN.NN .. N N N N N N N NN«N.«NNNN NNNNN.N. NNNNN.NN N N N N N N- N NNNN.«NNNN- .IINNNNN.N NNNNN.NN N N N N N N NNNN. NNNN« NNNNN.N NNNNN.NN N N N N N N NNNN. NNN«« NNNoN.N. NNN«N.Nm N -N IIII I M IIIIII M N IzIII N I II.I N I- NNNN.N««NN II II NNNNN.N NNNNN.NN N N N N N N N NNNN NNNNN NNNNN.N. III N«NNN.«o N N N N N N N ««NN.NNNN« NNNNN.N. NNNNN.NN N II N N -N- N N NIII NNNN.NNNN« INNNNN. N NNNNN.NN N N N N N NN«N.NNNNN INNNNN. N NNNN«.«N N N m N N N N «NNN.NNNNN NNNNN. N. NNNNN.NN N I . III I N--I:II: N IzI - N -IIIII N NNNN.NNNN« INNNNN. N. NNN«N.NN N N N N N N N ; NNNN.noco« II NNNNN. N IIII NN««m.No I N m N N N N N ««NN. NN«NN NNNNN. N. NNNNN.NN N II N N N N INII «NNN. NNN«N NNNNN. N NNNNN. no N m M M .N N N NNNN. NNNN«IIIIII NNNNN. N IIIIII NNNNN. on III. IIIII.I-I I I - N o N NN«n. NNNo« NNNoN. N. «mNoN. No u m M m N N N NNNN. NNNNNN NNNN1 N. NN«NN.NN N m N N o N ««NN. NNNNNNN NNNN1 N NNrnN.no N N N M N N N NNNN. NNNNN NNNN1 NI. NNN«N.NNI N N N N N N N NNNN. NNNNN NNNN1 N. NNNNN.NN IIN -IN N -IIII: N;-IIIII N - -I I N IIII-I,N.:-INNNN.NNNNNN NNNNN.N NNNNN.«N N N N N N N N NNNN. NNNNNN NNNNN.N. NNNNN.NN N N N N N N«NN. .N«NNN NNNNN.N. NNN«N.NN N N N N N N N N«NN. NNNNN NNNNN.NI NN«o«.Nm N N N N N N N NNNN .NN«NNN NNNN1 NI NN«NN.NN N N N N N N NN«N. NNNNNN NNNN1 N. NNNNN.NN I --III I IIII N ,III:, M .1.- - “I III m II " IIII- N II: I N NNNN. N«NNN NNNNN.N. NN«NN.NN N N N N N N N «NNN. NNNNN NNNNN.N NNNN«.NN N N N N N u N NN«N.NNNNNN NNNN1 N. «NNNN.«N N N N N N N u NMflN NNNNNN NNNNN. N “NNNN. no N N N N N N N N N.NNNNN NNNNN.N NN«NN. NN «NI- «NNN NNNNN NNNNN N. NNNNN.NN vu NNNN.«NN NNnN.NNN nmNN.Noo «NNN.noo NoNN.oNo NmNn.oNN NNNN.NNN N«NN.NNN “NNNHNNN .mfiu u . "MNM.Mwu NmNo”N«N “fig“ NNNN.N«N NN«o.onN n« . M«Nw.”uu N« . :33 N« . NNMM.MWM N«««.aNo «N««.NNN NNNN.«NN noNN.mNN moflnnnNN NMNm.MMM N«NN.NNN “Mucunoo o mm cNN mmNN.ooN mm. oNN «NN«.NNN N “MN“.MMM NoNN.NNN NNNN.NNNN NN«N. .mNmN NN .3 N N « NanN.NNNN NNNNHNNNN NNNN “mmNHN«NN NNmm.MMMM NNNN. N NN« mafia NNNo. wnnuI ooNo. nN«N 1E32 ( ) N N N N N N N NNNN. NNNN . -- N NNNNN N NNNNN.NN o NNN . M M H M M N N NNNo .NavnN NmNNN.N NNNNN.Nn o||:Non.mmm N N N N N N N NNNN.NNNNN NNNNN. N. NNNNN.NN o NNNN.N.N N N N N N m N .N.N.NN:NN NmoNN.N. NNNNN.NN N .Noo.NN. N N N N N IIN N.NN.NNNNN NN«NN.N In.---NcNNN.No N nuNNNo.NNv N N N N N N N NNNN.NNNNN NNNNN.N N:NN.om o NNNN.NNN N N N N N N N NNNN..NNNN NNNNN.N. -NNNNN. No 0 NNN N. NNN N N N N N N N NNNN.¢.NN. NNNNN.N NNNNN. NN N-- NNNN. 09. N N N N N N N .NNN NNNNN NNNNN.N. NNNNN. oo N NNNN. NNN N N N N N N N N.NN. .NNNNN NNNNN. N. NNNNN. Nm N NNNN.N.N N N N N N N N onoN. NNNvN NmNNN. N Navac. No N None. .Nn N N N N N N N NNNN. NNNN. NmNNN. N anvn. No o NNNN. NNN N N N N N w NI NoNN. NNNNN NmNNN. N. NNNNN.NN nail o .N0N.NNN N N N N N N NI N.NN NN.NN NNNNN.N. NNNNN.NN uNa,zNNNn.NmN -; N N N N N N N NNNN.NNN.N NNNNN. N N.NN..NN o NNco..NN N N N N N N N NNNN.NNNNN NmNNN.N. NNmoNJNN o .NNN.Nov N N N N N N N NNNN.NNNNN NNNNN.N NNNNN.NN m.:.Nm.m.NN. N N N N N N N Nooc.oNoNN NNN: .N mNmNN.Nm N N.NN.N.N N N N N N m MI N.NN .NNNN -NNNNN. N. NNNNN. No unauo NNNN. .NNN N N N N N N .NNo. NNNNN NNNNN. N NNNNN. No 0,- Nnmv. NNN - N N N N N N NNNN. NNNNN NNNNN. N. NNNNN. Nm N NNNN. NNN N N N N M M N NNNN. NoNNN NNNNN. N.I NNONN om N ..NN. NNn N N N N N N N NNNN. N.NNN NNNNN. N NNNmN. om NI;.NNNN. NNN :ziuu: N N N N N N N NNNN. N.NNN NNNNN.N. NNNN. on N NNNN. NNN N N N N N N NNNN. NNNNN NmNNN N--a;uluNNNNN. Nm o NNNN.NNN N N N N N N NNNN. .NNNNN NNNNN.N NNNNN .an It! o--A.NNv..NN.inII: N N N N m N N NNN.. .NNNNN NNNNN.N. NNNNo. Nm N NNNN.NNN N N N N N M N ...N. NNNNN NNNNN.N. NNNNN.NN o NNNN.NN. N N N N N N NNNN. NNNNN ! NNNNN.N I- N.NNN.Nm.. -NI: NNNN.NNN N N N N N N N NNNN.NNNNN NNNNN.N NNNNN.NN N N.NN.N.N N N N N N N NI oNoN.¢NNNN INmNNNJ. £330.... o noNoJmn N N N N N N N N.NN.NNNNN NNNNN.N NNNNN.NN :u:. N N.NN.NN. 1.1.- N N N N N N N .NNN.NNNNN NNNNN.N. NNNNN.NN N NNNN.NNN N N N N N N N NNNN.N.NNm NNNNNNN. NN«NN.om o NNNN.NNN N N NNNN NNNNN NNNNN.N NNNNN.NN N NNNN.NNN4 N .. :.:1. - , :11111. l-i|:,, 9-1:. N m M w M N N N.NanNNNNN NNNNN.N N.NNN.NN NN NNN..N.N :13:11 N N N N N N N NNNN.NNNNN NoNNN.N. N.NNN.NN NN NNNN.NNN N N N N N N N NMNNNNNNNN. NNNNN. m. NNNNNNNm NN .Nnoo.Noo N N N N N N N NNNN.NNNN. NNNNN N NNNNN.NN NN NNNN.NNNN - N N N N N N N NNNN.¢NNNN NNNNN. N. NNNNo.oN NN NNN..NNN N N N N N N. N11. NNNN NNNNN III: NNNNN.N 3.:1! .NNN..NN |::u NN NNNN.NNN N N N N N N NNNNHNNNN. NNNNN.N NNNNN.NN NN .N.N.NNN . N N N N m N N N.NN.NNNNN NNNNN.N. NNmnN.No NN NNNo.NNN N N N N N N N Nnoo NNN: N033? NnNanoN- NN 336; N N N N N N N NNNN.NNNNN NNNNN.N NNNNN.NN NN NNNNNNNN N N N N N N N NNNN. NNN». NNNNN.N NNNNN.Nm NN NN...NNN N N N N N NA Nils NNNN. NNan .--. NNNNN.N.- - vNNoo.No NN nvvn.NNN N N N N N N N NNNN. NNNNN NNNNN.N NNN.N.NN I: NN NNNN.NNN I-I - N N N N N N N NNNN. NNNNN NoNNN.N. .NNNN.NN NN NNNN.NNN N N N N N N N “NNN. NNNNN NNNNN.N. nmNNN.om NN .NNN.NNN N N N N N N NNNN NNooc NNNNN.N oNNNN.NN MNIINNNN.NNN- III. N N N N N N N NNN». N.N«N NNNNN.N. NNNNN.NN NN NNNN.N.N N N N N N N N NNNN. NNNN. NNNNN.N .mNNcN. Nm NN NNNN.NNN N N N N N N NI NNNN.NNNNN NNNNN.N -NNNNN. an --I- NN NNNN.NNN ..... N N N N M N N NNNN NNNNN NNNNN.N. NNNNN. Na NN NNNN.NNN N N N N N N N Nvmn.mmwmn NoNNN.N N.NNN. No NN NNNN. NNN N N N N N N N NNNNHNNNN. NNNNN.N. NNNN. on NN NNN NNN N N N N N N NNNN.NNNNN NoNNN.Nu coch. vn NN NNNN. on. N N N NNN. N.NN. NNNNN.N NNNNN.NN NN NNN». NNN 'I"‘ “'4. DO 0". \'. 1533 ( ) «mmdc.oc “ m N m m N M MMNNHNNNNN NNNNN.N N N N NN NNNNN N N N N N N N NNN..NNNNN NNNNN.N. N N N N N N N NNNN.NNNNN NNNNN.N m N N-; NM NI1;11: N.--III:.N-2: NNNN.NNNNN NNNNN.N. N N N N N N N NNNN.NNNNN NNNNN.N N N N N N N N NNNN.NNNNN NNNNN.N N N N N N N N NNNN.NNNNNI. NNNNN.pw N N N N N N N NNNN.NNNNN NNNNN.N. N N N N N N N NNNN.NNNNN NNNNN.N N .:..-- N ,inluu N ls-:.; N .--ugN N - ...... , .N -:::-- N -, NNNN.NNNNN ..N- NNNNN.N N N N N N N N NNNN.NNNNN NNNNN.N. N N N N N N N NNNN.NNNNN NNNNN.N N .N N N N N lNinu.NNNN.NNNNN- NNNNN.N. N N N N N N N NNNN.NNNNN NNNNN.N. N N N N N N N NNNN.NNNNN NNNNN.N -Ni::II:-N; N NysnluI-N ,l... N N;:n NNNN.NNNNN NmNNN.N. N N N N N N N NNNN.NNNNN NNNNN.N “ m H m N M N NNNNHNNNNN NmNNN.N N N NNNN NNNNN N N N N N N N NNNN.NNNNN NNNNN.N N N N N N N N NNNN.NNNNN NNNNNHN. N- ,N N N .N- --vzziln;a Naz..NNNN.NNNNN :Icz‘ NNNNN N. N N N N N N N NNNN.NNNNN NNNNN.N N N N N N N N NNNN.NNNNN NNNNNnN “ N m M M M w NNNonogqu nmNNN 01 N NNNN NNNN N N N N N N N NNNN.NNNNN NNNNN.N N N N N N N N NNMNJWNNNN NNNNN.N. NNNNN.NN NNNNNNN. ouwovNNN NNNNN.NN NNNNN.NN NNNmN.NN NNNNN.NN NNNNN.NN NNNNN.NN- NNNNN.NN NNNNN.NN NNNNN.NN NNNNN.om NNNNN.NN -NNNNN.NN NNNNN.NN 1‘ NNNNN.Nm NNNNN.NN NNNNN.NN NNNNN.NN NNNNN.NN‘ NNNNN.NN NNNNN.NN NNNNN.NN NNNNN.NN NNNNN.NN NNNNN.NN NNNNN.NN «NNNn.Nn NNanNNNt NmNNN.NN I N NNNN.NNN N ; NNNN.NNN N NNNN.NNN N NNNN.NNN N NNNNHNNN NNNN NNN N NNNNHNNN N Nara“ N NmNNHNNn N NN N.» N “Nam N. NNNN.Non‘ N NNNN.NNn N NNNN.NNN N NNNN.NNN N NNNN.NNN N N N N N N N N N N N N-- N N N N NNNN.NNN NNNN.NNn NNNN.NNN NNNN.NNN NNNN.NNN NNNN.NNN NNNN.NNN N INNNn.NNn NNNNNoNn NNNN.NNN NNNN.NNN ’U‘IO. A. '- .b- 184 ~( ) NNN.N NNNNN.N NNNNxmmN.NN N NNNNNNNN.NN -mp1 NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNNN NNNNN NNNNN.N NNNNNNNN.N N NNNmNNNN.N mum: NNN.N NNNmp.N NNNNNNNN.NN N NNNNNNNN.NM NNNi NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N Nu NNNNN NNNNN.N NNNNNNNN.NN N NNNNNNNN.NN mum: NNN.N NNNNmuN NNNNNNNNNNMNN N NNNNNNNNNNNNN NN: .NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NN NNNN... NNmqumwilllllltmeNNme.NNNNN N NNNnNNNN.mmmNN .wa NNN.N NNNNNNN NNNNNNNN.N .I-N NNNNNNNN.N NNNN NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNN NNN.N NNMNN.N NNNNNNNN.NN N NNNNNNNN.NN -NNN N»..N NNNNN.N.|u:u:::|sNNNNNNNN1N N NNNNNNNNNN Nu. NNN.N NNNNN.N NNNNNNNN.N ‘N NNNNNNNN.N NNN NN~NN1||IIIIIII.NNmrN.¢:IIIIIaII~NNNNNNN.NN N |!NNNNNNNN.NN N. NNN”pIIIIIIIIII.NNNNN.Nallanantl.NNNNNNNN.N N NNNNNNNN.N-- I:.:u:u-u-u..NN NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N N NNmNN NNNNN.N NNwNNNNN.N N NNNNNNNN.N§ NNN; :oNo.N-II:I nnnNN.N.IlaulltiuNmNNNNNN.N extnlanl.sl-zm.1|1.n-cyNNNNNNNN.N- is l -- .um NNN.N NNNNN.N NNNNNNNN.NN N NNNNNNNN.NN NN NmmNN NNNNN.N NNNNNNNN.N N NNNNNNNN.N u- NNNmNz|.; NNNNN.Nutzu-:z:|aNNNNNNNNunmgxlsulzlz;.:;N.;;;.;;‘:NNNNNNNN.NN :a 1:1. 1111111 K aNN NNNN.N. NNNNN.NN NNNNNNNN.NNNN N NNNNNNNN.NNNN N NNN.N . NNNNN.N NNNNNNNN.NNN N NNNNNNNN.NNN N .NNNmsgugNzwphuNmNNNNN NNNNNNNNN N mmmNNm-Nmmmazlluunq:.NNNNNNN;:::u,NNNNmeiulemeuau‘x 111111 .NNNNNNNNI--- NNNNNNNNNNNN .NNNNNN No .NNNN NN NNNNNN h~uoxt Nod Nx N. NNNNNNNN NzNNzNNNN NNNNNN NNNNNNJNNN 2N.) NNNNNNNNN NNN II Cl. — m N N N N m o z N N N N N N N N N a N N N z N muzouwn onNVQN mg~> pzwxcau pa<4 muzua aumc<4w NNNNNNNN NNN: NN . NNNN at.» szuxau NNNNNNzN NxNaNN u. NNN .1535 ( ) -- -N -- z‘- :. NNN.Nl:|u::I:1; NNNNN... NNNNNNNN.N :Ia.::n-.:;:.N.--- NNNNNNNN.N - .él;;.{ NNNNN NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N Nmua Nnoqq. NNNNN.N::1I1|III.NNNNNNNN.N- .N ---oNNNNoNN.N :sa...|i;g--+-. :.- umuN NNN.N|:a.||uu- NNNNN.N NNNNNNNN.N ,1..:_:;:: N - . - NNNNNNNN.N - NNN NNN.N NNNNN.N . ooNNNoNo.N N ooNNNNNN.N NNNN NNN.N NNNNN.N NNNNNwNN.NN N NNNNNNNN.«N NNN- .Iy'-nn!e--: -:.--- NNN.N-;I---. NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNN NNNN.N. NNNNN.NN NNNNNNNN.NNNN N NNNNNNNN.NNNN gm .NNm.N NNNNN.NIIIII;:.|.NNNNNNNN.NN N NNNNNNNN.NN -.n-:;;n.-.NNNN : --ullaz. -; i.{NNN.N-IIIana.;.,NnnNN.N ‘.:I. NNNNNNNN.N::iiaa:IIn!9-N:::|I. --NNNNNNNN.N -- - :1: -- noun NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNNN -NNN.N;, -‘NNNNN.NN ;.;NNNNNNNN.N.-,- unlz:NI|i.::s: NNNNNNNN.N -q..-- lllllllll NNN. .‘-l..on.;. - NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNNN NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNN NNNNN NNNNN.NIIIIIIIIIINNNNNNNN.N N NNNNNNNN.N :,;:i1:|; NNN 1.!as:,NNN.N.:11|:-¢.: NNNNN.N ,N NNNNNNNN.N ,- N NNNNNNNN.N .N NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNNN NNm.N NNNNN.mnlulzs-ausNNNNNNNN.N-- N .NNNNNNNN.N!;.;n|-:nn1;:..:‘n.w--:::.: NNN .s'n'NNN.pillsxn:ann:NNNNN.N.1xu-!.- NNNNNNNN.N -1. -..-N::-.III:NNNNNNNNN.N.::'.i--2:9 ..--:, NNN NNN.N NNNNN.N NNNNNNNN.N N NmNNNNNn.N No NNm.N NNNNN.Naunlnulc.NNNNNNNN.N N NNNNNNN~.N NNN. NNNupal. .NNNNNNN‘:- In: .NNNNNNNNN.NNzau-nI:I!n|. m.-u-.n : NNNNNNNN.NN.:sa-siulrt-;:;.|-; .1 - no NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NN .sNNNNN . NNNNN.N ---NNNNNNNN.NNNN N NNNNNNNN.NNNN .:- N- :; NNN.N zzzrnun:sl.NNNNN.N :;'i...- NNNNNNNN.N I: 1..-.on‘ N NNNNNNNN.N mNumN no... NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNN. N.N.N NNNNNNN NNNNNNNNNN N NNNNNNNN.N :NNNN- NNNHN .NNNNNNN::|::||nI:NNNNNNNNJv3|u|1::|:-.n; N|.u:.yl:1NNNNNNNN.N . ‘:--1 |.; . - NNN NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNNN NNNNN NNNNN.N NNNNNNNNNN N NNNNNNNN.N\ ‘meNI NNN.N NNNNN.N :mNNNmnmmwNN N NNNNNNNN.NN 'I O. O. u 186 ( ) 02 mm: ..... NN i‘em- mNmNNNN> NmnomNNu mN N awmz .N Nx wNmNNNN> NanmwNNu m. -NNNN NN Nx mNmNNoNN NNNNNNNu m. xxmu NNN .N Nmu:wNNNNNNN .NNNNNNN N. Nzw>2N .N Nx wNmNNNN> panamNNu a. mmzoame -xzmnxl-mmlumeNNNNN.>¢outhu m. pmszr .N Nx mgmNNNN; >woumNNo m. N N N .i:-!ll . #1:» N N N NNuwNum NNzNo azN azNNNNuNNNmz zwmzhmn mzoNNUNmmNzN NNN NNN; puwuum zNNNNuNNNwN NN macaw uzNzNszz NNN NNNNNNNN.NNNNNN NNNNN NNNNNNNN.NNN NNN NNNNNNNN.NNNNN NNNNN NNNNNNNNN NNN.N NNNNN.N NNNNNNNNNN N NNMMNNNN.N NNNNNN NNN.N NNNNNHN NNNNHNNN.N N NNNNNNNNNm . NNNNN- No... NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNNNN NNN.N NNNNN‘N NNNNNwNm.N \N NNNNNNNNNN umau NNNNN NNNwN.N NNNNNNNN.N N NNNNNNNNNN -- .uzmeNN: NNN.N NNNNN.N NNNNNNNN.N N NNNNNNNN.N NNNN NNNqN NNNNN.N NNNNNNNNNN N NNNNNNNN.N .qu: NNN.N NNNNN.N NNNNNNNN.NN N NNNNNNNN.NN NNN ‘ '01" W‘ h Wu