A CENTRAL- PLACE ANALYSJS OF SELECTED INDUSTRIAL MARKET STRUCTURES Thesis for the Degree of Ph. D. MICHIGAN STATE UNTVERSITY PAUL THEODORE NELSON 1970 LIBRARY NIiChigau St 2",: University This is to certify that the thesis entitled A CENTRAL—PLACE ANALYSIS OF SELECTED INDUSTRIAL MARKET STRUCTURES presented by PAUL THEODORE NELSON has been accepted towards fulfillment of the requirements for Ph.D. degree in Business Administration DateAApril 28. 1970 0-169 IINEING BY HUAE a SONS' BDUK amntnv mc. . usmav amozns "IMSNIT. ”cm“! .;~— ‘ I 5’ . Iii/IT fist-«31.1“? 11‘ mi. T r . 1. «2} ...L. 5‘. ‘ “Mien ‘. tuna-1.49:. ‘1’. ad y' 3 :5 . ,, ‘ r' lute: farm ,5“ m brush 'jL V\ "’é‘. v m m. qua“ " .‘ "1h .‘ . "- ‘ ' 1,}. \. .. . 4' 11-" T .Is , ‘e ' ‘3' 01' 56-1» 1; . . 1.. W‘ " .01. s. i ‘1"). , frat» H,» fir-.- ' J Jungian? .i a- ;. . 1*.38v’5z'iptjflt. vV ~.‘£,.-;'mv; army. 3} over exam-g ~‘ math! to -;3c:rfr:7rntt Hts“ Pet "crumt 3.1.va 7 ,. useful ”new-“ inr anatysm «- gm . “ I‘M GS'tl‘ 111‘? (A',." .L"- the Eitadtififl‘f ”‘ ' 1 21‘ m1 produces ABSTRACT A CENTRAL-PLACE ANALYSIS OF SELECTED INDUSTRIAL MARKET STRUCTURES BY Paul Theodore Nelson This research was undertaken to study the spatial dimension of marketing. Although a significant body of knowledge and theory of economic activity over space already exists, it has been developed almost entirely outside the field of Marketing. As a result, the con— cern has been primarily with the cost factors of loca- tion. Only recently has attention turned to revenue, or market factors. This study is concerned entirely with market factors. The basic question underlying the research is: What is the effect of location on market penetration? To answer this question one must have some basis for the locational, or spatial, analysis of markets. Central- place theory, from the field of Geography, offers a theoretical description of economic activity over space. This research seeks to determine whether central-place theory provides a useful framework for analysis of spatial relationships as they actually exist in the marketing of selected industrial products. Paul Theodore Nelson Specifically, this study is concerned with the market- ing of industrial products through industrial distributors. Three industrial products and the distributors through which these products are marketed form the basis for this research. The products and the related distributors are: (1) grinding wheels marketed through general-line industrial distributors, (2) ball bearings through ball bearing special- ist distributors, and (3) steel bars through steel service centers. The research is dual—focused in that it is con— cerned with both the sellers of these products and the industrial buyers. The scope of the study is restricted to a geographic area consisting of approximately the southern half of the state of Michigan. Three hypotheses drawn from central-place theory con- stitute the foundation for the research design. With regard to the industrial distributors it is hypothesized: (Hypothesis H1) that the measure of a distributor's estab- lishment is related to the size of the city in which the distributor is located; and, (Hypothesis H2) that the measure of a distributor's market territory is related to both the size of his city, and the measure of his establishment. These two hypotheses are tested using a rank correlation analysis of empirical data gathered through interviews with selected industrial distributors. Paul Theodore Nelson With regard to industrial buyers it is hypothesized: (Hypothesis H3) that industrial buyers' preferred sources of supply which are located at some city other than his own 3 will be located at a larger city. This hypothesis is tested 4 on the basis of data drawn from thirty-six industrial buyer interviews using a X 2 one-sample test of significance. Two primary conclusions are reached based on the II. analysis of the research results: I. It is concluded that the industrial buyer behavior is entirely consistent with central-place theory, and that the analysis of buyer behavior on the basis of central-place theory provides significant insight for practical decision making and for theory development. It is concluded that the spatial relationships of the three market structures are not consistent with central-place theory; or, that central-place theory does not provide a description of the spatial rela- tionships of these market structures as they actually exist over space. While this conclusion stands in direct conflict with the first conclusion, further analysis of the results of the market structure research on the basis of central-place theory yielded insight into the causes of the apparent lack of order expected under the theory. A CENTRAL-PLACE ANALYSIS OF L SELECTED INDUSTRIAL MARKET STRUCTURES BY Paul Theodore Nelson A THESIS Submitted to 4* y a“ ‘.A 4.! éfifia? ‘ ' Michigan State University jigs Mm partial fulfillment of the requirements ~. ' i‘EVif’Q L for the degree of i y t :L“ ' nocmon or PHILOSOPHY if3‘7 5" Department of Marketing and ‘ 'Qfia - ‘ Transportation AdministratiOn- ‘ 1970 ACKNOWLEDGMENTS This thesis is the combined result of many persons too numerous to acknowledge individually. However, a very special note of thanks is due the members of the Disserta— tion Committee. Dr. Donald J. Bowersox, Professor of Marketing and Transportation Administration, Michigan State University, as chairman of the committee devoted what must have seemed endless hours of reading, editing, and meeting. His gene- rosity in time and effort as well as the expert guidance he provided are sincerely appreciated. The other two members of the committee, Dr. Edward W. Smykay and Dr. Donald S. Henley, Professors of Marketing and Transportation Administration, Michigan State University, provided invaluable assistance and counsel during the prepa- ration of this manuscript. To all three, and to all the others who helped, my deepest thanks. The final acknowledgment is reserved for my family, my wife Sue, and our children Douglas, Cynthia, David, and Eric. It was their support, encouragement, and sacrifice that made it possible to reach this goal. ii A TABLE OF CONTENTS ‘ . mmDUCTION . I I o I o o o I a I o o I o I o a I C O 1 Chapter I. ’THE SPATIAL’DIMENSION OF MARKETING. . . . . 3 Location Theory: Historical Perspective Private Enterprise Over Space Marketing Industrial Marketing Focus of the Research Summary II. THE LOCATIONAL ANALYSIS OF MARKETS. . . . . 26 Economics The Locational Interdependence Approach The Market Area Approach Geography Central-Place Theory Regional Science Behavioral t Mathematical Marketing Summary 1;: THE RESEARCH DESIGN . . . . . . . . . . . . Scope of the Research y_‘ Industrial Market Structures ,. w~}_‘- Area of Inquiry ‘Wflisé' .." ’ ' - F. The Research Problem ‘5‘ , :r-_ , ‘" : 111 Chapter III. (Continued) Methodology Determination of Market Structures Determination of Hierarchy of Functions Hypotheses and Tests Collection of Data Buyer Data Method of Data Collection Summary IV. RESULTS OF THE RESEARCH . . . . . . . . . . lll Grinding Wheel Distributors Product Distribution Research Results Ball Bearing Distributors Product Distribution Research Results Steel Bars Product Distribution Research Results [ Industrial Buyers Summary V. ANALYSIS OF RESEARCH RESULTS. . . . . . . . 152 ‘ Hypothesis H1 D Product Specialty Scale Measure ' Composite Size Measure Hypothesis H2 H2a H2b Comparative Analysis of Distributors Grinding Wheel Distributors Ball Bearing Distributors Steel Distributors iv .(Cbntinued) Summary ‘gyi. CONCLUSIONS . . . . Primary Conclusions Buyer Behavior Market Structure Limitations of the Research and Suggestions for Future Research “ii-8T0? REFERENCES . . . . . v‘v ’W—v—‘v v — ‘3 _ Contributions of the Research The Industrial Buyer - Hypothesis H3 Discussion and Corollary Conclusions 170 195 197 YA Juli. In I I l LIST OF TABLES Table Page 1. Central-Places in Research Area . . . . . 83 2. Product Specialty Scale . . . . . . . . . 92 3. Calculation of Hierarchy of Function. . . 94 ; 4. Product Frequency . . . . . . . . . . . . 97 5. Product Lines of General—Line Industrial Distributors . . . . . . . . 116 6. Calculation of Hierarchy of Function For Grinding Wheel Distributors . . . . 120 7. Market Territory Summary of Grinding Wheel Distributors . . . . . . 124 8. Location of Principal Outside Competition Grinding Wheel Distributors . . . . . . 126 9. Additional Product Measure of Steel Bar Distributors. . . . . . . . . . . . 138 10. Market Territory Summary of Steel Bar Distributors. . . . . . . . . . . . 144 11. Location of Principal Outside Competition Steel Bar Distributors. . . . . . . . . 145 3 12. Location of Alternate Outside Source For Steel Bars. . . . . . . . . . . . . 146 1 13. Summary of Buyer Data on Control Products 150 Exhibit Appendix II. General—Line Industrial Distributors (Grinding Wheels) Establishment Data. . 199 III. General-Line Industrial Distributors (Grinding Wheels) Market Territory Data 200 vi Table XI. XII. XIV. XVI. Industrial Place Rank . . . . . . . . . . Example of Calculation of Spearman (rs) Rank Correlation Coefficient. . . . . . Steel Bar Distributors Establishment Data Steel Bar Distributors Market Territory Data. . . . . . . . . . . . . Product Groups Selected by Industrial Buyers . . . . . . . . . . . Test of Hypothesis H3 . . . . . . . . . . Location of Industrial Buyers' Preferred Sources for Grinding Wheels . . . . . . Location of Industrial Buyers' Preferred Sources for Ball Bearings . . . . . . . Location of Industrial Buyers' Preferred Sources for Steel Bars. . . . . . . . . Page 201 202 205 206 208 209 210 212 214 LIST OF FIGURES Figure Page 1 1. Hotelling Model . . . . . . . . . . . . . 31 2. Smithies Model. . . . . . . . . . . . . . 35 3. Effect of Freight Rate. . . . . . . . . . 36 4. Variation in Marginal Cost. . . . . . . . 37 5. Devletoglou Model . . . . . . . . . . . . 42 6. Hoover Models . . . . . . . . . . . . . . 45 7. Losch Models. . . . . . . . . . . . . . . 49 8. Christaller Model . . . . . . . . . . . . 53 9. Distribution of Market Potential — Harris 61 . 10. Shopping Preferences of Rural Residents . 67 ; 11. Probability Contours. . . . . . . . . . . 71 ' 12. Combined Probability Contours . . . . . . 71 ' 13. Research Area . . . . . . . . . . . . . . 79 14. Christaller Hexagonal Network . . . . . . 81 E 15. Abstraction of Research Area. . . . . . . 84 . 16. Second-Order Boundaries . . . . . . . . . 85 E 17. Third-Order Boundaries. . . . . . . . . . 85 ’ 18. Shopping Preferences for Clothing . . . . 105 19. Sales Territories: General-Line Industrial Distributors . . . . . . . . 125 viii Page Sales Territories: Steel Bar Distributors . . . . . . . . . . . . . . 143 21. Purchasing Agents' Preferred Sources: All Industrial Distributors. . . . . . . 149 Exhibit Appendix I. General-Line Industrial Distributor Questionnaire. . . . . . . . 197 VI. Steel Bar Distributor Questionnaire. . . . 203 IX. Industrial Buyer Questionnaire . . . . . . 207 XIII. Purchasing Agents' Preferred Sources: Grinding Wheels. . . . . . . . . . . . . 211 XV. Purchasing Agents' Preferred Sources: Ball Bearings. . . . . . . . . . . . . . 213 XVII. Purchasing Agents' Preferred Sources: Steel Bars . . . . . . . . . . . . . . . 215 ix INTRODUCTION Economic activity is carried on over space. A con- siderable body of knowledge has been developed regarding the economic costs of overcoming the friction of space and the effects of those costs on economic activity. Whether a particular activity is viable, however, depends on both cost and revenue. This study is concerned solely with revenue, or market factors. Of critical importance to both the marketing student and practitioner is the question: What is the effect of relative location on market penetration? This question is the foundation for this dissertation. In order to begin to consider this very complex question it is necessary to develop some understanding of the order and the regularities in the spatial relationships of marketing. This study is a search for that order. The research is founded on the premise that the concep- tual framework provided by central—place theory offers par- ticular promise as a basis for the locational, or spatial, analysis of markets. The objective of the research is to determine through the analysis of empirical data drawn from three selected market structures whether central-place theory does, in fact, provide a useful basis for the spatial analysis of existing market structures. i 2 “frscope of the research is limited to a consideration '. marketing of industrial products through industrial 3‘ibutors. Geographically the study is restricted to ;AState of Michigan. The first two chapters, Chapter I and Chapter II, Ifghesent the background and perspective for the study. ‘_¢' Chapter III is a detailed description of the research 'design. The remaining chapters, Chapter IV, Chapter V, apd Chapter VI, contain the results, analysis and conclu- 3 sions of the research. 02-. "i a .3 CHAPTER I THE SPATIAL DIMENSION OF MARKETING This study is concerned with marketing. More specifi- cally it is concerned with marketing over space. It is a study of the geography of marketing. Unfortunately the term marketing is plagued by a great diversity of opinion and uncertainty as to just what it means. Even within the aca- demic field of study called marketing one would be hard pressed to find two people who would agree precisely on one definition. It is almost as if marketing were in the eye of the beholder. This is not intended as a criticism or a plea for the correction of a weakness. Indeed, it may be a strength. The point is, however, that before proceeding the reader should have some idea of what is in the eye of this beholder. Therefore, the definition — my definition — is that marketing is the satisfaction of needs through markets. A market exists when there is a seller and a buyer with choice. The choice is usually among alternative sellers but the definition holds even if that choice is either buy or not buy. Thus any discernible entity, a person, a "firm", or a state engaged in economic activity which meets this criteria is engaged in marketing. Some of the relevant 3 4 implications of this definition will be discussed in later sections of this study. For the moment, however, the important thing is to note that, as used here, marketing is defined very broadly. One aspect of marketing which has received precious little attention is the study of its spatial dimension. Marketing activity takes place over space. The importance of this fact has long been recognized and has been the sub- ject of considerable study and theory development by econo- mists and geographers. The focus of their study of economic activity is, however, quite different from that of the mar— keting student. The importance of space in marketing has received very little recognition or attention. Marketing, of course, is a very new academic discipline and as such has been characterized by many fragmented approaches to its study. As a result, the small amount of attention to space, or location, has been highly particularized. For example, the two notable areas within marketing where the importance of location has been recognized are retail store location and distribution center location. Because they have grown out of very practical problem solving, the intent generally has not been to integrate or generalize for theory develop- ment. In one sense this study suffers from the same failing since it is focused on the marketing of industrial products. In another sense, however, the basic question underlying this study and its attendant research is generalized to all 5 marketing. One of the primary objectives of this disserta- tion is to call for attention to the importance of location to all of marketing, if you will, to marketing geography. Location is directly related to both cost and revenue of economic activity. Most of the existing location theory and location analysis is based on cost factors - transfer costs, production costs, and agglomeration cost factors - as they vary over space. It is only recently that attention has turned to revenue, or market factors. The great preponderance of this work has been done by economists and their contributions have, indeed, been significant. Perhaps of greatest signi- ficance of all is that they have recognized the importance of the spatial dimension of the market ahd have devoted their attention to it. The economist, however, by imposing his microeconomic model of pure price competition yields descriptions of spatial equilibrium which are unsatisfactory from a market- ing point of view. The reality of imperfect competition and, in fact, oligopolistic competition over space is of extreme importance in marketing and cannot be ignored. This kind of competition is different from pure price com- petition in two ways. First, price is but one among many of the factors which determine the quantity of a good demanded. Equally important are factors such as product quality, promotion, packaging and location. Second, this kind of competition is far from impersonal but is instead characterized by rivalry among competitors. Thus, the question: When rivals are competing for the same market, what is the effect of their locations on their market penetration? Or, restating the question from another perspective, is the purchase decision of a buyer related to the relative locations of his possible suppliers? This is the underlying question with which this dissertation is concerned. It is in this sense, as noted above, that this study is generalized to all marketing. The question can be applied to the locations of retail stores, distribution centers, or production locations. The research, however, is focused on the marketing of industrial products. This restriction of focus is for two reasons. First, this allows the opportunity to abstract to a considerable degree from price and product. Homogeneous products are most prevalent in industrial marketing. Furthermore, a number of these products and industries are characterized by an oligopolistic structure whose few mem- bers attempt to "meet competitor's prices." Second, the market area for an industrial marketing facility as contrasted with that of a consumer marketing facility usually involves a much broader expanse of space and greater distances. 7 The combination of these two reasons therefore yields the following model for inquiry: There is a broad geograph- ical market area in which each of the buyers of a particular industrial product face sets of "possible" suppliers. These "possible" suppliers are rivals competing for their share of this market. The product is relatively homogeneous, and a number of different suppliers are able to meet the buyer's requirements. The rivals compete; price does not preclude choice. One final condition for this model has been implied in the previous discussion but should be made explicit at this point. That condition is that the study involves a private enterprise economy. This point will be discussed in a later section of this chapter. To put the problem in a managerial perspective, con- sider the case of a firm contemplating entering a new market. TWO questions must be answered by this firm. First, will the market support a new entrant? In other words, is there a reasonable likelihood that entry into this market will be profitable? And, the second question is: Where should the facility be located? If market penetration (or market share, or drawing power) is not related to location, then the questions can be answered in sequence. It is the feeling of this author that too often firms first determine whether there is suffi- Cient demand to support a facility and then, on the basis 01? cost, determine its location. The thesis here is that 8 the two questions are not sequential if market penetration varies with location. If so, then each feasible location must be analyzed on the basis of both cost and market factors. At this point, one further clarification, or definition, is necessary for viewing the research question. When asking whether location is related to market penetration, the word "location" in fact means relative location. When one is considering the location of a facility, he is considering that location relative to the locations of his competitors, the locations of his other facilities, and the locations of all buyers in that market. Likewise, when the focus is on a particular buyer, the consideration is with the location of that buyer relative to the locations of all possible suppliers. Location Theory: Historical Perspective Intelligent inquiry into any question should begin with an appraisal of the existing situation. This is more than an audit of what we have right now. Instead it is more like asking both where are we and how did we get here. It is for this reason that this section offers an historical perspective. Essentially this is a review of the development of location theory. But, this is not intended to serve as a review of the literature. Rather, the intent is to 9 concentrate on the various thrusts or vectors of development which have occurred in the broad area of location theory. Such an approach must necessarily involve reference to the literature but only as it is incidental to creating an overall perspective of the various directions of development. For those desiring a thorough founding in the litera- ture they are referred as a start to Losch,1 Isard,2 and Greenhut.3 References more relevant to marketing may be found in the extensive bibliography contained in the doc- toral dissertation of Donald J. Bowersox4 and in the recent resource bibliography by Revzan.5 In attempting to develop an historical perspective of location theory, attention first turns to its start. In all of the literature on the subject, two names stand out as the 1August Losch, The Economics of Location, translated from the second revised edition by William H. Woglom with the assistance of Wolfgang F. Stolper (New Haven: Yale University Press, 1954). 2Walter Isard, Location and Space Economy (New York: John Wiley and Sons, 1956). 3Melvin L. Greenhut, Plant Location in Theory and Practice (Chapel Hill: The University Of North Carolina Press, 1956). 4Donald J. Bowersox, "Evaluation of Alternative Solutions to the Food Distribution Center Location Problem," unpublished Ph.D. dissertation (East Lansing: Michigan State University, 1960). 5David A. Revzan, A Geo ra h of Marketin : Resource Bibliography (Berkeley: Institute of Eusiness and Economic Research, University of California, 1968). 10 notable early contributors. Their work is often referred to as the "classical" location theory. The first is that of Johann Heinrich von Thunen, and the other, Alfred Weber. Von Thunen's famous work, The Isolated State,6 was published 7 in 1826. Weber's Theory of the Location of Industries was published eighty-three years later in 1909. Before going into a consideration of this "classical" theory it seems appropriate to note that the Theory of Comparative Advantage should really be included as a part of the body of location theory. Certainly John Stuart Mill's8 famous example of the comparative advantages of producing corn and cloth in England and Poland is evidence of its relevance. Furthermore, in writing about von Thunen's work, Michael Chisholm9 notes that both Ricardo and von Thunen introduced the concept of Economic Rent quite 6J. H. von Thunen, Die isolierte Staat in Beziehung auf Landwirtschaft und NEEiOnal 6kon6mie, Rostock, 1926. (An English translatiOn appeared in 1966 as von Thunen's Isolated State, edited by Peter Hall.) 7Alfred Weber, Uber den Standort der Industrien, Erster Teil, Reine Theorie des Standorts (Tubingen: Mohr, 1909). This book has been translated By C. J. Friedrich as Alfredieber's Theory of the Location of Industries (Chicago: University of ChiCago Press, 1928). 8John Stuart Mill, Princi les of Political Economy with Some of Their Applications to SociEI’Philosophy, Book III: ExEHange, COIlected Works, ed. J. M. Robson (Toronto: University of Toronto Press, 1965), pp. 589-90. 9Michael Chisholm, "Johann Heinrich vOn Thunen," Readings in Economic Geography: The Location of Economic Activity, ed. Robert H. T. Smith, Edward J. Tsaffe, Leslie J. King (Chicago: Rand McNally and Co., 1968). - 1-— fi’ --...- 11 independently of each other even though Ricardo's work was done a few years earlier. For Ricardo the nature of Econo— mic Rent depended on differences in soil fertility, whereas for von Thunen it depended on differences in location. Returning to the "classical" theorists, it is a note- worthy curiosity that both von Thunen and Weber were Germans. Added to this is the fact that August Losch, probably the greatest general theorist in the area, was also a German. Although the language barrier was probably a contributing factor, speculations on the reasons for this curiosity will be left to others. The fact that the two, von Thunen and Weber, stand as the classical contributors is provident in that they repre— sent two quite different approaches to the general location problem, and further, in that these two approaches are dominant in the later development of theory. Von Thunen's approach, very simply stated, was to take the location as given, and then determine the type of prod- uction which should be carried out there. Weber on the other hand began by taking a particular type of production given and sought a location for it. This in itself is not enough to make the approaches different. In fact, at this point one would have to say that the two are opposite but mirror images of each other. In addition both are based entirely on cost factors. The primary variable in the Weber problem is transfer cost; and, in von Thunen's problem the variable is Economic Rents, 12 which in turn depend on transfer costs. The real difference in the two approaches is due to two other factors. One factor is related to the difference in the basic assumptions. Von Thunen assumed a homogeneous land plane whereas Weber assumed the opposite. The second factor concerns the basic difference in the type of economic activity each was concerned with. Von Thunen was talking about farming and economic land use. For him the problem was determining what kind of use broad expanses of land should be put to achieve maximum economic rent. Weber was r’ essentially concerned with determining the location of an industrial activity in the face of uniquely (or unevenly; i.e., not a homogeneous plane) located raw materials and markets. The problems were, thus, entirely different. It should be noted that von Thunen's work was done before the start of the Industrial Revolution and at that time it was natural that agriculture be of prime concern. Weber, however, was right in the midst of booming industrial development. As indicated above, the two problems were entirely different; but, both were concerned with the location of economic activity. Von Thunen's approach was a macro- approach, considering the whole of "the isolated state." Weber's was a micro—approach, considering the factors influencing the location of a single firm, or producing facility. These two approaches constitute the major thrusts in the development of Location Theory after 13 von Thunen and Weber, and they provide a useful basis of classification. Another basis of classification is the field in which the contributor worked. Location Theory, until very recently, has been almost exclusively the province of the economists and the geographers. At first glance, an easy distinction to make; but, as soon as some of the individuals are studied the distinction gets very fuzzy. The existence of a field of study called Economic Geography along with its associated books and journals, has provided an arena in which men from both fields comfortably fit. Therefore, the distinction must be made on the basis not of the man, but the orientation of the discipline and the differences in the nature of inquiry. Geography until recently has actually had a retarding influence on the development of Location Theory. In spite of the fact that the orientation of geography must by defi- nition include the study of the location of economic activi— ty, the nature of inquiry which characterized the field was of little value. Although harsh, it seems fair to say that geography for too long was oriented to the uniqueness of each region, area, and location, and the need to describe it. Fortunately, this has changed and the work being done is truly exciting and promises important theory development. One notable example of this new thrust is the Central-Place Theory which is discussed in detail in Chapter II, and in fact, provides the conceptual framework for the empirical 14 research in this study. Because of the fact that the micro-approach is focused on the "firm," geography by its very nature will usually be found under the macro-study classification. The direction of the study of economics has certainly not been heavily oriented toward location or space. In fact, just as in marketing, recognition of the importance of the spatial dimension is lacking, even now. Yet, the credit for the beginnings and the very major part of the development of Location Theory must go to the few economists who sought to determine the relevance of space to economic activity. As already indicated in the discussion of von Thunen and Weber, who must both be classified as economists,_a’ the nature of economic inquiry accommodates both the micro— and macro-approach. Coming out of the "classical" period of location theory the thrust was almost singly Weberian, a cost-oriented theory of the firm. There were two serious shortcomings which were recognized in the late 19203. One was the realization that cost alone was not enough; that the rational economic man attempts to maximize profit which involves both cost and revenue. The other shortcoming was a failure to cope with locational interdependence. Weber had begun to attack this on the cost side in his consideration of agglom- erating factors, but even more critical was the interdepen- 10 dence side of revenue. Harold Hotelling was the first to loHarold Hotelling, "Stability in Competition," Economic Journal, XXXIX (1929) 41-57. 15 attack both with his "ice cream vendor" model. This was really a pioneering effort in market analysis and is also more fully discussed in Chapter II. The point of mentioning this here is that his article appeared just at a time when micro-economic theory was undergoing a dramatic change in recognizing the existence of imperfect competition. The 11 and Robinson12 best known works are those of Chamberlin but it was Hotelling who earlier demonstrated that the mere addition of a spatial dimension to economic theory yields a form of monopolistic competition. From this point in time on (about 1930), the impression is one of many vectors of effort and development. 0n the micro-side we find one principal vector highlighted by the name of Greenhut.l3 His methodology is to apply spatial variables to the production functions and resultant cost curves. Greenhut's excellent work is probably the most relevant and significant to marketing of all, and yet, it is because of the inadequacy of the "pure price competition" model that this dissertation is being written. Another set of vectors starts on the micro-side and crosses over to the macro—side. One is that of Losch who 11E. H. Chamberlin, The Theory of Monopolistic Com etition (Cambridge, Massachusetts: Harvard University Press, 1933). 12Joan Robinson, The Economics of Imperfect Competition (London: Macmillan & Co., Ltd., 1933). 13Melvin L. Greenhut, o . cit. Also see Melvin L. Greenhut, Microeconomics and the Space Economy (Chicago: Scott, Foresman, 1963). 16 started with the micro-economic demand cone and proceeded to develop from it a very rigorous, highly mathematical general theory, a macro-theory of the economics of location. Another member of this same set is Edgar M. Hoover.l4 Starting with Weber's assumption of linear transfer costs he considered the effect of the actual situation of non-linearity and then moved on to a von-Thunen-type exposition of economic land use. He concluded by posing the problems of regional conflict in economic development. Somewhat in line with Hoover is Walter Isard.15 He, too, started with Weber by suggesting what he called substi- tution analysis. Although a more thorough consideration of cost factors it is essentially a heuristic approach to the Weberian method. From here, however, he departed on a macro-basis by employing the dynamics of input—output analysis. Not stopping here, Isard has gone on to become a bedrock of the Regional Science Association. Finally, the other major vector is the macro-thrust of the new geographers. Discussion of this thrust is deferred to Chapter II. Whereas, the micro—economic thrust is of most immediate relevance and significance to marketing, the new geography seems to offer the greatest future promise 14Edgar M. Hoover, Location of Economic Activity (New York: McGraw-Hill, 1948). 15Isard, o . cit. Also see Walter Isard, et al., Methods of RegionaI Analysis: An Introduction to Regional Science (New York: thn Wiley and Sons, Inc., 1960). 17 for marketing, if you will, for the foundation of Marketing Geography. Summarizing this section of historical perspective, what has been described is essentially a two-pronged development of location theory, one a micro-approach, the other, a macro—approach. This study is a macro—study since the specific research is concerned with relationships of market structures over space. It is a study of the geography of markets. And yet, in very general terms, it is an attack on the two weaknesses of micro-theory by studying the relationship between relative location (locational interdependence) and market penetration (revenue). Private Enterprise Over Space In the introduction to this Chapter it was noted that this study is concerned with marketing in a private enter- prise economy. Implicit here is the assumption of competi— tion. The concern is with a market area in which several rivals vie for a buyer's business. This is not pure or price competition; instead, competition here means rivalry. This distinction is so very vital; in fact, it is the inter— face between Economics and Marketing as academic disciplines. Writing in the May 1965 issue of Economica, Nicos E. Devletoglou attacks this point in very relevant terms: Am... . 18 . . . over the years economists have suffered a growing tendency to regard the concepts of "price competition" and "competition" as synonymous. As a result, the term "spatial competition" is gradually disappearing from our vocabulary, though sporadically one still encounters it in the class- room and in print. But if this anomaly persists, an undesirable imbalance in microeconomic theory may become entrenched. A change in emphasis, therefore, is mandatory. Chamberlin17 and others recognized that the two price theory models of pure competition and pure monopoly were not adequate, that products are not always homogeneous. The result of product differentiation was imperfect com- petition which he called monopolistic competition. They recognized that this was not the impersonal competition of the pure price competition model, but was instead quite personal. This was a recognition of the existence of rivalry, one firm vis—a—vis its competitors. In addition to price the rivals compete on the basis of differentiation. The equilibrium price level in monopolistic competition is higher than that under pure price competition. But, this added economic cost is justified on the basis that differen- tiation offers a variety of choice, and in choice there is utility. This is still a very muddy area particularly as the number of firms becomes less and industrial structure 16Nicos E. Devletoglou, "A Dissenting View of Duopoly and Spatial Competition", Economica, May 1965, p. 140. 17E. H. Chamberlin, op. cit. Q F- V l9 approaches oligopoly. The important point here, in layman's terms, is that rivalry and differentiation go together and the result is variety or choice when product differentiation is involved. Spatial competition is differentiation, or rivalry over space. However, analysis of this situation using price theory, yields exactly the opposite result; that is, the consumer has no choice at all. The result is a number of spatial monopolists. This is the foundation of Losch's general theory and, at the same time, its very weakness. For a state controlled economy where choice can be eliminated this may be an ideal central planning guide.18 However, in a private enterprise economy, neither can the entrepreneur be forced to locate by any plan, nor is the concept of no consumer choice a realistic one. There are only two ways in which the consumer of a homogeneous commodity can have a choice of sellers within a given market area. Either all suppliers to that market must be located in the same place, or if they are at dif- ferent locations price discrimination must exist; or in more palatable terms, rivals must be willing to "meet competition." A thorough consideration of this problem is included in Chapter II. At this point it is necessary only to 18Greenhut comments on Losch's theory in this regard in Plant Location in Theory and Practice, p. 41. 20 establish the basis for the consideration of a buyer facing several possible suppliers over space. There are rivals who are in fact engaging in price discrimination. The objective is not to make a value judgment but instead to observe what exists in reality. This is not an attempt to deprecate the efforts of the economists. It is, in fact, quite the opposite. Economics is the foundation of marketing. Here we are concerned with the goods market of microeconomic theory. When differentiation exists competition is imperfect and rivalry exists. It is precisely because competition is imperfect that Marketing exists as a discipline. There are factors other than price that determine the quantity demanded, and when rivalry exists the more important question from the firm's point of view is from whom? This is where marketing comes in. Else—ties This study is concerned with the marketing effort of "the firm". One of the basic tenets of the modern (managerial) marketing concept is that the focus of the marketing effort must be on the customer. Within the goals and objectives of the firm, and within its physical and human constraints, the planning and carrying out of the marketing effort must be directed toward satisfying the wants and needs of the customer. The key to successful 21 marketing is knowing what variables affect the purchase decision and, in light of that knowledge, manipulating those variables in such a way as to achieve the goals and objectives of the firm. Many of the variables are, of course, uncontrollable. These include economic, political, social, ethical, and legal variables to name a few. These are usually referred to as environmental variables. It is the controllable variables that marketing management must manipulate. Product, price, and promotion are controllable variables. Location, too, is a controllable variable. Manipulation of controllable variables in a private enterprise economy results in differentiation among rivals. Manipulation of the location variable results in spatial differentiation. The objective here is to try to gain some insight into how this location variable affects the purchase decision. Industrial Marketing In order to examine the effect of a single variable it is most desirable to be able to hold the other variables constant or at least to abstract from them as much as possible. Industrial marketing is particularly well suited to this end. It is not that all industrial marketing is characterized by the conditions being considered here, for in fact it is not. It is rather that within the broad 22 area of industrial marketing there are some products and industries which do fit the basic requirements. Those two variables which this study seeks to abstract from are price and product. To abstract from product it is focused on homogeneous products. Many industrial products meet this condition. They are ones which are bought to specification to fill a particular need. The product of any one of a number of the existing producers or sellers may satisfy the buyer's requirement. So far as product is concerned, he faces a number of "possible" suppliers. To abstract from price, the delivered price to the buyer must be the same from all his "possible" suppliers. For this condition to exist, the producers or sellers must engage in what the economist calls price discrimination. This usually takes the form of freight absorption, price equalization, or zone pricing. The purpose is to be com- petitive or to "meet competition." Thus, there must be rivals at different locations competing for the same market of buyers. There is no doubt that this does exist both in consumer marketing and industrial marketing. This is not collusion, nor is it illegal. It is often called "adminis- tered pricing" which carries with it evil connotations. In many cases, however, it is sound marketing management and necessary administrative practice. The incidence of this kind of pricing practice is highest in oligopolies. The fewer the number of firms involved in a market, the more intense or personal the rivalry. As a result, there is 23 little doubt when facing few rivals that any change in price will be immediately countered. Ordinarily one looks at oli- gopolies on a non-spatial basis and concludes that size of firm as a result of scale economies determines whether we have few or many firms. There is, however, a very important spatial dimension here. A given region, city, or even village will support only so many firms engaged in a parti- cular activity. There are thousands of barber shops through— out the United States; but, there is no doubt that within each community they are a part of an oligopoly. This is a special case of marketing, and it is ad- mittedly a limitation. The objective, however, is not to study the entire marketing system; instead, it is a study of just one of the controllable variables affecting the system, location. The marketing of certain kinds of indus- trial goods fit this abstract model quite well. Focus of the Research The preceding remarks have been very general. Their purpose has been introductory. The intent has been to provide a general statement of the importance of the Spatial dimension in marketing, and, to discuss the background and the factors bearing on the location problem facing marketing management. The spatial dimension touches all aspects of marketing. Even limiting this study to the revenue, or "market" factors, 24 and, further limiting it to industrial marketing still leaves an area of great breadth. This particular study makes no pretense of being an all-encompassing work. Instead, it attempts to chip away at one little corner of a very large subject. It was stated earlier that the underlying question with which this dissertation is concerned is the relation- ship between relative location and market penetration. This question should be viewed as the stimulus for this study. The specific focus of this research is on the existing market structures for three industrial products. It is empirical research of actual markets as they exist over space. The objective is to study the spatial relationships of certain characteristics of these markets as they vary over space. The purpose is to gain insights into the order, or regularities, of these relationships. Summary This study, in essence, involves the locational, or spatial analysis of markets. This is a special sub-set of location theory. This chapter has presented a general treatment of location theory and the spatial dimension of marketing. Chapter II is devoted to specific considera- tion of the various approaches to the locational analysis of markets. 25 This first chapter is best summarized by enumerating the goals of this dissertation. 1. To emphasize the spatial dimension of marketing, if you will, marketing geography. 2. To place a special emphasis on Industrial Marketing. 3. To provide insights for solution of practical problems facing marketing management in evaluating marketing performance and planning new facilities. 4. To provide insights for the further development of Location Theory. 3" CHAPTER II THE LOCATIONAL ANALYSIS OF MARKETS The preceding chapter presented a general treatment of location theory and the spatial dimension of marketing. This chapter is focused specifically on markets. The loca- tional analysis of markets is but one part of the broad body of literature and theory related to the location of economic activity. The concern here is with the market for a given location or set of locations. The locations of raw materials, their transfer costs, production costs, and agglomerating cost factors are not involved. The concern is also with how the market for one location is related to or affected by the location of a competitor. Thus, the concern is also with locational interdependence. Locational interdependence is related to both cost and market (or revenue) factors. Here the concern is with the locational interdependence of market factors. The purpose of this chapter is to bring together various efforts and diverse approaches to the locational analysis of markets. This is essentially an attempt to integrate studies in Economics, Geography, and Marketing. Use of the word "integrate" is perhaps presumptuous in that it implies 26 27 a tying together of the elements to be integrated. There is no pretense at any kind of integration beyond physically tying the elements together in the same chapter. But, this in itself is inherently integrative in that it provides perspective for marketing in general and for this study in particular. The chapter is divided into three parts: (1) Economics, (2) Geography, and (3) Marketing. Each contains descriptions of the various approaches to the analysis of markets in the respective fields. The treatment is purposely cursory since the goal is perspective and not depth of understanding. Finally, the most important point with regard to this chapter and its goal of providing perspective is that it is not intended to be a critical analysis. It is expository. The nature of the inquiry in each of the three areas is entirely different. Each approach is of eminent value and is included because it is noteworthy. In fact, more than noteworthy, it is essential that anyone working in any of the three areas be aware of the work that has been done in the other two. This is a marketing study. It was stated in the previous chapter that certain economists' descriptions of spatial equilibrium are not satisfactory from a marketing point of View. This is true. But, in no way is this a criticism of the approach or the results, nor is it a denial of validity. It is instead an attempt to assess the validity of the approach to the marketing question under O 1 I < 28 consideration. Similar comments will be encountered in the Geography and Marketing sections. They are made in the same spirit and with the same intent. This chapter stands for the advocacy of studying the spatial dimension of economic activity, and that it be an interdisciplinary study. Economics Melvin L. Greenhut, in his book Plant Location in Theory and in Practice,1 distinguishes between "the theory of market areas" and "the theory of locational interdepen— dence" as the two approaches to locational (or spatial) analysis of markets. This is a reasonable and logical distinction and it is on this basis that this section of the chapter is presented. Greenhut acknowledges that studies in both areas are very closely related but finds clear distinction in the analytical frameworks of each. The market area approach assumes fixed locations and is essentially an analysis of short run phenomena.2 . . . from this point of departure the size and shape of firms' market areas are derived.3 1Greenhut, Op. cit., p. 24. 21bid., p. 25. 3Ibid., p. 37. II t '. 29 On the other hand, the locational interdependence school hypothesizes either movable locations (without cost) or planned future locations. Its framework is inherently designed for long-run analysis; its conclusions must be considered by those who hope to deduce a general equilibrium theory within a market area framework.4 and, The interdependence approach . . . seeks to find reasons for a particular location; . . . in other words, the factors causing concentra- tion and dispersion are highlighted. These two approaches differ in analytical patterns; they are alike in emphasizing the monopolistic aspects of space. This section of the chapter is divided into two sepa- rate parts, one dealing with locational interdependence, the other with the market area approach. The Locational Interdependence Approach. Until very recently this school has been characterized by the use of a linear market. A line is drawn and consumers are assumed to be evenly distributed along the line - from one end to the other. The question is: Where will sellers be located in an equilibrium state: The "market" is absolutely static but the firm or competing firms are completely mobile. Equilibrium is defined as that state in which further move— ment offers no advantage to the firm or firms competing in the market. The model is highly abstract and normative. 41bid., p. 25. 51bid., p. 37. 30 In 1929, Harold Hotelling contributed what is generally regarded as the pioneering effort in this school. It appeared as an article in the March issue of The Economic Journal.6 In a way it appears that Hotelling actually backed into this position of prominence as a location theorist. The Opening sentence of his article provides the clue: "After the work of the late Professor F. Y. Edgeworth one may doubt that anything further can be said on the theory of competition among a small number of entre- preneurs."7 Thus, his concern was not focused on spatial relationships but instead was focused on the problem of imperfect competition, Specifically oligOpoly and its special case, dquoly. In fact, the major part of the paper is devoted to a mathematical analysis of the effect of space on the equilibrium price that will obtain for spatially separated duopolists. In this sense his work was not pioneering but rather it was an extension of the thread begun in 1838 by Cournot in his famous treatment of the duopoly problem.8 Nevertheless, as a result of intro- ducing the spatial dimension to the duopoly problem he went on to open a new door to academic inquiry, and this was a pioneering step: 6Hotelling, op. cit. 7Ibid., p. 41. (my underlining.) 8A. Cournot, Recherches sur les Principes Mathema- tiques de la Theorie des Richesses (Paris: Hachette, 1838.) 31 As a further problem, suppose A's location has been fixed but that B is free to choose his place of business. Where will he set up shop? As promised, the following treatment of several members of this school is highly simplistic. Those desiring more should refer to the sources cited. Hotelling began by postulating a market described by a line bounded at both ends. Buyers of the commodity are uni— formly dispersed along this line and each has an absolutely inelastic demand for one unit of the product. Sellers A and B are located as indicated below and their cost of production is zero. Each buyer must pay a transfer cost of C per unit of distance. . a . x l y . b . A [ B Market of length = 35 In this example a = 4, b = l, x = 14, y = 16 Figure l Hotelling Model Each unit of length of the market represents one buyer (thus, a total of 35 buyers). Finally, the case rests on two critical ananormative assumptions: 1. Each seller seeks to maximize his profits by adjust- ing his price with respect to the other's price. Here price is the only variable. F.O.B. price is the same to all buyers. 9Hotelling, op. cit., p. 51. 32 2. Each buyer buys from the seller whose F.O.B. price plus transfer cost is lowest. Thus each "market" is mutually exclusive. Given these conditions and assumptions, Hotelling proves that there is an equilibrium price for each seller. The values of x and y are determined by these prices (pA and p3). In his example shown in Figure l where c = 1 (transfer cost per unit of distance), pA = 36 and PB = 34. Whereas prices can be maintained above these values, they ”will never remain below them."10 He then moves on to the question quoted earlier "where will he set up shop?"11 The answer is quite simple and obvious. If A is fixed B will locate one unit away from A toward the center of the market. But relaxing the assumption that A is fixed there is quite a more difficult problem. Hotelling argues that while (to him) it would be socially more desirable that they locate at the quartiles (since total transport cost to society would be lowest), the result of spatial competition is that they will instead cluster. Here the marketing argument might be that the value of choice might be greater than the additional cost Of trans- port. However, choice is not involved in this model. Hotelling further hypothesized clustering when more than two sellers are involved. The clustering in all cases would 1°1bid., p. 49. 111bid., p. 51. 33 be at the center. He finally concluded by acknowledging that an elastic demand for all buyers would result in some diSpersion but B "will not go as far from A as the public welfare would require. The tempting intermediate market will still have an influence."12 By introducing the spatial dimension, Hotelling demonstrated that the equilibrium solution was definitely dependent upon the relative location of competitors. This was the first important statement of locational interdepen— dence in the field of Economics. While it may have inspired considerable interest it can hardly be said that the article inspired much creative activity. Hotelling's article appeared in 1929, and yet Devletoglou writing in the 1965 article previously cited states: Unlike price analysis, however, the theory of location or of quality under conditions of imperfect competition has never been a cause celebre. Except for two articles written on the subject several decades ago, spatial com- petition, in the present context, cannot be said to have attracted to date more attention than has been afforded to it by occasional deferential referenfes to Professor Hotelling's views on centre-clustering. 3 The two articles Devletoglou refers to are those by A. P. Lerner and H. W. Singer, "Some Notes on Duopoly and 121bid., p. 56. 13Devletoglou, op. cit., p. 140. 34 "l4 and, Arthur Smithies, "Optimum "15 Spatial Competition, Location in Spatial Competition. Each of the contributors in this area of locational interdependence has essentially taken the basic Hotelling model and proceeded to analyze variations of the original conditions. Lerner and Singer, by placing a finite upper limit on the inelastic range, concluded that ". . . on the assumption that A takes B's position as given (i.e., loca— tion and price), he will not behave as in Hotelling's hypothesis, but will either keep at a distance or attempt to drive him out."16 COpeland considered the case Of infinitely elastic demand concluding that, in the absence of freight absorption, sellers would disperse in accordance l7 Smithies' treatment was with the dispersion of buyers. notably expansive. He imposed equal and identical nega- tively leping demand curves for each of the evenly distri- buted buyers. His basic model is presented on the following page.18 Note that while it is still a linear market, the 14A. P. Lerner and H. W. Singer, "Some Notes on Duopoly and Spatial Competition," Journal Of Political Economy, April 1937. 15Arthur Smithies, "Optimum Location in Spatial Com- petition" Journal of Political Econom , June 1941. 16 Lerner and Singer, op. cit., p. 154. 17Morris A. Copeland, "Competing Products and Monopolis- tic Competition," Quarterly Journal of Economics LV (1940-41). 18Smithies, op. cit., p. 426. 35 model is in two dimensions. ‘This has become a popular convention since it is graphically clear and descriptive. 0‘0 (I’d p =aq+b Figure 2 Smithies Model Smithies then added three considerations. First, he consi- dered variations in rivals' expectations under: (1) "full quasi-COOperation," (2) "quasi-cooperative as to prices and competitive as to locations," and (3) "full competition" which is, according to Smithies, equivalent to both Hotel- ling's and Lerner and Singer's treatments. He argued non- mathematically that the result is that the full quasi- cooperative solution will be the same as the two plant monopolist with locations at the quartiles and as conditions approach ”full competition" the result is that "equilibrium will be achieved with equal prices and equal territories 36 19 The effect closer to the center than to the quartiles." of demand elasticity is to exert a pull from the hinterland and away from "center-clustering." He noted that under Hotelling's zero elasticity "this means that a producer by altering his position does not affect his position in the hinterland."20 Second, Smithies considered variations in the freight rate showing that the higher the freight rate, the stronger the likelihood of dispersion of rivals. This is similar to COpeland's argument and is logical when pushed to the limit, i.e., there is no point in sharing the inner portion of the market. b Rather Than A B Figure 3 Effect of Freight Rate Third, in similar fashion, he demonstrated that variations in marginal costs yield similar results. 19Ibid., p. 432. ZOIbid. 37 -----~ Rather NO Than Sale A B A E Figure 4 Variation in Marginal Cost Without doubt, the best treatment of locational inter- dependence is given by Greenhut in two landmark volumes previously referred to: Plant Location in Theory and Practice, and Microeconomics and the Space Economy. He essentially accomplishes three things: (1) he integrates and polishes the works of the previous contributors, (2) he puts the school in perspective vis-a-vis the market area school, and (3) particularly in his second volume, he gives very rigorous treatment to the effect of space on the deter- mination of the profit-maximizing price level of the spatial monopolist. As a result, Greenhut presents the clearest and most complete statement of locational interdependence available. 38 Greenhut prOposes two sets of "factors" influencing locational interdependence which he calls "The Pure Factors" and "The Imperfect Factors." The following is a very brief summary of these two sets of factors:21 A. Pure Factors 1. Elasticity of demand curve — The more elastic, the greater the tendency to concentrate, and vice versa. 2. Shape of.marginal cost curve - If "negatively leping throughout almost all ranges of output and (if) F.O.B. mill pricing is followed," then this condition "encourages all competitors to disperse." "If marginal costs are increasing, the opposite type of conjecture is admissible." Even though each would gain by dispersing, the larger market areas involved tend "to create uncertainty as regards dispersion of rivals, which in turn promotes concentration." 3. Height of the freight rate - The lower the rate, the greater the tendency to concentrate, and vice versa. 4. Unequal costs at alternate locations - Tends to promote concentration at low cost centers (agglomerating cost factors). B. Imperfect Factors - these are market imperfections which tend to influence concentration or dispersion of rivals. It seems that here Greenhut is acknowledging the existence of imperfections which tend to weaken interdependence. These imperfections pertain to (1) industrial structure, such as 21Greenhut, Plant Location in Theory and in Practice, pp. 39-40. 39 imperfect or monopolistic competition, oligopoly, etc., as considered by Smithies, and (2) marketing efforts in both pricing and in non-price competition. 1. Where entrepreneurs believe personal contact with bankers, suppliers, and customers is more important than their location vis-a-vis their rivals. 2. Uncertainty regarding rivals (i.e., Smithies) where "firms are highly competitive in location as well as price, they will tend to concentrate." 3. The greater the differentiation in product, the less the locational interdependence. 4. Basing point or equalizing delivered price systems tend to discourage dispersion. Greenhut begins by concurring with Hotelling, Smithies, and Losch, that given equal negatively sloping demand func- tions, plus all the other assumptions of the conventional model described earlier, "A spatial scattering of sellers is, therefore, the rule . . ."22 He then goes on to bring in the importance of locational interdependence by saying: "But once all factors influencing locational interdependence are considered, concentrations Of like firms at a few pro- duction centers may be explained as was true in the Hotelling's case."23 Thus, to Greenhut, locational inter- dependence results in a tendency for rivals to concentrate. The condition of locational interdependence that results from these factors is the determinant of the sought 221bid., p. 39. 23Ibid., p. 39-40. 40 (but never reached) state of equilibrium in terms of price, quantity ppd_location of sellers. The state is never reached because that is not the objective. The objective is instead a theoretical construct describing the underlying system and not the vagaries of reality. It is, therefore, pure price theory with the added dimension, space. It is microeconomics, it is highly abstract and normative, and concerned with long run equilibria. Greenhut is an advocate of the locational interdepen- dence approach and his second book, Microeconomics and the Space Economy is a rigorous extension of his advocacy. His "general theory" is a combination of locational interdepen- dence and the Weberian least—cost approach. Here, however, it is vital to note the distinction in the nature of inquiry! Greenhut's is pep a general location theory; it is instead a general price theory. What he has done is to introduce the spatial dimension into both the goods and factor markets of price theory. Thus, expressed in very simple terms, he has applied location theory to the development of micro- economic price theory. In this sense it is of the same vital importance to marketing in space as traditional price theory is to marketing. In like manner, the descriptions of spatial equilibria are as unsatisfactory in marketing as are the equilibria of traditional price theory. The nature of inquiry in marketing is quite different. Equilibrium is not a goal in marketing although descriptions of it must be viewed and understood as 41 parameters or constraints. It is rather an understanding or description of the effects of all variables affecting the marketing effort - including price — which marketing seeks. From the point of View of this study, one of the main deficiencies in the locational interdepence approach is that it provides no way to assess market penetration. For example, the assumption is that a market with all sellers of equal size located at the same point will split the market pro— portionately. And, if sellers are spatially dispersed they become Spatial monopolists. When price discrimination is introduced in the form of basing point, zone, or equalized systems, the unanswered question is what is the effect of proximity, customer, service and so on? Herein lies a good part of the research of this study. Before leaving the locational interdependence approach, the article of Devletoglou must be discussed. It is included in this section since it is at the level of first reading little more than a rebuttal of Hotelling's centerwclustering thesis. It is in the Cournot-Hote11ing-Smithies tradition restricted to consideration of duopoly. Equally sterile is the model, with everything else remaining equal, evenly scattered, and constant except for the locations of the two sellers who are symmetrically mobile. Finally, the method- ology is rather simple geometric manipulation. And yet, with the introduction of one very simple and rational concept he has produced a very insightful article! 42 He calls this concept the "minimum sensible constraint of indifference" which he describes as follows: Quite simply, the principle put forth is that some positive minimum, however small or large, exists for the consumer where the difference in distance between patronizing one store rather than the other can be axiomatized as inconsequential, and thus too weak a . . . . 2 criterion for practical chOice or revealed preference. 4 The result of this is two hyperbolae between which is the (shaded) ”doubtful area" where the choice of the buyer is uncertain. As the sellers move symmetrically toward the center, the "doubtful area" increases. Once the distance separating the two becomes equal or less than the "minimum sensible,” the "doubtful area" includes 311 of the market area. (B)(B) (A > (A) (A L (Bl) \ . .n. _ 9 I -. . . I u ‘ ‘ . ~~- 1‘. L’ , ' i. ' (‘1 * b. ' ‘ I, - . i Minimum sensible =29. Figure 5 Devletoglou Model 24Devletoglou, Op. cit., p. 142. 43 Devletoglou uses this construct to submit the con— clusion that "the spatial equilibrium between two sellers of the same commodity is a dispersed one."25 This article obviously crosses the line into the market area school. And even though this transgression is inciden- tal to the article, it has definite bearing on the forth— coming discussion. It is also closely related to the overlap or "umland" region which concerns both geographers and marketers. While it may contribute little to the locational inter- dependence school of thought since it essentially confirms Greenhut's statement that "a spatial scattering of sellers is the rule, "it does accomplish the author's primary con- cern: "to stimulate thought and discussion on so deserving a t0pic . . ." as spatial competition. Furthermore, it offers significant marketing insight. For example, there is no reason why the same construct might not be very useful in studying product differentiation where instead of distance, the "minimum sensible" might be some measure of difference in prOduct attributes such as quality, useful life, etc. One final comment regarding the locational inter- dependence school. Even granting that this is a long-run approach, the notion of perfect mobility of sellers seems extreme. Certainly, the ice cream vendor is mobile, but the 251bid., p. 158. 44 steel mill certainly is not. Perhaps the run is too long to have any value. The Market Area Appgpach. At this point it would be well to reinforce the distinction between the two approaches to market analysis being considered here. Using Greenhut's distinction, the interdependence approach starts with a "market" and seeks reasons for the equilibrium location of rivals within that market. Once defined, the market areas for each of the rivals is per force determined. On the other hand, the market area approach begins with the "market" 33d assumes the locations of the rivals as already established. They seek to find those forces affecting the market boun- daries between rivals. Their interest is therefore the size and shape of the firm's market area. But, both are founded on the notion that spatial differentiation results in spatial monopoly. This was the principal failing from a marketing standpoint in the inter— dependence approach. It is also the failing of the market area approach. This is, in fact, a denial of the kind of rivalry the marketing student is concerned with. Devletoglou attacked this notion and suggested instead a "doubtful area" rather than a clearly defined boundary. Clearly, the exis— tence of more than one "possible" supplier is untenable under either of these two types of analyses except when two or more suppliers are located at the same point. While less 45 restrictive, Devletoglou's innovation allows the "possible" notion only within the "doubtful area." Nevertheless, just as was the case in the discussion of the interdependence approach, the work done in the market area school is extremely relevant and important to marketing. Both are the result of recognizing that cost alone is not enough for the evaluation of a location. There have been two principal contributors to the market area school of analysis, Edgar M. Hoover and August Losch. The two are quite different in their approaches, yet both have focused on the importance of the size and shape of the firm's market area. The frequently cited drawings shown below are the best graphic description of the nature and scope of Hoover's work in this area.26 M N Figure 6 Hoover Models 26Greenhut, Plant Location in Theory and Practice, p. 28. 46 Since x and y are spatial monOpolists, the determinant of their mutual market boundary is delivered price. Thus, the principal determinants, assuming F.O.B. pricing, are the firm's F.O.B. price and the freight rate. Figure A shows the two monOpolists splitting the market equally under con- ditions of equal price and equal freight rate gradients. Either lower price or lower freight rates for y result in the market boundary shown in Figure B. Figures C and D show the results of variations or irregularities in freight rates such as zones, mileage brackets, and gateways. The arc, MlNl, might represent the result of a cheaper mode such as water shipment to a port deep in the rivals' territory. By applying this method of analysis to an existing set of buyers and sellers, a comparative evaluation of locations can be made in Weberian fashion. However, since buyers are not evenly scattered, the immediate result is that pro- duction cost and demand characteristics must be used to determine the market boundaries. Hoover used a convention which he called the "margin line" to account for this. The end result, however, is still a clearly defined set of boundaries which describe the Size and shape of each firm's exclusive, monopolized market area. Such a cursory review of Hoover's work does not even begin to do justice to it. Hopefully this has been enough to convey the desired per- spective of the approach and to give some glimpse of its value and limitation in marketing application. 47 August Losch is much more difficult to handle. He is included in this section because the size and shape Of mar- ket areas is the foundation of his general theory. He did not, however, begin by assuming locations of sellers as given; nor could his theory in the least be termed short- run. Finally, rather than micro, it is a macro-theory. Losch began by postulating a perfectly homogeneous plane.27 He then set out to consider the purely economic reasons for spatial differentiation which must take place. He, thus, excluded historical and geographical factors and accidents. The two pure economic forces which produce a market area for a given point of production for a manufac- tured good are the "advantages Of mass production" and "shipping costs." The former work toward concentration and the latter toward dispersion. Combining a production func- tion of economies of scale with a price funnel reflecting transport costs he arrived at his "demand cone" which des- cribes the extent of the market area. AS long as profits are made, new entrants will emerge until all areas are covered and no pure profits are made. He then elaborately proved that a honeycomb pattern of hexagons will result since this shape will minimize total transport costs. 27See Stephan Valvanis, "Losch on Location," Readings in Economic Gepgraphy: The Location of Economic ActiVity, ed. Robert H. T. Smith, Edward J. Taaffe, Leslie J. King, (Chicago: Rand McNally and Co., 1968). 48 This market area part of his general theory while highly abstract is nevertheless completely deterministic. For this reason, Greenhut claims that Losch failed to take into account interdependence; and, under Greenhut's notion of interdependence he is correct. Yet, it would seem that Losch might well have argued that the hexagonal network taking into account the entire plane is in itself graphic expression of interdependence. The reason.for this conflict is that the two work toward entirely different ends. Green- hut is concerned with microeconomics; he is concerned with the problem of the fipp attempting to maximize profits over space in competition with rivals. Losch on the other hand was optimizing for society's benefit, allowing no pure profit and minimizing total transport costs; rivalry is not a part of his theory. Losch's goal was to derive a picture of the economic landscape which would evolve subject only to pure economic reasons. This he did in purely theoretical fashion by rotating different sized hexagonal networks around a common production point: in such a way as to get six sectors with many and six sectors with only a few production sites. With this arrangement the greatest number of locations coincide, the maximum number of purchases can be made locally, the sum of the minimum distances between industrial locations is least, and in consequence not only ship- ments bug also transport lines are reduced to a minimum. 8 28Losch, op. cit., p. 124. 49 The resulting cogwheel or corridor pattern therefore explains the agglomeration of economic activity and the resulting concentrations of population. Theoretical pattern of an Theoretical pattern of an economic landscape. economic landscape, but without nets Figure 7 Losch Models Greenhut's criticism is entirely valid so far as analysis of the firm's market area is concerned in a private enterprise economy. The same disagreement exists with regard to the general theory. Greenhut: It seems that attempts like that of Losch to explain a spatial long—run equilibrium by ascertaining the size and Shape of a firm's market area must remain of little value . . . It is worthy of special note that Losch's system though inapplicable to a private capi- talist economy is patently of value in economic planning. In fact, this is the use to which Losch eventually dedicated his theory. p. 29Greenhut, Plant Location in Theory and in Practice, 41. 50 In summary, the economists of both schools of thought have contributed greatly to the study of economic activity over space. So far as marketing over space is concerned their analysis provides long-term factors to consider in locational adjustments and short term analysis of existing market conditions. However, their devotion to the mono- polistic aspect of spatial differentiation still leaves in doubt any consideration Of what we have referred to here as market penetration. Geography_ Losch provides a convenient transitional link between the Economics section and this one. As indicated earlier, the geographers until recently had been concerned with the uniqueness of locations and the need for describing them. This was the result of a philOSOphical approach called "environmental determinism," Under this approach man's economic (even human) activity at any one location was strictly determined by a unique set of physical environ— mental circumstances. This, of course, was wholly unsatis— factory since it ignored (1) the obvious interdependencies among locations, and (2) even more critical, the fact that the very subject of study - man's activity — is the result of man taking decisions within the constraints of such things as environmental factors and subject to the effect of locational interdependencies. Thus, what was missing 51 was a search for order. There was a need to answer questions about the nature Of man, his behavior, his economic activity which might explain regularities over space. There was, in fact, a reluctance on the part of geographers to become social scientists concerned with the behavior of man. This is exactly what the economists had been doing by imposing order through the concept of economic man - von Thunen, Weber, et a1. Yet as Rushton notes: Many geographers abandoned interest in theory construc- tion because they could not accept the postulate of economic man. Central-Place Theory. One who did not shy from theory construction is Walter Christaller31 and it is with his name that one associates what has become known as "Central— Place Theory." The theory is important in the present con- text for two reasons. First, it marks the beginning of the new thrust in Geography; and second, it is extremely relevant to market area analysis. This is not to say his was a behavioral approach, it was not. But, what it was, was an attempt to consider the nature of economic activity and from that deduce a pattern of order describing the economic landscape. Here then is the link between Losch and Chris- taller - again note the dominance of the Germans. 30Gerard Rushton, "Characteristics of Location Theories," unpublished paper, Michigan State University, 1968. 31Walter Christaller, Die zentralen Orte in Sud deutschland (Jena: Fischer, 1933). 52 Christaller began by assuming a homogeneous plane of consumers for whom a number of goods and services must be provided. He then ranked these goods and services in descending order Of the minimum-sized market area necessary for them to be provided on a viable, or profitable basis. He defined this ranking of goods and services as a hierarchy with that good or service requiring the largest market as the highest-order good and the one with the smallest minimum market as the lowest-order good. Christaller then argued that the location providing the highest—order good, or, the "most national commodity," defines a central—place of highest-order from which all other lower-order goods and services will also be provided. He further postulated that the most efficient market area for this most national good must be hexagonal, resulting in triangular relationships between central places. The problem then was to determine the location of lower-order activities (and, therefore, lower-order places). Given three central-places, he theorized that the location of the next, or second-order, center would be at the mid-point of the equilateral triangle. Using this newly determined location plus the three original centers, a new set of smaller equal-size hexagons are drawn. Each is one-third the size of the original hexagons representing the market areas for the highest-order good. This smaller hexagon now determines what goods will be provided at the newly determined location. Any good whose minimum market 53 size is larger than the new (smaller) hexagon will be pro- vided only at the highest-order places. Any good whose minimum market size is equal to or less than the new hexagon will be provided at p252 the new, lower—order place 229 the highest-order places. Repeating this argument, the locations of third- order places are determined. And, again, the new, even smaller hexagons determine the goods and services which will be provided at these third-order places as well as being provided at the highest-order and second order places. Christaller arrived, therefore, at a description of the economic landscape involving a hierarchy Of places and a hierarchy of hexagonal markets which determine the goods provided at these places. Figure 8, shown below, demon- strates Christaller's principle.32 . Highest—order places 0 Second-order places 0 Third-order places Figure 8 Christaller Model 32B. J. L. Berry, GeOgrapny of Market Centers and Retail Distribution (Englewood CIiffs, N.J.: Prentice-Hall, 1967. See partiCularly Chapter III, "Classical Central- Place Theory," pp. 59-73. 54 According to Berry, "In Christaller much of the underlying theory is implicit, and it was Losch who, in an independent derivation, made it explicit." He summarizes the two: Both theorists agree on the spatial arrangement of stores required for the optimal distribution of a single good to a dispersed pOpulation. However, their arguments diverge significantly when they seek to obtain locations for many kinds of goods considered simultaneously. Losch derives corridors of concentration in which are found centers of specialization through rotation of networks, whereas Christaller derives a landscape of central-place hierarchies. Both theories are highly abstract and norma- tive. They are also both macro-theories. And, again, they emphasize the monopolistic aspect of spatial differentiation; market areas for each location Of activity are mutually exclusive. Thus, they are Optimizing for the welfare of society - the greatest number of firms with the smallest market areas possible - rather than Optimizing a single firm's objective whether it be maximum profit or market share. It is interesting that Christaller, a geographer, derives an economic landscape based, just as Losch's, on the production function and transfer costs. It was stated that Christaller's is not a behavioral theory, but in one sense this is not true. He does assume the behavioral model 33Ibid., p. 59. 55 of the rational economic man as the consumer minimizing travel or transport costs, thus supporting the clear dis- tinction of market areas. In other words, Devletoglou's doubtful area is eliminated; decision making on the part of the sellers in selecting locations is precluded. If there is to be more than one firm competing in this mutually exclusive spatial market they must, by definition, cluster at the central place. Yet, in spite of its abstract and normative aspects, central-place theory has proved to be an extremely good representation of consumer retail and service economic activity over space. The two related concepts of centrality and the hierarchy of places are the great contributions of this theory. As mentioned in Chapter I, the great preponderance of location work in marketing has been done in the two areas of retail location and distribution center location. Central- place theory has provided a theoretical basis for some of the work done in retail location analysis. Not surprisingly, much of this work has been done by geographers. An example of this work is provided by Brian J. L. Berry's, Geography 34 of Market Centers and Retail Distribution. In this book he succeeds in integrating the theoretical base with a variety of empirical studies some of which are in addition across culture and across time. . 34B. J. L. Berry, Geogra hy of Market Centers and Retail Distribution (Englewood CIifEs, N.J.: PrenticeJHall, 1967). 56 The best summary to this section on central-place theory is contained in the first paragraph of the preface to Berry's book: The thesis of this book is that the geography of retail and service business displays regularities over space and through time, that central-place theory constitutes a deductive base from which to understand these regularities, and that the con- vergence of theoretical postulates and empirical regularities provide substance to marketing geo- graphy and to certain aspects of city and regional planning. The foregoing treatment of central—place theory has purposely been more complete than that given the other approaches to the locational analysis of markets. The reason for this is that central-place theory provides the theoretical basis for the empirical research involved in this dissertation. There are three other areas in geography in which work has been done which is related to market analysis and which deserve mention here. The rather cursory treatment of each of these areas is not to demean their importance. Regional Science. It is really rather difficult to classify this area of study under Geoqraphy since it inte- grates all of the social sciences; but, since one might describe its goal as optimizing society over space the 351bid., p. vii. 57 classification is partially justified. Two forces have been developing which have resulted in the emergence of this relatively new area of study. On the one hand, there are the complexities of highly industrialized urbanized societies resulting in fantastic social, economic, and political problems. If trend be destiny, we are doomed to central-places which will become so colossal that their inhabitants will either smother each other or come to a halt in one grandiose traffic jam. Students in this area seek bases for planning to replace this chaos with order. On the other hand, the undeveloped and underdeveloped societies in this world are no longer isolated. Inter- national political and economic forces compounded by advanced transportation and communication have suddenly thrust upon these societies roles of world citizenship. In many cases, these are roles for which they are hardly ready to cope. These societies cry out for bases of planning for deve10pment. The field of regional science is new. The variables with which they wrestle are social, political, and economic. All of the related disciplines are involved. Theories such as those of Losch and Christaller will assume increasing importance and perhaps soon dictate governmental policy. The implications for marketing and the analysis of markets are Obvious. Already in England there exists a system of "location points" aimed at imposing order on the growth and 58 distribution of economic activity. It is not inconceivable that in the near future our own country may see the estab- lishment of new production facilities or even the expansion of existing ones in metropolitan centers either precluded by edict or effectively discouraged by tax law. All of this is far removed from'the micro-study of the firm's market, but its possible impact on that study is dramatic. Behavioral. This section is captioned as it is for want of a better word. Actually, empirical studies such as those reported by Berry are studies of consumer behavior. However, the reason for including this section is to comment on a new area of study related to the decision process. Granted, Berry's is a study of the consumer's decision as predicted by central-place theory and depending on the notion of the rational economic man. This new area is concerned, however, with decisions Of men who do not have perfect information, men whose decisions are a reaction to uncertainty or risk. An example of a study in this area is the one done by Julian Wolpert on the behavior of farmers in Middle 36 Sweden. Wolpert states: "The overall objective is to be able to substitute a workable spatial and behavioral model 36Julian Wolpert, "The Decision Process in a Spatial Context," Annals, Association of American Geographers, Vol. 54, no. 4, 1964, pp. 537-58. 59 of the decision process for the untenable structure of classical theory." He proceeds by first showing that the population does 225 behave as rational economic man; that he "does not achieve the fruits of his rational actions, i.e., optimum productivity from a given set of resources." He then attempts to demonstrate that the decision process (of the farmer) has a spatial dimension, that, in fact, there is a regularity in the variation of the process over space. One of the prominent factors contributing to the variation in the decision process was the "change and lag in the communication and perception of information." This is directly related to the great amount of work being done in the area of diffusion best exemplified by Everett Rogers, Diffusion of Innovations.37 A very thorough treatment of diffusion over space is found in Torsten Hagerstrands, The Propagation of Innovation Waves.38 It is expected that this area will be of growing interest to geographers; in fact, the word, diffusion, has a very definite spatial connotation. But, here, just as it is true in so many other examples in the social sciences, it is essential that disciplinary boundaries be bridged. Tying this to the 37Everett Rogers, Diffusion of Innovations (Free Press of Glencoe, N.Y., 1966). 38Torsten Hagerstrand, The Propagation of Innovation Waves (Lund Studies in Geography, University Of Lund, Sweden, 1952). 60 decision process by using the act of purchase and time of purchase as measures of diffusion automatically has value in the analysis of market areas. Mathematical. This word has been purposely selected to caption this third and final area under Geography rather than the term "quantitative". Too often that word is used to denote the application of statistical methods of analysis to measurements of behavioral variables. What is meant here is a non-statistical, mathematical approach. Both Losch's derivation of the hexagonal network and Devletoglou's "doubtful area" are examples of such an approach. The applications of mathematical methodology and theory construction have been slow in coming to geography but are now increasing in incidence. So far as the analysis of markets is concerned two examples will be discussed in this chapter, one in this section and one under Marketipg. The two selected in no way exhaust the set. It is felt that the two examples selected simply demonstrate the theoretical foundations of much of the work being done. The example selected for presentation here comes from the article by Chauncey D. Harris entitled "The Market as a Factor in the Localization of Industry in the United States."39 The concept of interest here is his measure of "market potential." 39Chauncey D. Harris, "The Market as a Factor in the Localization of Industry in the United States," Annals, Association of American Geographers, Vol. 44 NO. 4, pp. 61 First, a brief description Of the concept. Harris defines market potential (P) for any point as the summation ( ) of markets accessible to that point divided by their distance (d) from that point. n . i=1 Starting with a given market area, the market poten— tials for a number of points over that area are calculated. Points of equal market potential are then connected resulting in a contour map of market potential. Validity for the concept of market potential as a meaningful index of accessibility to markets would seem to rest on a progressive decline in quantity of goods moved with increasing distance.4 Figure 9, below, is an example of the distribution of market potential for the United States, based on Retail Sales in 1948.41 Percent below New York City by land 'ronsport COI°IOIl Figure 9 Distribution of Market Potential - Harris 401bid. 41Ibid. 62 As Harris points out, the model is very similar to those used in physics for calculating the strength of a field whether electrical, magnetic, or gravitational. In geography these are commonly referred to as gravity models. The obvious shortcoming of the model is its inability to take into account aspects of competition or rivalry. Instead it seems to do little more than present a contour map of market concentration. In no way does it seem to provide the basis for a decision rule which would lead the locator to select the point of highest market potential unless he would in fact be the sole seller or spatial monopolist. The model seems to provide little basis for competitive decision making or for the determination of market boundaries or market penetration. The important point here, and the reason for inclusion, is the demonstration of the application of the gravity model. Furthermore, just because this measure of market potential may prove less than valuable to the marketer, it does accomplish something for the geographer by providing a graphic description of one aspect of order over space. And, as for the model itself, it will be seen in the following section on Marketing that the gravity model may prove to be of significant practical value in location problems. Summarizing this section of Geography, we see a field whose past offers little but whose future seems very bright for Offering much that is of value in analyzing 63 competitive economic activity over space. The future promise of the field seems to stand as a signal to marketers to be aware of developments there. Marketing The purpose of this chapter has been to integrate various approaches to the locational analysis of markets. In focusing on marketing a critical distinction must be made. This distinction is best made clear by contrasting the two areas of marketing in which the spatial dimension has been recognized. The first of these two areas is commonly referred to as warehouse (or distribution-center) location. A great deal of work has been done here ranging from very simple gravity techniques to linear programming, total systems analysis, and simulation. Here, however, the market is assumed as given, or, more exactly, the market share for each destination is assumed given and fixed. Thus, this becomes a Weber-type problem Of finding the least cost location or set Of locations for "serving" the market. This problem is actually entirely internal to the system of the firm. The assumption is that the location of the warehouse will have absolutely no effect on demand or market penetration. The subject with which this dissertation is concerned is exactly the opposite. Here the concern is with the 64 firm's revenue and not its costs. For this reason there is no discussion of the warehouse location problem; nor is there discussion of related problems such as the trans-shipment problem or salesman routing problem. Perhaps one further note of clarification is required at this point. In the paragraph above it was stated that the warehouse, or distribution center, location problem will not be considered. And yet, the empirical research in this dissertation is based on industrial distributors. Essentially, this is a problem in semantics. In retail marketing the buyer-seller interface is at the "store". The distribution center is just that; it distributes to the store. This research, on the other hand, involves industrial marketing. The industrial distributor referred to here should really be called an industrial "store," it is at the buyer-seller interface. The other area in marketing which has been concerned with its spatial dimension is retail marketing. It is in this area that the work which has been done relevant to the locational analysis of markets is to be found. For consi- deration here, the work is classified into two categories: (1) empirical research, and (2) theory development. Within the first classification there is a rather significant amount of research which has almost entirely been done since World War II. It is rather interesting in reading Berry to find that, at least in his Opinion, work in this area has been progressing in two separate camps. 65 The group which he is writing about are geographers, he calls them marketing geographers. But, he then points out that: . . the work of the marketing geographers paid at best only lip service to the existence Of a field Of marketing and an expanding body of literature called marketing science emanating from the schools of busi- ness and economics. The groups are almost identical in their objectives and interests in the area. It would be very difficult to dis— tinguish whether an article were written by a marketer or a geographer. The only real difference might be termed motivation or orientation. Whereas the geographer would probably be satisfied with an understanding or explanation of order in economic space, the marketer while seeking the same may not be satisfied until he can translate this into implications for the marketing strategy of the firm. Work in the area of retail marketing has principally been directed toward the measurement of the "trading area." Two main measures have been used, drawing power and market share (or market penetration). Combining these measures allows trading area maps to be drawn describing the primary, secondary and fringe trading areas for each store. Most of the empirical studies have been devoted to the study of such measures, the factors affecting them, and how they vary over time with changes in the infrastructure, the competitive structure, and population distributions. Filling—station, 42Berry, op. cit., p. 126. 66 supermarket, and shopping center studies are most typically representative of this first classification. Essentially, it involves measuring marketing variables over space and attempting to empirically determine significant relationships. So far as the second classification is concerned, that Of theory develOpment, there is practically no content at all in terms of volume. But, ask almost anyone in marketing to give you one example of theory in marketing and he will probably respond: "Reilly's Law," (perhaps somewhat face- tiously). This may say something about marketing theory; but, the real curiosity is that it was in 1931 that William 43 J. Reilly wrote The Law of Retail Gravitatiop. Since that time until quite recently little more was even said about his "law." In fairness to marketing it must be acknowledged that it was, even as late as the 19403, an infant discipline. It is perhaps too much to expect a new discipline to recog— nize the importance of the spatial dimension when others much older still suffer from this failing. "Reilly's Law" was expressed in the form of mathemati- cal relationships. The underlying mathematical model is the gravity model and as denoted by the title this is the basic substance of his "law" or theory, that the behavior of the retail buyer can be explained by the gravity principle. 43William J. Reilly, The Law of Retail Gravitation (New York: Reilly, 1931). 67 Two equations are particularly of interest and direct— ly related to the preceding discussion. The first is what he called the "breaking-point" equation: D = Miles between A and B B 1 + /Size of A Size of B This states that the retail trading area boundary between Center A and Center B is DB miles from B. Size can be measured in terms of many variables: population, number of central functions, square feet of retail sales area, etc. However, regardless of the measure used, there is one unique breaking—point. As discussed earlier, when expanses of rural area separate two central-places the concept of a boundary is viable. The empirical research done by Berry in Iowa provides a graphic description of this phenomenon as shown in Figure 10 below.44 “cl" I X1 1’" “i Figure 10 Shopping Preferences of Rural Residents 44Berry, op. cit., p. 17. 68 Actually, in this context, the central-place theory of Christaller might be viewed as theoretical confirmation of this part of "Reilly's Law." In spite of the apparent similarities and complimen— tarity between the two, it becomes obvious that Reilly's Law for breaking-points cannot be pushed far. By attempting to aggregate all retail trade, the suggestion is that a boundary exists between any two centers. But, according to central-place theory, market boundaries exist between goods or functions of equal hierarchical order. Thus the breaking- point equation appears valid only when seeking boundaries between equal-order central-places pp as between specific functions as shown in Figure 10 from Berry's text. The second mathematical relationship provides a theoretical basis for market relationships in those areas where no clear-cut boundary can be defined, as for example within metropolitan regions. Two centers attract from intermediate places approxi- mately in direct proportion to the sizes of the centers and in inverse proportion to the square of the distances from these two centers to the imtermediate p1ace.45 T , T = proportions of trade from intermediate A B place attracted by centers A and B 45Reilly, Op. cit. 69 'U "U H Sizes of A and B Distances of A and B from the intermediate place. 0 U ll While this does take into account the fact that there is overlap and no one place has a monOpoly over an area, it does suffer two weaknesses. The first is the same as dis- cussed above involving aggregate retail trading; that is, it ignores the hierarchy of places and functions. The second weakness is that the model is deterministic. It is toward this second weakness that the first really significant contribution since Reilly's is directed. This refers to the work of David L. Huff46 which is centered on the following probability model of consumer behavior over space: This equation says that the probability of a consumer located at (i) buying at a store located at (j) is equal to the quantity Sj / Tij divided by the sum over all j's, all possible locations at which he might possibly buy. The (S) represents a measure of size, the (T) some measure of distance or time and the ( ) an empirically determined value. 46David L. Huff andearry Blue, A Programmed Solution for Estimating Retail Sales Potential (Lawrence, Kansas: Center for RegIOnal Studies, the University of Kansas, 1966). 70 By applying this model to a given market area, a set of contour maps can be drawn which show the probabilities of an individual located anywhere on the map Of buying from a particular location. Figure 11 is an example of such a map showing the contours of the probabilities of buying from location 1, given the existence of identical competi- tors at locations 2 and 3. Combining the maps of all three sellers located in the market results in the map shown in Figure 12. Used in conjunction with maps of estimated or actual "industry demand", calculations of expected values of market penetra- tion can be made for use in analysis of performance or for the planning of new locations. It is still too early to tell whether empirical research will confirm the validity of the model; but, at this point, it does seem to hold exciting promise for the practitioner as well as the academician. This section began by focusing on the work in retailing. Furthermore, every subject discussed from Reilly, through the supermarket and shOpping center studies, to Huff has been in the context of retail marketing. Unfortunately, the result of this situation is that it may be assumed by many in marketing that market area study (or more generally the spatial dimension of marketing) is a special sub—set of retailing. There is one whole other part of marketing which has been ignored. The desire at this point is to turn that 71 Figure 11 Probability Contours Probability of buying from 1, given 2 and 3 .50 .33‘ 333 1 1- .331 03‘2'1 .50 .50 2 3 (I II Figure 12 Combined Probability Contours P1 = Probability of buying from 1, etc. 72 perspective around and note that while the spatial dimension of all marketing is extremely important, only consumer mar- keting has been looked at. Industrial marketing has been ignored. Before concluding this section, it should be acknow- ledged that there has been an untold amount of research and analysis in this whole general area of which few will ever be aware. This is the work done by individual firms, and by consulting companies for firms, related to the spatial aspects of the firm's activities. The quality of work no doubt ranges from very good to very bad. But, regardless of the quality, it is unfortunate that there is not a desire on the part of private enterprise to make public and share some of the knowledge gained from such research. While it is understandable that no firm should give away its differential advantage, it seems equally clear that the practitioner's paranoia about secrecy may in the longer run be costing them more than the risk of disclosure. Ehrenberg attacks this situation in the August 1968 Journal of MarketingResearch47 and suggests that need for secrecy is more often imagined than real. One can only hOpe that this condition will improve with time. 47A.S.C. Ehrenberg, "The Great Confidentiality Nonsense," Journal of Marketing Research, August 1968, p. 331. 73 Summapy This chapter has looked at three different academic disciplines. In each one different approaches to the locational analysis of markets are found. Within the field of Economics there were the two approaches, interdependence and market area, differing in methodology but alike in seeking descriptions of equilibrium founded in micro- economic price theory. In Geography there is a recently emerged but promising approach searching for spatial order and factors contributing to that order. In addition, this approach carries with it the potential for integrating the efforts of many disciplines in studying behavior in markets over space. Finally in Marketing there are two areas in which concern for location and space are evident; but, only one of these two is really related to the locational analysis of the market. Warehouse and distribution center location problems have been described as internal least—cost problems. In the other area, generally restricted to the retail mar- keting of consumer good, the approach is focused on the consumer and on the variation in his behavior over space. Most of the work consists of empirical studies and derived rules of thumb for application by the practitioner with little theoretical support. In short, Marketing as an academic discipline is lacking in recognition of the 74 spatial dimension. This is one of the two motivating factors in this dissertation. The other is that the little work done in this area to date has been done in consumer marketing. The desire here is to direct attention to that imbalance through an emphasis on Industrial Marketing. These first two chapters provide the background and perspective for this dissertation. As such, they stand as the foundation for the empirical research undertaken. Chapter III is a description of the design of the research project. CHAPTER III THE RESEARCH DESIGN Chapter I was devoted to a discussion of the importance of the Spatial dimension of marketing. One very important aspect of this dimension for marketing management involves the question of the relationship between relative location and market penetration. In order to answer this question one must have a basis for the locational analysis of markets. Chapter II presented an overview of various approaches to the locational analysis of markets in three areas: Economics, Geography, and Marketing. Central-place theory, described in Chapter II in the section dealing with Geography, seems to offer promise as providing a useful basis for the locational analysis problem facing marketing management. The work of B. J. L. Berry with retail markets stands in evidence of its value.1 The primary purpose of this dissertation is to deter- mine whether central-place theory provides a useful basis for the analysis of industrial markets. The research is, therefore, a test of the application of central—place theory. 1Berry, Gepgraphy of Market Centers and Retail Distribution, 9p. cit. 75 76 This will be accomplished through the evaluation of empirical data drawn from existing industrial market struc- tures to determine whether these markets fit the theoretical framework provided by central-place theory. The premise is that if these markets do fit this framework, then central- place theory, in the words of Berry previously quoted, "constitutes a deductive base from which to understand these regularities . . ."2 the regularities, or spatial relationships of market structures. If they do not, the reasons for the lack of fit as revealed by the investigation is expected to provide new insights for the locational analysis of industrial markets. The foregoing introductory remarks to this chapter constitute a general statement of the research problem. The remainder Of this chapter is devoted to a discussion of the specific research design involved in this dissertation. The discussion is divided into the following four sections: . Scope of the Research The Specific Research Problem Methodology . Collection of Data. DNNH 0 Scope of the Research It is the purpose Of this section to define the parameters of the research project. Since the research 21bid., p. vii. 77 consists of the analysis Of existing industrial market structures, the parameters must be defined in terms of: (1) the definition of those market structures, and (2) the geographical space, or area of inquiry involved. Industrial Market Structures. This research is focused on the marketing of three industrial products: (1) grinding wheels, (2) ball bearings, and (3) steel bars. The industrial buyer with a requirement for any of these three products may buy either directly from the producer, or from an industrial distributor who stocks the product. Within the framework of this dissertation, these are defined as two different types of market structures. This research is restricted to the marketing of these three products through industrial distributors. The market structure for each product is defined as consisting of those establishments (industrial distributors) within a given geographical area which stock the particular product. Two criteria governed the selection of the three products. The first criterion was that the product be relatively homogeneous so that the buyer would conceivably have several possible suppliers. In the context of this research, a homogeneous product is defined as one sold according to standard specifications and standard- or administered-prices. The second criterion was that the products be selected from each of three industrial goods classifications: 78 (1) industrial supplies (grinding wheels), (2) component parts (ball bearings), and (3) component materials (steel bars). Customarily there are three other classifications of industrial goods: major equipment, minor or accessory equipment, and raw materials.3 Although they are frequently marketed through distributors, both equipment classifications were excluded on the basis of a lack of homogeneity. Raw materials were excluded because they are rarely handled by industrial distributors.4 Selection of the products from each of the three goods classifications adds breadth to the study. Product from each of the three classifications also reduces the possi- bility of overlap of distributors carrying each product. That is, there is a higher likelihood that the three market structures will be mutually exclusive such that the distri- butor who stocks grinding wheels probably does not stock ball hearings or steel bars, etc. Area of Inquiry. The region over which the study was conducted is displayed on the map on the following page. It includes the major portion of the southern half of lower Michigan. 3Alexander, Cross, and Hill, op. cit., p. 22. 4Industrial Distribution: 1968-69 Directory of IndustriaI’Distributors (New York: McGraw—Hill, Inc., 1968). 79 Saginaw Muskegon Gra d aanids Flint lensing alama OO 5 Battle Jackson. OAnn Creek ArbO Figure 13 Research Area Since central-place theory provides the foundation for the research the determination of the area Of inquiry was made on the basis of two criteria from central-place theory. Before proceeding with a discussion of the two criteria and their relation to central-place theory, it is necessary to clarify some of the terminology which will be encountered in the discussion of the research. Hierarchy: A ranking based on some measure such as population, market size, degree of specialization, etc. Order: Denotes position in ranking, i.e., highest-order is the highest ranking. 80 Place: A metropolis (Standard Metropolitan Statistical Area - SMSA) city, town, village, or hamlet; thus, a concentra- tion of people and activities. Central-Place: Denotes relative position of places. A hamlet is a central—place in relation to the rural area surrounding it. A village is a central-place in relation to the hamlets and rural area surrounding it, and so forth. Function: The activity of providing a good or service. Here the term is operationally defined in terms of the industrial dis- tributor establishment stocking one of the three particular products. Population: Unless otherwise stated refers to human population on the basis Of the 1960 U.S. Census of Population projected to 1970. To review briefly from Chapter II, central-place theory is based on the assumption that the goods and services re- quired by society can be arranged in a hierarchy, or ranking, based on the minimum market size (areal extent) necessary to carry on the activity of providing each good or service. This ranking is referred to as a hierarchy of functions. The theory postulates that the place at which the highest- order function is carried on defines the highest-order central-place. The theory further postulates that on the basis of the hierarchy of functions required by society there will be a hierarchy Of central-places and that these places will be located in hexagonal relationship to each other. The following figure demonstrates the principle. 81 Figure 14 Christaller Hexagonal Network 'With this theory providing the basis for determination of the research area, the following two criteria were established: 1. The area should include one place which might be defined as the highest-order central-place and, further, be broad enough to include a sufficient number of other places of varying size so that a hierarchy of places will exist. 2. The spatial relationship of these places to each other must be appropriate for analysis. With regard to the first criterion, Detroit, Michigan, 5 as the fifth ranking SMSA in the United States, fulfills the requirement of being a highest-order central-place. 51960 U.S. Census of Population. 82 Furthermore, the research area displayed in Figure 13 contains all incorporated places of 25,000 and larger population in the state Of Michigan. Table 1 shows a ranking of all places of 25,000 and larger population on the basis of four different measures. As noted on Table 1, there appear to be at least three distinct levels of hierarchy involved in this area. The unlisted places under 25,000 provide at least one more level of hierarchy. To analyze the Spatial arrangement of these places in accord with the second criterion, a one—hundred-twenty degree (120°) sector of Christaller's hexagonal network was laid out. It was found that by abstracting from actual location to this network a very close fit of hierarchy Egg spatial arrangement could be obtained. This is shown in Figure 15. Only one cell does not fit, the cell labeled B.C.-J.-A.A. for Battle Creek, Jackson, and Ann Arbor. This cell stands as a proxy for the three places. The network boundaries were then abstracted to the map. Figure 16 is a display of the second-order network boundaries. Detroit's first order boundary is not shown. Finally, Figure 17 is a display of the third-order boundaries along with a display Of the abstract pattern from which it was derived. 83 Taflel Central-Places in Research Area .momaoo one to soouom .m.o .eoma .xoom ouoo Shae monsoon .soaz .mcflmcmq umom .wuflmuo>wco ououm comesowz .coumomom mo coamfl>wo .soummmom OHEocoom was mmocflmom mo omouom .moma ..om sun .uomnumom HOOflumHumum somGEOHzm AHHV ANHV Aoav Amy Ame Abv Amy Aev Amv Amy Ame Adv mva.mma mma.mHH omm.mmH www.mvm www.mom Nm¢.Hmm mmh.vov wNH.mmv mmatmhv mam.mvH.H omw.mhh Hmvtomfl.m Aooomv UOGU< nooam> Shadow HHOHO .um How mum monomflm cousm uuom mucooo coonamu new one monomwm xomuo mauumm I mmzm uoz I a “may AHHV Aoav Ame Amv Ame Ame Amy Rev Ame Amy Adv mcwusuommocoz omw.m Heh.m Hom.wa hov.ma moa.mN www.mw Huh.mm www.mm Hmm.mN mnm.mw mom.em mamtmmv moomoamfim a Add Aoav ANHV Amv Amy Any Amy Amy AHHV Avg Amv ANV Adv moe.H moo mae.a mmm.a mmo.H mmo.~ mmo.~ amm.a oso.e hmmth hmm.oa mootmw OHOmOHosz momhoamem o Ham ANHV AHHV AOHV Amy “mo Any Amy Amv Avv “my Amy Adv .xcmu monocoo A v Hmm.moa mmm.mo~ vvm.vMH omo.mmH www.mma mmm.HmH hmo.mHN Nmm.HNN mmo.omm mam.hom ovo.NNm hNh.MhH.v coauoofloum cowumasmom chmH "ouoz spec mom «cons: Doom c0mxomn “romeo mauuom comoxmsz oonEmHmm snowmmm wooed com mcwmcoq HOOHOIOchB ocean mowmmm ocmuw noonolocoomm phoneme “opponpmoooam 4m2m 84 Kalamaz-é Highest—Order: P lace ® Hexagon .2! Second-Order: Place I. Hexagon ¢=======D Third-Order : P lace o Hexagon Figure 15 Abstraction Of Research Area 9 0 g. .0: 85 I ’ \ Mugkegon. .. - -( ’ /\/ Grand \ ’ Port EBRapldS \ Huron 4D " ’ \ ,Lansing \ \ Creek I > \o.._.._°.)( Jac' son OAnn /’ Kalamazoo \ Arbor / \ / \t ....... __-.____/ Figure 16 Second-Order Boundaries g. c Saginaw ° glamazcao Figure 17 Third-Order Boundaries 86 Since the cell boundaries are arbitrarily drawn, one final step toward conformity was taken by defining the research area in terms of county line boundaries. In so doing, stability Of the geographical control units was obtained.6 The area selected therefore meets both criteria. Although the area could have been expanded, it is felt that this would not significantly influence the results of the research. The Research Problem With the parameters of the research established, the specific research problem can now be defined. This is a test of the application of central-place theory as a basis for the analysis of the spatial relationships existing in three industrial market structures. Thus, the research consists of three separate tests, one for each market structure (product). The specific research problem is stated in the following terms: Given: 1. A geographical area containing a hierarchy of central-places dispersed over this area; and, 2. A market structure consisting of all the industrial distributor establishments stocking a particular industrial product. 6Donald J. Bowersox, Edward W. Smykay, Bernard J. LaLonde, Physical Distribution Management (New York: The Macmillan Company, 1968), p. 91. 87 According to central-place theory: 1. There exists a hierarchy of functions such that lower-order places will carry on lower-order functions (i.e., provide lower-order goods); and, higher-order places will provide these same lower- order goods, but, in addition they will also provide higher-order good, thus, carry on higher-order functions; and, 2. The hierarchy of function is determined by the market area required to carry on the function. The research problem is to determine, within the existing market structure and within the given area: 1. Whether there does exist a hierarchy of functions (establishments) which is related, according to central-place theory, to the order of place in which they are located; and 2. Whether the market areas of these functions (establishments) are related to (l) the hierarchy of the function, and (2) the hierarchy of the place. The Objective of this research is to attempt to answer these two questions on the basis of empirical data drawn from the industrial distributors who make up this market structure. Rephrasing the problem in more literary terms, there are two questions involved. The first asks whether the distributor of, for example, grinding wheels in Jackson, Michigan, is the same "kind" of distributor as the one in Grand Rapids, or if, in fact, there is a hierarchy of establishments (functions) which can be established on some measure of the data to be collected. If such a hierarchy can be established the question also asks whether, according to central-place theory, the highest-order establishment 88 (function) in Grand Rapids is of a higher—order than the highest-order establishment in Jackson. It is important to note at this point that only the highest-order establishment is relevant since according to the theory it is highest—order function which determines the hierarchy of place. The second question asks whether the territory (market area) which a distributor serves is related to (l) the hierarchy of his establishment (function), and (2) the hierarchy of place at which he is located. As in the first question, it is only the highest-order function at each place that is relevant to the research. A market consists of sellers ppg_buyers. To this point, all of the attention has been focused on the sellers. There is, also, in central-place theory an implied statement of buyer behavior. This can best be described in the following way. Within the context of central-place theory, the seller's territory can be defined as "looking down" from his order of place. That is, his market includes only places of lower order than the place at which he is located. The buyer's set of possible suppliers can then be defined by those sellers who consider him to be in their territory. The deduction, therefore, is that buyers "look up" for their possible suppliers. This concept of buyer behavior seems particularly relevant in industrial marketing since, in most cases, the producers' and distributors' salesmen go pp the buyer. This is in contrast to consumer buyer behavior where the consumer goes to the store. Or, in other 89 words, the consumer makes the decision of which stores constitute his set of possible buyers. Since buyer behavior is vitally important in analyzing markets, and since central-place theory does provide a possible framework for the analysis of that behavior, the research will consider one final problem. Given: Determine: A set of spatially dispersed industrial buyers who buy from industrial suppliers. Whether the set of possible suppliers as defined by the industrial buyer is con- sistent with central-place theory, i.e., all possible suppliers are located either at his same place or at some higher-order place - whether he does "look up". The consideration of this problem will be based on the analysis of the results of interviews with a sample of indus— trial buyers who are located in the research area and who buy from industrial distributors. Methodology The research problem is operationally defined in terms of three (3) hypotheses. Hypothesis Hl Hypothesis H2 concerns the relationship between the hierarchy of functions and the hierarchy of places, in each market structure. concerns the relationship between size of market area and (1) the hierarchy of functions, and (2) the hierarchy Of places, in each market structure. 90 Hypothesis H3 concerns the relationship between hierarchy of the locations Of the possible suppliers perceived by the industrial buyer and the order of place at which he is located. Before proceeding with a discussion of these hypotheses and the testing of them, two methodological steps must first be discussed. First, the establishments which make up each market structure must be determined. And, second, the hierarchy Of functions in each market structure must be determined. Determination of Market Structures. A market structure is defined as all of the industrial distributor establish- ments within the research area which stock the particular product. There are three products: grinding wheels, ball bearings, and steel bars; thus, three market structures. Two sources were used to generate the list of distri- butors for each market structure. They are: (1) the Indus- trial Distribution 1968-69 Directory Of Industrial Distri- butors for Michigan7 and (2) the yellow pages of the telephone directories for all incorporated communities in the research area over 10,000 pOpulation. Each distributor which indicates in either directory that it stocks one of the three products was included in the initial set Of the establishments making up the appropriate market structure. The only exception to this procedure is in the case of the 7920 Cite, pp. 154-690 91 Detroit SMSA where only selected firms were included in the initial listing. The reason for this exception is discussed under the comments dealing with the determination of the hierarchy of functions which follows immediately. Determination of the Hierarchy of Functions. This second step was to determine on the basis of the data to be collected whether, in fact, a hierarchy of functions (establishments stocking the product) could be developed for each of the three market structures within the research area. To determine whether a hierarchy could be developed, it was decided that establishments would be measured on two dimensions. The first is a measure of the degree of spe- cialization. The technique employed is a scaling technique 8 The scale used is a one- similar to the Guttman scale. dimensional, five-point scale of product size running from very common sizes to very special sizes. The conceptual basis for the scale is identical to the inventory phenomenon in which some items are "bread and butter items" that turn over very quickly, others move more slowly, and still others are very special items which move very slowly. Viewing this 8This is a method of measurement developed for latent structure analysis in social science research. For a des- cription of its use see: Latit K. Sen, Charles P. Loomis, J. Allan Beegle, "Central Place Hierarchy in Non-Metropolitan Michigan: A Scalogram Analysis," unpublished monograph, Agricultural Experiment Station, Michigan State University, East Lansing, Michigan, 1969. 92 phenomenon on a spatial basis, the very common items can be considered as ubiquitous - found everywhere — while at the other end of the continuum the very special sizes may be found at only a few locations. Each scale point includes two item sizes to lessen the possibilities of scale discontinuities. The scale used for steel bars is shown below as an example: Table 2 Product Specialty Scale (Common Sizes) (Very Uncommon Sizes) Items 1;),1 74 u I"; 1d"; x 7: yulefikyi l 71 ’7: “(71 g. 5 Angle Angle Angle Round Flat g 71x2” 3h. ‘lq i171 IAIXVY (x7117! Flat Round Flat Channel Channel Rank 1 2 3 4 5 Each establishment was first asked whether they stock either of the first two items (Rank = 1). To be included in the ranking, the firm had to respond affirmatively to the first scale item. A negative response eliminated the estab- lishment from the entire study and considered it not to be a part of the market structure. After that, each successive scale point of higher rank was checked. A negative response to both sizes in any rank terminated the search and the establishment was assigned the rank of previous positive 93 response. This, then, established a hierarchy of establish- ments on the degree of specialization dimension. The second dimension measured involves establishment size. The ranking is made on the basis of a combination of three measures: 1. Square feet of warehouse space. 2. Number of outside salesmen selling the products. 3. Number of trucks used for delivery of the product. Taking each of the three measures separately, the largest value obtained from all establishments in the par- ticular market structure determines the upper value of the scale. The scale is then divided into four equal parts and assigned values running from tOp to bottom from 1 to 4. The rank value for each establishment is then determined by summing the values from each of the three scales. Table 3 presents an example of the calculation of establishment size hierarchy of function. 94 Table 3 Calculation of Hierarchy of Function Highest values: Square ft. of warehouse - 100,000 Number of salesmen - 24 Number of trucks - 15 Sq. Ft./Value No. Salesmen/Value No. Trucks/Value 75,001-100,000 1 19-24 1 13-15 1 50,001- 75,000 2 13-18 2 9-12 2 25,001- 50,000 3 7-12 3 5- 8 3 0- 24,000 4 0- 6 4 0- 4 4 Value Establishment X Sq. ft. 40,000 3 Salesmen 8 3 Trucks 3 _£ Total 10 = rank value To summarize these two steps in the research design methodology, first, all of the establishments in each of the three market structures were determined. Then two measures were devised for evaluating those establishments in terms of a hierarchy, or ranking. These measures fulfill two purposes. First, they allow the selection of the highest order establishment in each of the twelve central-places (cities) in the research area. In other words, it is on the basis of these measures that the largest, or highest-order grinding wheel distributor, ball bearing distributor, and steel bar distributor in each city were selected. AS noted earlier, it is only the highest- order function at each place which is relevant to this study. Had there been a conflict between the two measures, separate hypothesis tests would have been conducted for each of the 95 two measures. This, however, did not present a problem as will be seen in the report of research results. The second purpose of these measures is that they provide the basis for the tests of hypotheses concerning the hierarchy of function. It is on the basis Of these measures that twelve establishments (the largest in each of the twelve cities) in each of the three market structures are ranked . One final note is required regarding the methodology to this point. This pertains to the creation of the degree of specialization scale. A fundamental assumption in this study is that the Detroit SMSA is the highest—order central- place in the research area. It follows from central-place theory that the highest-order functions should be located there. It was, therefore, decided that the creation of the degree of specialization scale would be based on extensive interviews with the highest—order establishments in Detroit. The determination of which establishments in Detroit were the highest-order establishments in each of the three market structures were made on the following basis: Grinding wheels - This product is sold through authorized industrial distributors. In the yellow pages of the Detroit telephone book six nationally known producers list their authorized distributors. These six constitute the set Of establishments in Detroit included in the study. Where more than 96 one authorized distributor was listed the deter- mination was made randomly by flipping a coin. The six are: Distributor Producer Industrial Abrasives Company Bay State Abrasives J. T. Wing Company Carborundum Cool-Cut Abrasives Company FMR Mel Shane, Inc. ITT National Grinding Wheels Sales Company National Tool Sales Company Norton Ball bearings - This product is sold primarily through specialty distributors called Bearing Specialists. Most producers appear to be willing to sell through any distributors. In the Detroit yellow pages three distributors are listed most frequently as the distributors for national producers. In addi- tion, these three are the only ones with branch operations in other parts of Michigan. They are: Bearing Service, Inc. Detroit Ball Bearing Company Michigan Bearing Company Steel bars (Hot Rolled Merchant Quality Bars) - This product is sold through industrial distributors called Steel Service Centers. The set here con- sists of two based on ten years personal experience in the industry: Joseph T. Ryerson and Sons, Inc. U.S. Steel Supply, Inc. The degree of specialization scales for each of the three products was developed on the basis of extensive 97 interviews with each of these firms. Each scale was based on the frequency of sale of items. The following table, Table 4, is the result of the interviews on which the steel bar specialty scale is based. The general method used in the interviews was to ask the distributors to select several of the most frequently sold items and several of the least frequently sold items. Then, between these two groups, three other groups of declining frequency were selected. Table 4 Product Frequency Hot Rolled Merchant Quality Carbon Steel Bars Size Frequency of sale for 1968 2" x 2" x 1/4" Angle 842 * a l x l x 1/8 Angle 401 * a 1-1/2 3(1-1/2 x 3/16 Angle 394 a 1/4 x 2 Flat 339 * a 1/2 x 2 Flat 241 1-1/2 x 1-1/2 x 1/8 Angle 238 * 3/4 Round 224 * 1/2 Round 220 3/4 x 3/4 x 1/8 Angle 133 * 2 x 2 x 3/8 Angle 127 1 x 1 x 1/4 Angle 118 1/4 x 2-1/2 Flat 116 * 1/2 x 1/2 x 1/8 Angle 59 1-1/2 Round 58 * 2 x 1 x 1/8 Channel 56 * 5/8 x 2 Flat 50 1-5/8 Round 14 5/8 x 4-1/2 Flat 14 * l x 3/8 x 1/8 Channel 10 * 1-3/4 x 1/2 x 3/16 Channel 10 * - Items arbitrarily selected. a - Most frequent of all bar sizes stocked. 98 Hypptheses and Tests; The research problem is Opera- tionally defined in terms of the following three working hypotheses. Hypotheses H1 and H2 are tested by the analysis of empirical data collected in personal interviews with establishments in each of the three market structures. Hypothesis H3 is tested on the basis of interviews with a sample of industrial buyers in the same research area. A discussion of the collection of data follows this section on methodology. Hypothesis H1: That the hierarchy of functions (rank order) composed of the highest-order function at each place is directly related to the hierarchy of those places (rank order) based on population. The null hypothesis is: HO: That there is pp association between the ranking based on the dimension of establishment measured (degree of specialization and/or size) and the population ranking of the places at which the establishments are located. This hypothesis will be tested by measuring the degree of correlation between the two ranks using the Spearman 9 The technique is Rank Correlation Coefficient (rs). explained below since it also is the basis for the statis- tical tests of Hypothesis H2. This is a nonparametric statistical test of measurements of two variables at an ordinal level Of measurement. The 9Sidney Siegel, Nopparametric Statistics for the Behavioral Sciences, (New York: McGraw-Hill, 1956). p. 202. 99 purpose is to determine: (1) a statistical measure (rs) of the degree of association between measures of those two variables, and (2) the significance of the statistic. The statistic, the Spearman Rank Order Correlation Coefficient (rs), is calculated as follows: 1. A list of all the establishments involved (N) in the rankings (hierarchies) is made. 2. Opposite each establishment is entered its rank on the one variable (in the case of H1, hierarchy of function) 32d its rank on the other variable (pOpulation). 3. The difference between the two rank values (di) for each establishment is determined and squared (diz). 4. The statistic (rs) is calculated: The significance of the statistic will be determined by testing the null hypothesis: Ho: That the two variables are not associated and that the observed value of r differs from zero only by chance. 3 The hypotheses will be rejected at them = .05 level of significance. This means that for a given number of establishments (N), the probability of a value of (rs) equal to or greater than a particular critical value, purely by chance, is less than .05 (p .05). The critical value is 100 obtained from the "Table of Critical Values of rs, The Spearman Rank Correlation Coefficient."lo Hypothesis H2: That the ranking Of the market areas as defined by the Operators of the highest order function at each location is directly related to the hierarchy of those places, and directly related to the hierarchy of those functions. A map of the area was presented to the respondent at each establishment, and, on it, he was asked to describe the limits of his salesmen's territories. The geographic area containing the aggregate of all salesmen's territories is defined as the market area of the firm. The respondent was asked to draw on the map the outer limits of this market area for his establishment. The map used was a conventional oil company highway map. Only those maps of the highest-order establishment at each place are used in the test of this hypothesis. This is the same set of establishments that is used in the test of Hypothesis H1. Square miles of market area is the measure that is used in testing this hypothesis. Each establishment is then ranked according to its measure of market area. Since this hypothesis concerns the relation of market area to (l) hier- archy of place, and (2) hierarchy of function, the hypothesis is rephrased in terms of two sub—hypotheses. 1°Ibid., p. 284. 101 Sub-hypothesis H2a: That the size of market territory is directly related to the hierarchy of the place at which the establishment is located. The null hypothesis is: HO: That there is no association between the ranking of market area size and the ranking of population. and, Sub-hypothesis H2b: That the size of market territory is directly related to the hierarchy of function of those establishments. The null hypothesis is: That there is no association between the ranking of market area size and the ranking (hierarchy) of function. Ho: Both sub—hypotheses will be tested using the Spearman rS statistic described above. The null hypothesis will be rejected at the = .05 level. The final aspect Of the research problem is concerned with the spatial behavior of the industrial buyer. In the context of this study the buyer's spatial behavior is defined by the locations Of those industrial distributors which the buyer selects as the most possible (or probable) sources of supply for a particular requirement. The research question is whether central-place theory provides a basis for the analysis of this behavior. The research problem is to describe the behavior of a sample of buyers and to determine whether their behavior is in accord with what might be expected on the basis of central-place theory. 102 A sample of three industrial firms was drawn from each of the twelve central—places within the research area: Ann Arbor Flint Lansing Battle Creek Grand Rapids Muskegon Bay City Jackson Port Huron Detroit Kalamazoo Saginaw The reason for using this cluster sampling technique is that each cluster represents one cell in the abstract hexagonal structure on which the research area was based. The result approximates a uniform distribution of the sample over the entire area thus reducing the possible effects of a unique variable at any one particular location. Although any area could have been used which meets the criterion dis- cussed earlier in the selection of the research area, this same area was used since data from the buyers could be collected at the same time as that from the distributors is collected. The objective is to study variations in behavior over space. Ideally each buyer would be identical in all respects except for his location. In an effort to abstract as much as possible from all other variables, two further restric- tions on the sample were imposed. First, the size of industrial firms selected were restricted to those that employ between 100 and 500 employees. The reason for selecting this particular range is that the firms were large enough to have a significant number Of dis- tributor requirements, but, at the same time, were not so large as to have an undue amount of purchasing leverage (re: General Motors). 103 Second, the sample was restricted to firms within the 33 to 38 Standard Industrial Classification (SIC) codes. The reason this particular range Of codes is that it is most representative of the manufacturing activity in the research area. The source of data for drawing the sample is the 1969 Directory of Michigan Manufacturers. Within the restrictions described above, a random sample of three industrial firms was drawn from each of the twelve places. The purchasing agent or buyer of each of the firms in the sample drawn was asked to select three products which he buys from industrial distributors' stock. He was then asked which three suppliers he considers to be the most probable or most possible sources of supply for each product, or in other words, his first three preferences in order of preference. The research problem is to determine whether, in fact, the industrial distributor does "look up" for his sources of supply. The problem is defined in terms of the following hypothesis: Hypothesis H3: That the preferred sources of supply as perceived by the industrial buyer will be located either at his same place or at some higher order place measured on population. The buyer for each industrial firm selected becomes an element in the sample; thus, a total possible n = 36. However, only those cases where one or more of the preferred 104 suppliers is located at other than the buyer's place are relevant to the hypothesis. The following dichotomous measure is used to test the hypothesis. Either: Category 1. All sources located at another place are at higher order places; or, Category 2. They are not. The null hypothesis is: Ho: That the preferred sources as perceived by the buyer which are located at some place other than his own are not related to the order of place of those locations as measured by population. This hypothesis is tested using a X2 one-sample test of significance. This technique is used to test whether a significant difference exists between the observed frequency of cases in Category 1 and Category 2 and the expected fre- quencies based on the null hypothesis. Under the null hypothesis the expected frequencies in each category are equal. The X2 statistic is calculated: _ 2 (01‘31)2 (02 E2) 2 _ X - + E1 E 2 where O and E1 represent observed frequencies and expecteé frequencies in Category 1, and 02 and E2, for Category 2; and, E1 = E2 under the null hypothesis. The null hypothesis will be rejected if the value Of X2== 6.64. This is the critical value of the Chi Square distribution fOr'OC= .01 level with one degree of freedom. 105 In addition to the question of whether the buyer does "look up", the direction in which he looks is also of interest. To provide some insight into this question, the data for all 36 buyers is displayed graphically on a map of the research area. This display is very similar to Berry'sll displays of his results of research in Iowa in which he asked a sample of consumers where they last purchased a number of different commodities and services. Figure 18, below, shows the resultant pattern of shopping preferences of urban resi- dents for purchases of clothing in 1960.12 ['1 Hunt]. "“30 - u lulu Figure 18 Shopping Preferences for Clothing 11B. J. L. Berry, Geography of Market Centers and Retail Distribution, op. cit. 121bid., p. 17. 106 It is anticipated that a similar display of industrial buyer data may provide the basis for certain inferences or conclusions to be drawn about buyer behavior, particularly with regard to direction. Collection of Data There are two distinct sets of data required for the research described. The first is the set of distributor ldata, the second, buyer data. Distributor Data. The distributor data required consists of three parts: 1. A measure on the degree of specialization scale. 2. A measure of size based on: a. Square feet of warehouse space. b. Number of trucks used to deliver the product. c. Number of outside salesmen selling the product. 3. A map describing the distributors market area for the product. The objective of each distributor interview was "hard data". There was no need to control respondents as would be the case in attitude or Opinion surveys. As a result, anyone present who could answer the questions was an accept- able respondent. The interview first asked to see the Sales Manager. If he was not available the interviewer asked to see an inside salesman familiar with the product. 107 Buyer Data. The buyer questionnaire is shown on the following page. The form consists of two (2) sections. Under Section I, the buyer was asked to select three products or product-line requirements which he buys regularly through distributors 229 for which he feels he has several possible sources of supply; i.e., that choice is not pre- cluded by exclusive contract. He was told that it would be preferable if the products selected were the kind that are purchased to standard specifications. And it would also be preferable if each of the three products involve different sets of distributors or types of distributors. With the three products selected, the buyer was then asked which three (3) distributors he considers to be his preferred sources of supply, in order of preference, for each of the products. If he appeared to be reluctant to give the names of the distributors, he was told that for purposes of the research names were not important, only the locations were needed. Section II of the questionnaire consists of an attempt to control response by product, and, specifically those products which were subject to test under Hypotheses H1 and H2. The respondent was told that the three products, - grinding wheels, ball bearings, and steel bars, are control items in the research and was asked if his company buys any or all of the items from distributors. If so, he was asked which three (3) distributors are his preferred sources of supply for his requirements. 108 BUYER QUESTIONNAIRE Name of Respondent Position SECTION I 1. Product: 1. 2. 3. 2. Product: 1. 2. 3. 3. Product: 1. 2. 3. SECTION II Grinding Wheels Ball Bearings 1. 1. 2. 2. 3. 3. Company Name Address Location Hot Rolled Steel Bars 1. 2. 3. 109 The data generated in this section, Section II, will be considered as secondary, or supplementary data. The rationale for this is that it was felt that asking a buyer about a particular product might arouse his suspicion and create a defensive bias in his response. In Section I of the interview, where the choice of product was his, it is felt that his response was less subject to this bias. In the conduct of the buyer interviews it was necessary that the respondent be either the Purchasing Agent or a buyer who is responsible for purchases from distributors. Method of Data Collection. Data was collected in the following three sequential steps. 1. First, in Detroit, as discussed earlier, six (6) grinding wheel distributors, three (3) ball bearing distri- butors, and two (2) steel service centers were interviewed. These interviews followed the questionnaire exhibited in all respects except for the specialty scale. It was on the basis of these interviews that the scales for each Of the products were established. 2. The next step was directed toward the reduction of the number of interviews required. This was done by tele- phoning all of the establishments at each place where the number of the establishments for any one product exceeded two. Each was asked over the phone to respond to the Specialty scale and to advise the number of outside sales— men they employ. On the basis of their responses, those 110 to be interviewed were reduced to the highest ranking two. The logic of this step is based on the fact that both of the hypotheses involving the distributors are focused on the highest order establishment at each of the twelve central- places in the research area. 3. This final step consisted of the actual data collec- tion from the distributors determined in Step 2 and from the thirty-six (36) buyers. Data from both were collected concurrently, through personal visits by the researcher. Summary This chapter has presented the design of the research involved in this dissertation. It has defined the SCOpe of the research, the Specific research problem, and the method- ology employed. Chapter IV is a report of the results of the research. Chapter V consists of the analysis of the results. Chapter VI presents conclusions drawn from the analysis, and com- ments regarding contributions of the research, limitations, and suggestions for future research. CHAPTER IV RESULTS OF THE RESEARCH This chapter is devoted to reporting the results of the research which was carried out in accordance with the research design as formulated in the preceding chapter. Analysis of these results and conclusions drawn from them are included in the fifth and sixth chapters. The data upon which the research is based were gathered by the author during the months of September, October, and November of 1969. This was accomplished through personal interviews with eighty-eight industrial firms. All of the firms involved in the research are located in one Of the twelve central-places in the research area: Ann Arbor Flint Lansing Battle Creek Grand Rapids Muskegon Bay City Jackson Port Huron Detroit Kalamazoo Saginaw There are two basic sets of data: (1) the industrial distributor data, and (2) the industrial buyer data. The industrial buyer data are based on interviews with purchasing agents of thirty-six firms, three firms in each of the twelve cities listed above. The industrial distributor data consist of three parts related to each of the three categories of distributor: 111 112 general—line industrial distributors, ball-bearing special- ists, and steel distributors. Fifty-two interviews were conducted with these distributors in order to determine the largest in each category in each of the twelve central places. In the research design each distributor establish- ment determined to be the largest of its category in a place is referred to as the highest-order function (establishment). Thus, in each of the three distributor categories, the data are drawn from the largest distributor in each Of the twelve central places. The presentation of the research results mirrors the composition of the data. The industrial distributor results are presented in three sections, one for each type of dis- tributor. Each of these sections consist of a discussion of (1) the product, (2) its distribution, and (3) the results of the research of the product distributor. The three in- dustrial distributor sections are followed by the final section of the chapter which presents the results of the industrial buyer research. All data which are not presented in the body of this chapter are found in the Appendix and are referred to by exhibit number. Grinding Wheel Distributors Product. The industrial product classification of this product is that of an operating supply. It does not become a part of the user's product and it is consumed in use. As 113 such it is referred to as an MRO (maintenance, repair, operating) item. The product does, however, have OEM (original equipment manufacture) application when it becomes a part of the product of grinding machinery manu- facturer. But, this application is customarily sold direct by the grinding wheel manufacturer to the grinding equipment manufacturer. The great majority of grinding wheel sales are as MRO items and these are almost entirely made through industrial distributors. Estimates made by the distributors interviewed regarding the percentage of total sales made by distributors ranged from seventy-five to ninety percent. The end use for this product is surface removal to finish (remove defects, rough areas, oxide, paint, etc.), to polish, and to size. In common parlance among distri- butors the name, grinding wheels, connotes a product for metal removal, although there are special wheels for concrete and masonry work. Most wood removal is done with coated abrasives like sand paper and belts. Within the industry, grinding wheels are technically classified as bonded abrasives. This is a homogenous product. A highly developed set of Industry standards has been develOped regarding sizes 1 and shapes of grinding wheels. Although each manufacturer 1USA Standard Specigications for Shapes and Sizes of Grinding Wheels, United States of America Standards Institute, Inc., New York, 1968. Sponsor: Grinding Wheel Institute. 114 uses somewhat different terminology and distinctive item- codings, each distributor has a conversion chart enabling him to determine his equivalent (identical) wheel in the brand he carries from the coding on another brand of wheel. A very high degree of brand loyalty exists both on the part of the seller (distributor) and on the part of the buyer (user). The distributor‘s loyalty appears to be organizational in nature; it is pi§_brand and he feels a relation to the producer's company. The impression is that this is not a product-based loyalty, but that the product is, in fact, homogeneous. On the other hand, user loyalty appears to be the direct result Of the grinding wheel operator and his reluctance to change. The Purchasing Agent may be convinced that any one of several brands would meet his requirements, but if the grinder operator is partial to a particular brand he can easily demonstrate that the others are inferior. The way the Operator uses the wheel will make all the difference in its performance. Improper Operating practice can literally break the wheel. Several grinding wheel salesmen mentioned this problem when attempt- ing to get a new customer to adopt their particular brand. There are many brands of grinding wheels produced ranging from those produced by small, local producers to familiar national brands. Within the research area, the following five brands were found to be dominant on the basis of distributor and buyer interviews: 115 Carborundum' Norton Avco—Bay State Simonds ITT-Peninsular Sterling The first two are believed to account for about 80% Of the total sales in the area. And, without doubt, these two are dominant so far as brand loyalty is concerned. The product is not complex. Product engineering require- ments are minimal. Operation is not dangerous, and risk to the work piece is usually slight. Principal service require- ments involve fine adjustments (grit, r.p.m., lubrication, etc.) to improve results and tool life. Distribution. The research is concerned with the indus- trial distributor who stocks grinding wheels. The principal outlet, or kind of distributor, is the general-line indus- trial distributor. These are the "general stores" for the industrial community. They are quite often called "industrial supply" or "mill supply" houses. Although there are some specialists who carry grinding wheels — mainly welding supply distributors and contractors products distributors - the major outlets are the general—line distributors. The general-line distributor customarily carries most or all of the product lines listed in Table 5. The percent- age figure next to the product line indicates that percent of the twelve firms in the research sample which stock that product line. 116 Table 5 Product Lines of General-Line Industrial Distributors Cutting tools and abrasives (100%) Power tools and accessories (100) Powered equipment (100) Pipe, tubing, valves, fittings (66.6) Fluid power products (91.6) Power transmission equipment (91.6) Bearings (50 Material handling equipment (100) Rubber and plastics (91.6) Ferrous and non-ferrous metals (83.3) Fasteners (100) Electric equipment (41.6) Safety supplies (100) Contractors' supplies (100) Misc. MRO items (100) Grinding wheels are an important product line for these industrial distributors: - One distributor with less than 20,000 square feet of total warehouse space said he does $300,000 in grinding wheel sales alone each year. - Another with approximately the same size warehouse estimated his grinding wheel sales at 30% of his total sales. ' Each general-line industrial distributor carries only one brand of grinding wheel for which he is the authorized distributor. In all eleven central places, excepting Detroit, each is the exclusive distributor for his brand. In Detroit, some producers have more than one authorized distributor. 117 The prices at which the distributor sells his grinding wheels are determined by the manufacturer; and, all prices are identical for the same item from every manufacturer. This statement is made on the basis of information from four different distributors. Price quotations were not obtained to verify this. Research Results. Each of the twelve general-line industrial distributors included in this part of the research are the dominant or highest-order establishment in their respective cities. This determination was made on the basis of size—measurement data: (1) square feet of warehouse space, (2) trucks used, and (3) outside salesmen. This was further verified through buyer interview with no exceptions encoun- tered. Exhibit II, of the Appendix, is a listing of these twelve establishments along with the measurement data collected. Each of these distributors is independent in that none is known to be owned or controlled by the producers of any of its product lines. Three, at Flint, Ann Arbor, and Detroit, are owned by the same parent organization; but, each of these three operates under a different name. They do, however, carry the same brand of grinding wheel. Three others of the twelve each has one smaller branch in the research area. The remaining six are local, independent operations with no branches. 118 Regarding overlap in the research design, three of the twelve grinding wheel distributors also operate steel service centers. But, in all three cases, the steel facilities are physically separate even though the salesmen do sell both grinding wheels 229 steel bars. So far as ball bearings are concerned, six of the twelve do carry a stock of bearings. They do so, however, only to complement a product line. None are a factor in the ball bearing market structure. In the Appendix, Exhibit I is the questionnaire used in gathering the data from grinding wheel distributors. Exhibits II and III display the data gathered from them in tabular form. Hyppthesis H1 The first hypothesis regarding the market structure of general-line industrial distributors is: That the hierarchy of functions (rank order) composed of the highest-order function at each place is directly related to the hier- archy of those places (rank order) based on population. As described in Chapter III, two measures of each dis- tributor were made which were to provide the basis for two separate rankings, or hierarchies of functions. The first measure was based on a Product Specialty Scale. This scale, which is found in the grinding wheel questionnaire (Exhibit 1), was developed on the basis of interviews with three Detroit area distributors. Although no "hard" data 119 was used, there was general agreement on the scaling from most common to least common.' As seen from the data in Appendix II, the product specialty scale measurements do not result in a hierarchy of function. There is, in fact, a negative hierarchy with seven of the twelve ranking as highest-order functions, three as second-order functions, one as third, and one as fourth-order. NO test was made Of the hypothesis using this measure. The second measure was a composite based on the three size measures of each establishment: (1) square feet of warehouse space, (2) numbers of trucks used, and (3) number of outside salesmen. The method of calculating the composite measure and the resulting rankings are shown on the following page for descriptive purposes. It is not included for ball bearing distributors and steel bar distributors. 120 Table 6 Calculation of Hierarchy of Function for Grinding Wheel Distributors Hierarchy of Function Place Sq. Ft. Trucks Salesmen Total (Rank)* Ann Arbor 4 4 3 ll 9 Battle Creek 3 4 4 ll 9 Bay City 1 l 3 5 1.5 Detroit 1 3 l 5 1.5 Flint 3 4 3 10 6.5 Grand Rapids 4 4 3 ll 9 Jackson 4 4 4 12 11.5 Kalamazoo 4 2 3 9 5 Lansing l 3 4 8 4 Muskegon 4 3 3 10 6.5 Port Huron 4 4 4 12 11.5 Saginaw 1 3 3 7 3 Bases: Square Feet Trucks Salesmen 85,501-114,000 = 1 10-13 = 1 15-19 = 1 57,001- 85,500 = 2 7- 9 = 2 10-14 = 2 28,501- 57,000 = 3 4- 6 = 3 5- 9 = 3 0- 28,500 = 4 0- 3 = 4 0- 4 = 4 * - Ties are assigned the average of the ranks which would have been assigned had no ties occurred. Sidney Siegel, Nonparametric Statistics for the Behavioral Sciences, (New York: McGraw-Hill, 1956), p. 206. The ranking of the twelve central places on the basis of population is shown in Exhibit IV. This ranking is referred to as the hierarchy of place. The literary null hypothesis is: H That there is no association between the hierarchy of ffifiction and the hierarchy of place. 0: This hypothesis was tested using the Spearman Rank Correlation Coefficient (rs) statistic. As stipulated in the research design the null hypothesis would be rejected 121 12, the critical value of at the = .05 level. For n .506.2 (For = .01, n 12, the critical value of H (I) ll r8 = .712.) These values are applicable in the tests of Hypotheses H1 and H2 for all three product distributors. Likewise, the Operational null is identical for all tests of H1 and H2: That the Spearman rank order coefficient of correlation (rs) between the two ranks is less than .506. If the observed value is equal or greater, then, that value is significant and the null hypothesis can be rejected. Testing this hypothesis using the composite size measure, a Spearman rS = .274 is obtained. This is not significant and the null hypothesis cannot be rejected. The calculation of this statistic is shown in Exhibit V for descriptive purposes. Since it was felt that a hierarchy of place based on human population might not be apprOpriate for industrial activity a second hierarchy of place was developed. This second hierarchy is referred to hereafter as Industrial Place rank. It was constructed by first ranking each of the twelve central places on three separate measures: (1) number of manufacturing employees, (2) number of manu- facturing establishments employing twenty or more employees,3 2Siegel, ibid., p. 284, Table 9. 3U.S. Census of Manufacture, 1963, Volume III, Area Statistic. 122 and (3) Merchant Wholesaler Sales, Class 508.4 Next, the rank value for each of the three measures were totaled for each place. The Industrial Place ranking is based on these totals. Both the Industrial Place rank and the census data upon which it is based appear in Exhibit IV. Testing Hypothesis H1 using this Industrial Place heirarchy yielded a higher Spearman rS = .41. This also is not significant and the null hypothesis cannot be rejected. Hypothesis H2 This hypothesis is concerned with the market areas of each of the twelve firms and states: That the ranking of the market areas as defined by the operators of the highest-order function at each location is directly related to the hierarchy of those places, and directly related to the hierarchy of those functions. The test of this hypothesis consists of measuring the degree of association between the ranking of the twelve firms in terms of square miles of market area and (l) the hierarchy of place (Hypothesis H2a), and (2) the hierarchy of function from Hypothesis H1 (Hypothesis H2b). 40.8. Census of Business, 1963, Wholesale, VOlume V, Area Statistics. 123 H2a Exhibit III contains the market area data for the twelve firms. The conclusion reached from visual observa— tion is confirmed by the statistical test of Hypothesis H2a. The null hypothesis is: Ho: That there is no association between the ranking of market territory as measured in square miles and the hierarchy of place. The Spearman (rs) rank coefficient of correlation is rs = -.007 using pOpulation as a measure of hierarchy of place. Using the Industrial Place rank described earlier, the coefficient of correlation is rS = .07. Neither result is significant and the null hypothesis cannot be rejected. Hypothesis H2b is concerned with the relationship between the market territory and the hierarchy of function measured in Hypothesis H1. The null hypothesis is: Ho: That there is no association between the ranking of market territory as measured in square miles and the hierarchy of function. The statistical test of this hypothesis yields a Spearman rs = .41. This result is not significant and the null hypothesis cannot be rejected. Figure 17 is a map of the research area on which all of the sales territories except Detroit are superimposed. Examination of this map confirms the test of the second 124 hypothesis in that there is no apparent order or regularity to the display. One aspect relative to central place theory is evident from the map and is summarized in Table 7. Table 7 Market Territory Summary of Grinding Wheel Distributors Other Central Places Population Distributor's In Market Territory Rank Place Lower-Order Higher-Order 1 Detroit Port Huron - 2 Grand Rapids Muskegon — 3 Flint — - 4 Lansing Ann Arbor - Jackson - 5 Ann Arbor Jackson - 6 Saginaw Bay City Flint Port Huron Lansing 7 Kalamazoo Battle Creek - 8 Muskegon - ~ 9 Battle Creek - - 10 Jackson - Ann Arbor 11 Port Huron - - 12 Bay City Port Huron Flint Saginaw Three (3) distributors consider places of higher-order to be in their market territories. Six (6) include other lower-order places in their market territories. Two other questions were asked in each interview with industrial distributors which it was felt might add insight to the research. The first question was: 126 Whom do you consider to be your principal competitors from out of town for your grinding wheel market here in (location of establishment)? The results of this question are shown in Table 8. The answers are given in terms of the location of their outside competitors. Of note here is the fact that each distributor who feels he actually has outside competition cited the loca- tion of his principal competition at higher-order places than his own. Table 8 Location of Principal Outside Competition Grinding Wheel Distributors Principal Principal Place Competitors pPlace Competitorsy Ann Arbor - Detroit Jackson - Lansing Battle Creek - Kalamazoo Kalamazoo - none Bay City - Saginaw Lansing - none Detroit — none Muskegon - none Flint - Detroit Port Huron — Detroit Grand Rapids - none Saginaw - none The second question asked was: If one of your customers had an immediate requirement for a Size or quality of grinding wheel which is not available from any local source, where would you suggest he try (or, where would you try for him) to find it? The response to this question by all the grinding wheel distributors was entirely consistent. First, they would get the item for their customer, and absolutely would not recom- mend a competitor or his brand. Second, and mirroring this 127 response, the only conceivable source in all cases was the distribution center maintained by the manufacturer Of their brand. There was no indication of cooperation among indus- trial distributors in other places who are distributors of the same brand except for those three previously mentioned with common ownership. Ball Bearing Distributors Product. This product has both OEM and MRO application. As an OEM item it must be classified as a component part. Practically all OEM sales are made directly by the ball bearing manufacturer to the manufacturer who incorporates it into his product. In MRO applications the industrial product classification is that of a supply. Most MRO sales are made through distributors. The end-use of this product is the reduction of friction as one part moves about another. The product is relatively homogeneous. There are well established standards of size and shape. These standards are written in both inches and millimeters (mm.) reflecting an international basis for their use. In OEM application users generally buy one brand exclusively for a particular application. But, in MRO use there appears to be no brand consciousness on the part of the buyer. Several buyers during the buyer interviews indicated that brand was not a concern in replacement. One said, "I just tell the 128 distributor the number of the bearing which failed and ask him to send me a replacement that will fit."5 There are many brands manufactured and marketed nationally. The major Detroit distributor advertises stocking "64 leading national lines of bearings."6 SKF has probably created the dominant image in the layman's mind, but others such as Fafnir, Hyatt, New Departure, and Timken are probably also familiar. The product itself is not complex. However, as a working part in a piece of equipment or machinery it may become a critical element. Considerable engineering and technical assistance are involved in OEM applications. So far as MRO applications and the ball bearing distributor are concerned, a considerable degree of product know-how is required. Distribution. The industrial distributor who stocks a full line of ball bearings is a bearing specialist. This is usually his only product line. There are other types of industrial distributors who stock a limited line of bearings as a complement to a particular product line or for a parti— cular market segment. Half of the general-line distributors discussed in the grinding wheels section carried a limited line. Automotive parts distributors carry only automotive 5Statement from buyer interview. 6Detroit Yellow Papas. Michigan Bell Telephone Company. 129 Specifications. But, for general industrial MRO needs, the industrial buyer relies on the bearing specialists. The bearing specialist carries all or most Of the brands produced. There are no exclusive authorized distributorships. The ball bearing producers determine the prices to be charged by the distributors. The prices for the same size bearing from all producers are identical. Again, price quotations were not obtained; the statement is based on information given by the bearing distributors and buyers. Within the research area there are several bearing specialists but all but two are local and located in Detroit.7 One of those two, the Michigan Bearing Co., has branches in four of the cities in the research area in addition to Detroit, its headquarters (Flint, Grand Rapids, Kalamazoo, and Saginaw). Detroit Ball Bearing Co., the other non-local bearing specialist, is the dominant distributor in all of the research area. This company has branches located in every one of the central places in the research area except for Port Huron where there is no bearing specialist. They have fiyg branches within the Detroit metropolitan area (SMSA). Because of this fact, the research sample consists of only one firm and therefore becomes a special case which is in general not suited to the research design. Also, while the location of the branches is public, in fact advertised 7Five Specifically identified as such in the Directory of Industrial Distributors, op. cit. 130 information, the company did request that the balance Of the data collected remain confidential. Hypothesis H1 Due to the nature of this situation, no attempt was made to construct a product specialty scale. In addition to the confidential nature of the information, this would have involved a lengthy study of the company's inventory system and each branch's place in that system. However, it can be reported that the company's inventory is computer- based; and that this, as well as all other aspects of the Operation observed, appeared to be well managed. On the basis of the data provided, a hierarchy of function based on size measure was developed and tested for association with the hierarchy of place. The null hypothesis is: ‘Ho: That there is no association between the hierarchy of function and the hierarchy of place. The test yields a Spearman (rs) Rank correlation coefficient r8 = .889 using human population as the measure of hierarchy of place. Using the Industrial Place rank rs = .921, the null hypothesis can be rejected at the = .05 level of significance. In fact, both values are significant at the a .01 level. 131 Hypothesis H2 H2a The null hypothesis of Hypothesis H2a is: Ho: That there is no association between the ranking of market territory as measured in square miles, and the hierarchy of place. Using population as the measure of place, a Spearman (rs rank correlation coefficient of r5 = -.20 is obtained. Using the Industrial Place measure the test yields rS = .08. In both cases the null hypothesis cannot be rejected. H2b The null hypothesis of Hypothesis H2b is: Ho: That there is no association between the ranking of market territory as measured in square miles, and the hierarchy of functions obtained in Hypothesis H1. The statistical test of this hypothesis yields a Spearman r3 = -.15. The null hypothesis cannot be rejected. Steel Bars Product. This product was specifically defined as Hot Rolled Carbon Steel Bars, merchant quality. This is a speci- fic description of a common product. Since there are many kinds of steel bars - special quality, alloy, stainless, tool steel, cold finished, ground, polished, etc. — this quality was selected to key the general-line steel distri- butor. The general-line distributor commonly carries 132 merchant bars, structurals, plates, and hot—rolled sheets. He may also carry the more special products. There are Specialist distributors, particularly bar specialists and sheet specialists; but, these usually do not carry the common items. There are no merchant bar Specialists. The industrial product classification is that of a component material. The product has both OEM and MRO applications; and, both applications are served both directly from producing steel mills and through distributors. Although subject to a broad range of applications, the end-use is that of a component material, to provide strength and durability. Practically all applications require fabrication. The product is absolutely homogeneous. Here the direct reference is to merchant bars, but this applies to all bars, structurals, plates, and sheets. The product is produced by the steel mills to standard specifications including chemistry, physical prOperties, dimensions, and tolerances. These standards are established and regularly reviewed by the producers' trade association, the American Iron and Steel Institute (AISI). Industrial users sometimes use the product of several mills indiscriminately in their fabrication process. However, the more common practice when possible is to retain the iden- tity of the supplying mill for ease in processing material claims and for vendor quality evaluation and analysis. 133 The user who buys from the steel distributor in most cases has no idea which mill produced the product. Further- more, in ordering from the distributor he orders by item, Size, and specification, ppp_by producer, i.e., brand. There is, in fact, no brand. There are 122 domestic steel producing companies listed in the weekly price list published by Iron Age magazine.8 For this particular product, 29 of these companies roll merchant bars at 46 mills throughout the United States. The product is relatively complex. Some degree of metallurgical knowledge is required of all steel distributors. Although many applications are routine, others are quite complex and critical. Services expected of the distributors involve assistance in welding practice, die practice, heat treating, etc. Distribution. Steel distributors are currently called "steel service centers." Until the 19503 they had been called steel warehouses. The trend now is toward the term "metals service centers" as aluminum, titanium, and other non-ferrous metals become important factors in their businesses. Roughly 15% to 20% of all the carbon steel produced in this country moves through steel distributors. The same percentage range applies to the specific product merchant 8IronAge, Published weekly by the Chilton Company, Chestnut and 56th Streets, Philadelphia, Pa. 19139. 134 bars. The balance is sold directly by the producing mills to the users. The organizational structure varies from very large national "chains" to small, local, independent steel service centers. Three of the largest national chains are owned by steel mills and are often referred to as "captives". Joseph T. Ryerson & Son, Inc., is the largest of all and is owned by Inland Steel Co. The other two are owned by United States Steel Corporation and Jones & Laughlin Steel Corpora- tion and operate under the same name as the parent company. A fourth producer, National Steel Corporation, has recently entered this category on a smaller scale with the purchase of a regional midwest chain. There are also two large, independent national chains: A. M. Castle & Co. and Earle M. Jorgenson Co. There are twenty regional chains with six or more branches. These are all independents. Finally there are smaller independents with five or fewer branches on down to the local independent with only one warehouse. This is not to imply that the local independent is always a small Operation. On the con- trary, he may have quite a sizable plant with a broad product and size range. In its 1969-70 directory, the Steel Service Center Institute which is the steel distributors' trade association, lists 392 active members in the United States and Canada. Within the research area the three large "captive" chains plus five of the regional chains operate plants in 135 Detroit. The balance of the research area is limited to the smaller independent organizations. Thus, the highest— order functions are located at, and pply_at, the highest- order place, Detroit. As noted earlier in the grinding wheel section, it is not uncommon for general-line industrial distributors or mill supply houses to carry the more common steel items as a complement to their other product lines. Some of these are actually the dominant steel distributor at their loca- tion, but this is usually because no steel service center exists at that location. This may be due to historical accident or insufficient steel demand to support a steel service center. Within the research sample for this product, four of the twelve steel distributors are general-line indus- trial distributors. The steel distributor serves both OEM and MRO require- ments. This applies to all steel items in general, as well as merchant bars specifically. The principal determinants of which source of supply - mill or distributor - a customer will use are quantity and usage rate. As a general rule, a mill requires a minimum order of a full truckload (32,000 pounds) or a full carload (40,000 pounds), and minimum items of three to five tons. In addition there are quantity price breaks. Quantity required plus usage rate determine inven- tory cost or "cost of possession". 136 Other contributing factors are: - Rush requirements. Minimum mill lead times are 2-3 weeks. - Preeproduction processing. Service centers perform such processing as: burning, shearing, sawing, punching, etc., not done by mills. - Credit. Distributors are willing to accept much greater risk than mills. Steel service centers are industrial distribution specialists. Like the bearing specialists, these are steel Specialists. However, the ball bearing specialist is much more like the general-line industrial distributor. The steel distributor is quite unlike the other two. Several primary distinctions can be made: - Steel items are large and bulky. They cannot be put in a drawer or on a shelf. - Large warehouse space is required. - Heavy material handling and processing equipment is required. - Safety hazards are involved. In short, the steel service center business involves running a plant. The other two types of industrial distri- butors being studied are involved with running stores. There is one other crucial distinction. The steel distributor is free to determine for himself the prices he will charge for his products. The steel distributor data are contained in Exhibits VII and VIII. Exhibit VII is relevant to measures of hier— archy of function and Exhibit VIII contains market area 137 measurements. Exhibit VI is the questionnaire used in the steel distributor interviews. Hypothesis H1 The first measure in the test of this hypothesis involves the product specialty scale. The scale used for steel bars is found in the steel distributor questionnaire, Exhibit VI. This scale was developed on the basis of actual, current data. The specific data is being held confidential at the request of the respondent. As in the case of grinding wheels, the measure on this dimension did not result in any differentiation among the twelve steel distributors. The results of this measure are displayed in Exhibit VII. Nine (9) of the twelve (12) scored five (5), the highest degree of specialization, and the other three scored four(4). NO statistical test of significance was made since there is pp association between this measure and any hierarchical ranking of places. Although it was not a part of the research design, an additional set of questions was added to measure product comparability for this particular set of distributors. The following four questions were asked: 1. What is the maximum thickness of plate which you stock? 2. What is the maximum section depth of Wide Flange Beams which you stock? 138 3. Do you stock the following sheet products: — Hot Rolled? - Cold Rolled? - Galvanized? 4. What is your shear capacity? The results of these questions are shown in Table 9. Table 9 Additional Product Measure Of Steel Bar Distributors Sheets Rank Place Plates WFB HR ER Galv Shear 7 Ann Arbor 6" 12" x x x none 11 Battle Creek 2 12 x #10x10'0" 9.5 Bay City 4 12 x x l/2"x8'0" 1 Detroit 16 18 x x x 1-1/4"x20'0" 4 Flint 8 18 x x x l/4"x10'0" 2 Grand Rapids 6 24 x x x 1/4"x12'0" 5 Jackson 10 12 x x 3/16"x10'0" 7 Kalamazoo 4 12 x x x none 3 Lansing 12 12 x x x 1/4"x10'0" 9.5 Muskegon 3 15 x x x none 7 Port Huron 1-1/2 24 x 5/8"x12'0" 12 Saginaw 1/4 none x x 3/16"x10'0" The following ratings were abritrarily established for each question. Plates: over 12" - 5 over 9" to 12" - 4 over 6" to 9" - 3 over 3" to 6" - 2 to 3" - 1 WFB: over 24" - 5 points 18" and 21" - 4 12" and 15" - 3 8" and 10" - 2 l 139 Sheets All three items - 5 Two Of three - 3 One of three — 1 points Shear 20'0" - 12'0" — 1/4" x 10'0" - 3/16" x 10'0" — others - points l—‘NbubUl The appropriate number of points was assigned to each service center. The points for each service center along with the points from the bar-product specialty scale were added together. The service centers were then ranked according to their totals. The resulting ranking is shown in the Table 9. Substituting this measure of product is the basis for a hierarchy, Hypothesis H1 was tested. The test yields a Spearman (rs) rank correlation coefficient of rs = .631. The null hypothesis, that there is no association between the hierarchy of function and the hierarchy of place, pap be rejected at the = .05 level. This test was made on the basis of population as the measure of hierarchy of place. The same hypothesis was tested using the Industrial Place hierarchy developed in the first section of this chapter. This test results in rs = .723. The null hypo- thesis is also rejected. This result is Significant at the = .01 level. The other measure of hierarchy of function used in testing this hypothesis is based on the composite measure of square feet of warehouse space, number of trucks used, 140 and number of salesmen. The data on which this test was based is included in Exhibit VII. The result of this test using population was rS = .179. This is not significant at the = .05 level, and the null hypothesis cannot be rejected. Testing on the basis of Industrial Place, rS = .495. This also is not significant at the = .05 level. Summarizing Hypothesis H1, the null hypothesis is: Ho: That there is pp association between the hierarchy of function of the highest- order establishment at a place and the hierarchy of place. Results of the test of this hypothesis are as follows: Hierarchy of Place Population Industrial Rank Place Rank Hierarchy of Function (Product-based) rs = .631* rS = .723** Hierarchy of Function (Size-based) rs = .179 rS = .495 Critical value ( = .05 level) rS .506 * - Significant. **- Significant at = .01 level. Hypothesis H2 This hypothesis is concerned with the relationship between the market territories of the twelve firms and: (1) their hierarchy of place (H2a), and (2) their hierarchy of function (H2b). The measure of market territory is made 141 on the basis of Square miles. The firms are then ranked according to this measure. Market territory data and rank are included in Exhibit VIII. H2a The null hypothesis of Hypothesis H2a is: Ho: That there is pp association between market territory rank as measured in square miles, and the hierarchy of place. This hypothesis was tested using both the pOpulation based hierarchy of place and the Industrial Place ranking. The result of these tests using the Spearman (rs) rank correlation coefficient are shown below. Hierarchy of Place Pppulation Industrial Place Market Territory rS = .541 rs = .70 Both results are significant at the = .05 level, and the null hypothesis can be rejected. H2b The null hypothesis of Hypothesis H2b is: Ho: That there is pp association between market territory rank as measured in square miles and the hierarchy of function from H1. Both the size-based and product-based measures of hierarchy of function were tested and the results of these tests are shown below. Size-based Product-based Market Territory r = .801 r = .631 142 Both results are significant at the = .05 level, and the null hypothesis can be rejected. The result using the size-based ranking is significant at the = .01 level of significance. Figure 18 is a map of Michigan with the territories of each distributor (except Detroit) superimposed upon it. Using the same type of summary as used in the grinding wheel section, Table 10 points up those central place aspects not evident from square mile figures or map display of these territories. DO 143 Figure 20 Sales Territories: Steel Bar Distributors f / I! I, ‘ f 1' f p l (I I , l I s \\ V ‘ 1 . 5 . y C \ n \\ \ '4 ‘ - , 2" I x H .1 31 - \ De it 0 \ \ 9’ , . \ \\ v ’ \, \ \- 144 Table 10 Market Territory Summary of Steel Bar Distributors Other Central-Places Population Distributor's In Market Territopy Rank Place Lower-Order Higper-Order 1 Detroit Flint, Lansing, Ann Arbor, Saginaw, Port Huron, Jackson, Bay City 2 Grand Rapids Kalamazoo, Muskegon 3 Flint Lansing, Ann Arbor, Detroit Saginaw, Port Huron, Bay City 4 Lansing Ann Arbor, Saginaw, Flint Kalamazoo, Battle Creek, Jackson, Bay City 5 Ann Arbor Jackson 6 Saginaw _Bay City Flint, Lansing 7 Kalamazoo Battle Creek 8 Muskegon - 9 Battle Creek — Kalamazoo 10 Jackson - Lansing, Ann Arbor 11 Port Huron - 12 Bay City - Flint, Saginaw, Port Huron 145 Six distributors consider other places of higher order to be in their market territories. All but two include other central places in the research area in their market territories. For additional insight regarding market territories, the same two questions asked of grinding wheel distributors were also asked of these steel distributors. The first question was: Whom do you consider to be your principal competitors from out Of town for your bar market here in (location of establishment)? Results of this question are shown in Table 11 in terms of the locations of the principal competitors. Table 11 Location Of Principal Outside Competition Steel Bar Distributors Place Compatitor Place Competitor Ann Arbor - Detroit Jackson - Detroit Battle Creek — Detroit Kalamazoo - Chicago Bay City - Detroit Lansing - Flint Detroit - none Muskegon - Chicago Flint - Detroit Port Huron - Detroit Grand Rapids - Chicago Saginaw - Flint The other question was: If one of your customers had an immediate requirement for a size or quality of bar which is not available from any source, where would you suggest he try (or where would you try for him) to find it? Results of this question are shown in Table 12. 146 Table 12 Location of Alternate Outside Source For Steel Bars Place Source Place Source Ann Arbor - Detroit Jackson — Detroit Battle Creek - Detroit Kalamazoo - Chicago Bay City - Detroit Lansing — Detroit Detroit - Chicago Muskegon - Chicago Flint - Detroit Port Huron - Detroit Grand Rapids Chicago Saginaw - Detroit In all_cases the distributor indicated he would offer to get the item for his customer, but would not be reluctant to suggest one or more possible sources in the location indicated. In all cases, except for Detroit, the source was another distributor unrelated to the respondent. In the case of Ryerson, in Detroit, the suggested source was the Ryerson plant in Chicago. Again note that in answer to both questions the distributor cited higher-order places. The foregoing completes the report of results of the research concerning the industrial distributor market structures. Part II, which follows, reports the results Of the industrial buyer research. Industrial Buyers Three (3) industrial buyers were interviewed in each of the twelve central places in the research area for a total sample of thirty-six (36). All of the respondents were 147 purchasing agents for medium-size manufacturing firms with between one hundred and five hundred employees. Also, each firm had a standard industrial classification (SIC) code between 33 and 38 inclusive. The data collected from these buyers consist of two parts as shown on the buyer interview form, Exhibit IX. The first part is general and is concerned with purchasing from all kinds of industrial distributors. The second part is specific in that it is directly related to the three control products of Part I of this chapter; grinding wheels, ball bearings, and steel bars. Each buyer was first asked to select three products or product groups which he regularly buys from industrial dis- tributors. He was told that three control products were being used and asked that these not be included in his selection. He was also asked that products be selected that involved different "kinds" of distributors, or, in other words, that each of the three products involve different sets of possible suppliers. Eight of the thirty—six could only think of two product groups and one had a contract with two distributors for all his industrial supplies. Thus, the total number of responses was ninety-nine products or product groups involving thirty- eight different classifications. Appendix X contains an alphabetical listing of the classifications selected. The buyer was then asked, of all his possible sources of supply, to select in order of preference the three most 148 probable sources of supply he would use for each Of the products. The data regarding the buyer's preferences were collected in terms Of place-name (cities) locations. This data is contained in Exhibit XI of the Appendix. The hypothesis, Hypothesis H3, being tested in this part Of the research is: H3: That the preferred sources of supply as perceived by the buyer will be located either at his same place or at some higher order place measured on pOpulation. Since each buyer is an element in the sample, there was a possible n = 36. However, eight (8) buyers selected all preferred sources at the same place as their own location; thus n = 28. The test is based on the following dichotomous measure: Category 1: A11 sources located at another place are at a higher-order place, or, Category 2: They are not. Using a X2 one-sample test of significance, as described in Chapter III, the null hypothesis is: Ho: That the preferred sources selected by the buyer which are located at some place other than his own are not related to the order of place as measured by population. The observed value X2 = 17.285 is significant at the = .001 level, and the null hypothesis can be rejected. One-hundred-nine preferred sources at other places were selected. Of these only three (3) were located at lower-order places. Figure 21 is a graphic description of these selections. 149 Figure 21 Purchasing Agents Preferred Sources: All Industrial Distributors Higher-Order Places:--D Lower-Order Places:'---O €HHCA6¢b 150 The data regarding the three control products was collected in the same terms as that described above. Each buyer was asked whether he bought grinding wheels, ball bearings, and steel bars from distributors, and, if so, to select his first three preferred sources. The data collected, along with graphical displays are contained in Exhibits XII and XVII. The data are summarized in Table 13. Table 13 Summary of Buyer Data on Control Products Preference Grinding Wheels n = 34 lst 2nd 3rd Higher Place 7 6 7 Buyer's Own Place 27 19 11 Lower Place Only one source - 9 Two sources - 7 Ball Bearings n = 33 Higher Place 3* 8 5 Buyer's Own Place 30 6 2 Lower Place - - - Only one source — 19 Two sources — 7 * - All three are located at Port Huron where there is pp bearing specialist distributor. Steel Bars n = 35 Higher Place 11 22 16 Buyer's Own Place 24 10 6 lower Place — — - Only one source — 3 Two sources - 10 151 Several notes regarding these data are pertinent at this point: 1. In pp case was a supplier from a lower order place selected. 2. Regarding buyers' lst preference selections: - For ball bearings, all were local except Port Huron. - Approximately 20% selected higher order places for grinding wheels. — Approximately one-third selected steel bar distributors at higher-order places. 3. Regarding sole sources of supply: - 58% of the buyers use only one source for ball bearings. — 26% use only one source for grinding wheels. - Less than 9% use only one source for steel bars. Summary This concludes the report of the research results. Chapter V, which follows, is devoted to the analysis of these results. CHAPTER V ANALYSIS OF RESEARCH RESULTS This research is founded on central-place theory. Essential to that theory and to the research hypothesis drawn from it is the concept of a hierarchy Of places. In each of the three hypotheses, the measure of the hierarchy of place is actually the independent variable. Since this measure constitutes such an integral part of the research, it is necessary to analyze the concept and the measure of it before proceeding with the analysis of the data and tests of hypotheses. The concept of a hierarchy of places is not a compli- cated or complex one. It is, however, an abstract concept which is very difficult to measure in real terms. The problem is further complicated by the spatial relationships of one place to another. The concept is based on the existence of highest—order places, second-order places, third-order, and so forth. These are discrete measurements. But, places as they exist in reality when measured on any basis describe a continuum rather than readily identifiable discrete groupings. Within the research area, Detroit by all measures is the highest- Order place. Grand Rapids by pp§p_measures is the 152 153 next-ranking place; and because Of its geographic location in relation to other places, particularly Chicago and Detroit, it is apparently a second-order place in the research area. Beyond these first two places, the distinctions become very difficult. In the first abstraction from the map dis- cussed in Chapter III (Figure 15), Flint is positioned as the same order of place as Grand Rapids. And yet, the Industrial Place measure developed in Chapter IV finds Lansing ranked above Flint. To avoid the problem created by the lack Of discrete measurements it has been a common convention in central-place theory analysis to use relative rankings as the measure of the hierarchy of place. This is the measurement method employed in this study. While it is not rigorous application of the theory, it does avoid arbitrary assignment to a particular rank-order grouping. Other weaknesses must also be acknowledged in the application of the theory. The model postulates a homo- geneous plane. Topographic features conflict with this postulate. The problem is further complicated by man. Par— ticularly in this country, pOpulation can be viewed in terms of growth and flow. The general east to west flow works against the notion of the homogeneous plane. Growth rates affected by many factors such as resources, entrepreneurship, accident, etc., further distort the basic hexagonal model. 154 Political boundaries have a significant effect. For example, the Detroit River as an international border stands as an impenetrable barrier to trade in the research study. No distributor studied has a sales territory in Canada, and no buyer buys from Canada. So far as city limits are con- cerned, Saginaw, Bay City, and Midland should perhaps be treated as one central-place. Finally, economic decisions (and accidents) have a great impact. Flint may be the size it is strictly because of General Motors. Probably Saginaw or Bay City should have developed as the other second-order place in the research area. All of these factors affect the measures of hierarchy of place. The point Of this discourse is to acknowledge that the measure of hierarchy of place is imperfect at best. While the ranking technique avoids the problem of discrete measurements, other factors are at work to distort measure- ments of place. However, the goal is not the validation of central-place theory. It is, instead, a test of the use of the theory as a conceptual framework for the analysis of certain aspects of economic activity over space. Couched as it is in terms of three working hypotheses, the research is analyzed in light of the results relevant to each Of these hypotheses. 155 Hypothesis H1 Central—place theory postulates that the hierarchy of function is directly related to the hierarchy of place. Two measures of each distributor (function) were taken: (1) product specialty scale, and (2) composite size. The hypothesis was tested by determining the correlation between the hierarchy of place and each of these two measures. Two measures of the hierarchy of place were tested Since it was felt a ranking based on population might not be the best measure in the industrial setting. The second measure is referred to as Industrial Place ranking and is a composite measure of number Of manufacturing employees, number of manufacturing establishments, and total merchant wholesaler sales. Product Specialpy Scale Measure. As reported in Chapter IV, this measure was not applied in the case of ball bearings. In the two cases where it was applied, grinding wheel distributors and steel bar distributors, there is absolutely pp_relation between rankings on this measure and the ranking of places on either basis, population or Industrial Place. Seven of the twelve general-line industrial distributors carry a complete line of the grinding wheels on the scale from most common to least common. Nine of the twelve general-line steel distributors also carry a 156 complete line. In the face of such results, one must ques- tion both the measurement instrument and that which is being measured. The scale of grinding wheel items was constructed on the basis of personal estimates by large Detroit distributors of sizes ranging from most common to least common sizes. While there was general agreement among the three, no hard data was available. The scale for merchant bars was constructed using actual current data. The last two items in the scale are, indeed, very uncommon. It would be very surprising if the majority of the distributors stocking these sizes received orders for them more than several times a year. While there might be an exception at one or two locations due to a particular end user, it is highly unlikely that this condition is common to all nine. Several distributors, when asked directly why they carried such uncommon items in stock, all responded that they had to because their competition did. None denied that they were uncommon sizes. One general—line distributor being questioned about his grinding wheel stock cited this as the major problem facing all distributors in that too much money is tied up in inventory which "never" moves. A better measure of this dimension would have included data on the amount of inventory on hand for each item, and the number of times in the past year the item was ordered. 157 Although it would be a better measure, it is doubted that even half of the distributors of either product could have provided the data. In four specific cases the response was to the effect that the respondent was sure he had the item in stock but wasn't sure where or how much. In only three instances were computer printouts of inventory in evidence; and, two Of these were Detroit distributors. The Kardex file system appears to be the general rule for distributors in all three types of distributors studied. It therefore appears that the majority of grinding wheel and steel distributors are attempting to maintain 100% service levels in these two products. Two factors appear to contribute to this situation. First, most are independent single-facility operations with no way to con- solidate inventories. Second, because they are independent and relatively small, the adOption of modern management science techniques has not occurred, as evidenced by existing inventory control methods. In addition to having no relation to placevrank, there is no distinguishable ranking of the measurements on the product specialty scale. Central-place theory suggests a pyramidal result. The results would indicate that there is no hierarchy of function involved. However, the additional measure of steel distributors on a product basis did yield a clear ranking; and, furthermore, that ranking is directly related to the ranking of place. 158 Comppsite Size Measure. The results of the test of Hypothesis H1 using the composite size measure are sum- marized below. Both the coefficient of rank correlation and whether or not (yes-no) the data support the hypothesis are Shown. Hierarchy of Function (Composite Size Measure) General-line General—line Industrial Steel Hierarchy Distributors Ball Bearing Distributors Of Place (Grinding Wheels) Specialists (Steel Bars) Population rS = .274, no r8 = .889, yes rS = .179, no Industrial rS = .41, no rs = .921, yes rs = .495, no Place There is an apparent paradox in these results. Although there is very high correlation for ball bearing distributors, it will be shown later that there is probably no hierarchy of functions among these distributors. The central-place logic is that if there is a hierarchy of functions, then the higher order functions will be larger. This result suggests that even in absence of a hierarchy of function there will be a relationship between size of establishment and place-rank. Two aspects of the application Of this measure to general-line industrial distributors deserve comment. A fifty-year-old, five-story warehouse with 100,000 square feet cannot be compared with a brand new 18,000 square foot high- cube warehouse on the basis of square feet. This does not measure the same thing. Questions of modernity, efficiency, and cost are involved. 159 The second aspect involves the number of trucks used. In this particular business there is a high reliance on delivery or parcel service. Thus, the distributor in Detroit, where such service is readily available, uses only six of his own trucks whereas the Bay City distributor uses thirteen. With regard to the measure of steel distributors, there is a problem with overlap. Four of the twelve are also general-line industrial distributors. Their salesmen sell both steel and industrial supplies and their trucks are used to deliver both. By taking only the eight steel distri— butors which are strictly steel service centers an entirely different result is obtained. The rank—order correlation coefficient is rS = .923 using either measure of place, population, or Industrial Place. There is no doubt that sales figures would provide a much more appropriate measure of size. But, Since most of these distributors are privately held concerns, there is great doubt that any would divulge such information. Hypothesis H2 Both of the tests of this hypothesis are concerned with the measure of market territory in square miles. This is actually another measure of hierarchy of function since central-place logic as expressed in the hexagonal diagram defines the order of function according to the size Of its market territory. 160 HES: The first test using this measure, Hypothesis H2a, is identical to Hypothesis H1 in that it tests the relation- ship of market territory as a measure of hierarchy of function with hierarchy of place. The results of this test are sum- marized below in the same form as in the analysis of Hypothesis H1. Size of Market Territory General-line General-line Hierarchy Industrial Ball Bearing Steel Of Place Distributors Specialists Distributors Population rS = —.007, no r8 = -.20, no r8 = .541, yes Industrial Place rS = .07, no r8 = .08, no rs = .700, yes The first fact which must be kept in mind in the analy- sis of these results is that this involves a measure of what the distributor perceives to be his territory and, in fact, directs his salesmen to cover. As was noted in Chapter IV, several distributors in both the steel distributor and indus- trial distributor categories consider higher-order places to be a part of their market. And yet, in pp case was this sub— stantiated in the buyer interviews. This indicates that the distributor overstates his effective market and this is reflected in the measurements. There are two other distortions of this measure which must be acknowledged. Both are related to the postulate of the homogeneous plane. First, one hundred square miles in 161 and around Detroit are not the same as one hundred miles in and around Muskegon in terms of market potential. Second, those distributors proximate to the rural hinterland to the north of the research area are inclined to include much of that area in their market territories. Because of these two factors, it is expected that, for an absolutely homogeneous function, there would be a negative correlation between a measure of market territory and hierarchy of place. This, however, is a contradiction in that it forces a ranking where one, in fact, does not exist. The negative result in the case of ball bearings supports this contention. The market territory data indicate that this is a low-order function with regard to the twelve cen— tral places involved in this study. The market territory for each ball bearing distributor includes only the central—place at which it is located. The one exception is Port Huron where there is pp distributor but which is served by a north- east suburb of Detroit. Furthermore, the average size of market territory of the ball bearing distributors at all eleven other places is 2,350 square miles. This same effect is no doubt reflected in the measures of the general-line industrial distributors and steel dis- tributors. However, the negative result for the industrial distributor cannot be interpreted as a lack Of a hierarchy of function. Seven of the twelve include one or more of the other central places. And, the average of the market terri- tories is 6,000 square miles. As a matter of comparison 162 ten of the twelve steel distributors include one or more Of the other central places; and, the average Of the market territories is 12,200 square miles. In light of the analysis of steel distributors under Hypothesis H1, the same eight steel distributors which are not also industrial distributors were tested under H2a. The results were rS = .928 using population, and r5 = .952 using Industrial Place. Thus, among these eight there is a very high correlation between market territory and hier- archy of place. ng, The second part of this hypothesis involves a test of the relationship between market territory and the hierarchy of function as measured by size from Hypothesis H1. The results of this test as reported in Chapter IV are summarized below: Market Territory General-line Industrial Ball Bearing Distributors Distributors Hierarchy of rS = .41, no rS = -.15, no Function from H1 Steel Bar Eight* Steel Distributors Distributors rs = .631, yes r8 = .90, yes * - Eliminating those which are also general-line industrial distributors. 163 Since both measures have been previously analyzed, the results Speak for themselves. 'Comparative Analysis of Distributors Grinding wheel distributors. The market structure as represented by these twelve distributors is not consistent with central-place theory. The data indicate a hierarchy of function, but neither measure of it, size or market territory, is related to place hierarchy. Furthermore, the two measures are not related to each other. Ball bearing distributors. If there is no hierarchy of function among the ball bearing distributor establishments, then this market structure is entirely consistent with central-place theory. However, since both H1 and H2 postu- late the existence of a hierarchy of function they are not appropriate to the analysis of this market structure. Steel distributors. The results of the research regarding the steel distributor market structure are not conclusive. The evidence, however, is strongly weighted in support of the contention that this market He consistent with central-place theory. While the results of testing Hypothesis H1 are not significant using either the product specialty scale or composite size measure, the test of this hypothesis using the more elaborate product measure does 164 yield significant results. Results Of testing both H2a and H2b app significant. Furthermore, by eliminating those four distributors who are also industrial distributors yields very high coefficients of correlation supporting H1, H2a, and H2b. The Industrial prer Hypothesis H3 The hypothesis being tested here is that when a buyer selects a source of supply located at some place other than his own he will select a source located at a higher—order place. The statistical test of the data supports this hypothesis. Of even greater significance is the fact that only three lower-order places were selected out Of the total one-hundred-nine sources located at other places. Thus, one—hundred—six were located at higher-order places. The map display of this phenomenon is a dramatic example of central-place theory in reality (Chapter Iv, Figure 21, page 149). The results related to each of the three con- trol products are even more convincing. In pp_case was a lower-order place selected. The implication from central-place theory is that when a higher-order source is selected then a higher—order func- tion is involved. While there is no doubt that this is an important factor in the results obtained, there is another 165 factor which must be acknowledged. This factor is the desire on the part of many buyers to have multiple sources of supply for a variety of reasons. This, however, does not detract from the results since when buyers do seek addi- tional sources outside their place they go to higher-order places. Two particular facets of the buyer's behavior as reflected in the data deserve comment. First, and most important, is the contradiction between buyer behavior and the distributor's perceptions of their market territories. Three grinding wheel distributors and six steel distributors consider higher-order places to be in their market territories. Yet, in no case is this substantiated by the buyer data. Second, the buyer data on the three control products, summarized in Chapter IV, are entirely consistent in their relationship on the following three measures: (1) total num- ber of higher-order sources, (2) the number of buyers who use only one source of supply, and (3) the number of buyers whose first preference is local. Ball bearing buyer data is lowest on the first measure and highest on the other two. The steel bar buyer data is just the opposite. And, the grinding wheel buyer data is in between on all three measures. This consistency plus the magnitude of the data indi- cates that there is very little, if any, hierarchy Of function in the stocking of ball bearings as perceived by the buyer. Furthermore, the fact that outside sources are 166 cited pp ail may well represent a desire on the part of the buyer for more than one source of supply. The question itself, asking for the first three preferences, may have forced a selection of more than one source. Or, the buyer may have responded with more than one preference to convey the fact that he is aware of alternate sources. The data support the probability that outside ball bearing sources selected represent multiple sources rather than higher-order needs. In all places where there is a second source available (Michigan Bearing Company with branches in Grand Rapids, Flint, Kalamazoo, and Saginaw) only one outside source was cited. This was a Kalamazoo buyer who cited South Bend as his third preference. Since it was not Detroit or Chicago it is speculated that the choice of South Bend represents a unique situation perhaps involving a particular brand or a personal relationship. On the same bases of consistency and magnitude of the data it seems quite clear that there ipja hierarchy of function among steel distributors as perceived by the buyers. Also, by referring to the map plot of these preferences, Exhibit XVIII in the Appendix, it is clear that Chicago and Detroit are highest-order sources. Grand Rapids and Flint stand out as second-order sources; and the selection of Lansing by a Battle Creek buyer indicates it may be of slightly higher-order than the other places. Comparing the grinding wheel buyer data to the other two sets of data leads to the deduction that the buyer J l . Illllllll | 167 perceives the stocking of grinding wheels as a function with a hierarchy somewhere between ball bearings and steel bars. However, a close look at the data, especially as dis- played on the map plot, Exhibits X11 and XIII, indicates this set of data is much more closely related to ball bearing buyer data. In fact, there appears to be a high probability that the buyer does not see any hierarchy of function among the general-line industrial distributors who stock grinding wheels. In two places, Flint and Muskegon, pp outside source was cited. In each of five more places, Grand Rapids, Jackson, Kalamazoo, Lansing, and Saginaw, only one outside source was cited. Fifteen of the thirty-four buyers (44%) feel their grinding wheel requirements can be met locally. It is possible that some of the outside sources cited reflect a brand preference either for a brand not carried locally or for a brand carried locally but exclusively thus neces— sitating an outside source. That fact that seventeen of the thirty-four buyers (50%) use two or three sources in their own city supports the contention that this is a homo— geneous product. It also indicates a desire for multiple sources. In both Port Huron and Ann Arbor where there is only one general-line distributor, the overwhelming preference for Detroit sources probably reflects this desire for mul- tiple sources. It is possible that those buyers who cited Detroit as an outside source may have had reference to grinding wheel 168 specialists in that city. If so, this would have to be considered a higher-order function although not within the general-line industrial distributor classification. Since the identity of sources is not available this speculation cannot be resolved. But, it must be acknowledged that there may be some slight degree of a hierarchy of function as perceived by the buyer of grinding wheels. Finally, there is one phenomenon which is not explained by the data and for which there is no basis for analysis. In the case of the grinding wheel buyer data it appears that Bay City is "tied" to Saginaw and Battle Creek is "tied" to Kalamazoo. These are the only cases where the outside sources are other than Detroit. This would indicate that proximity imposes some degree Of hierarchy on a function. And yet, in looking at the steel bar buyer data where a hierarchy apparently does exist, pp buyer in Bay City cited Saginaw and pp buyer in Battle Creek cited Kalamazoo as outside source preferences. With a bare modicum of data this part of the research provides considerable insight into industrial buyer behavior. It seems to follow that further research in greater depth might produce much greater insight into this behavior. 169 Summapy This chapter has presented the analysis of the research results Obtained. The final chapter, which follows, presents the conclusions drawn from the research along with comments on the contributions of the research, its limitations, and suggestions for future research. CHAPTER VI CONCLUSIONS This research was undertaken to study the spatial dimension of marketing. Specifically, the study is focused on the spatial relationships in the marketing of industrial products through industrial distributors. The study is dual-focused considering both the industrial distributors and the industrial buyers who buy from them. The objective of the research is the analysis of spatial relationships. This chapter is devoted to assessing the results of the research in light of that objective. In addition, specific comments and conclusions are made regarding the products and distributors involved in the research. Al- though neither the products nor the distributors were the objects of the study, it is felt that these additional com- ments add substance and value to the findings. Primarpronclusions The central research question asks whether central- place theory provides a useful framework for the analysis of spatial relationships. Analysis of the research results leads to the following two primary conclusions: 170 171 I. That industrial buyer behavior is entirely consistent with centralhplace theory, and that the analysis of buyer behavior on the basis of central—place theory provides significant insight for practical decision— making and for theory development. 11. That the spatial relationships of the three market structures are not consistent with central-place theory; or, that central—place theory does not pro— vide a description of the spatial relationships of these market structures as they actually exist over space. While this conclusion stands in direct con- flict with the first conclusion, further analysis of the results of the market structure research on the basis of central-place theory yields insight into the causes of the apparent lack of order expected under the theory . Discussion and Corollary Conclusions The two primary conclusions which are stated above form the basis for this section. Both conclusions are discussed in detail, and corollary conclusions to each are presented. Buyer Behavior. The first primary conclusion actually consists of two parts. First, it is concluded that industrial buyer behavior is consistent with central-place theory; and, second, that using central-place theory as a framework for analysis does provide Significant insight for decision- making and theory development. With regard to the first part, that the buyer's behavior is consistent with central-place theory, the evidence is con— vincing. For the "all—other-products" category, one-hundred- six out of one-hundred-nine outside sources - as opposed to local sources — are located at higher-order (larger) places 172 than the place at which the buyer is located. For the three control products all outside sources are located at higher-order places. Conclusions corollary to this part are the following: I01. That the buyer does "look up". He may have many reasons for using an outside source: no local source; local source unacceptable; a higher-order need (specialized) than local sources can supply; or, a desire for multiple sources. But, regard- less of the reason, when the buyer does select a source of supply outside his own place, he con- sistently selects a source at a higher—order place. This finding is not surprising in light of the previous work done by Berry, referred to earlier, which is focused on consumer behavior.1 As such, this research stands as a complement to his efforts. That the number of outside sources used by a buyer provides a basis for evaluating the hierarchy of a function. At the one extreme, if all buyers' needs for a particular product or product group can be satisfied by local sources, it follows that those establishments satisfying that need are homogeneous with regard to that product or product group. There 1B. J. L. Berry, Geogpaphy of Market Centers and Retail Distribution, pp. cit. 173 is, therefore, no hierarchy of function. In Christaller's terms, this would be a low order function found at all places.2 At the other extreme, when a clear hierarchy of function does exist, some needs for a particular product may be met locally, but higher-order or more special needs can be met only at higher-order places. The buyer's responses regarding ball bearing suppliers is evidence of the lack of a hierarchy of function. In all cases the first preference was a local source and the total number of outside sources was relatively low. On the other hand, almost one-third of the first preferences for steel bar sources were outside sources; and well over half of all three preferences were outside sources. While it is recognized that outside sources are Often the result Of a desire on the part of the buyer to use more than one source of supply rather than as a result of a hierarchy of function, it appears that the two are directly related. The data indicate that where there is a clear hierarchy of function there is also a greater tendency to 2W. Christaller, Die zentralen Orte in Suddeutschland, op. cit. 174 use more sources of supply. The mean number of sources cited for steel bars is 2.54, while for ball bearings a mean of 1.64 sources are used by each buyer. The actual difference is probably considerably higher since it is likely that many of the 22 steel buyers who cited three preferences may well actually use four or more. Only seven of the thirty-three ball bearing buyers use three sources . The second part of first primary conclusion is that using central-place theory as a basis for the analysis of buyer behavior provides significant insight for practical decision making and for theory development. So far as decision making is concerned it seems that two conclusions follow from the results of the study of buyer behavior. I03. In evaluating marketing performance, the indus- trial distributor should be aware of the fact that buyers do "look up". If the distributor is attempt- ing to sell in a higher—order place than his own he should very carefully consider the marginal cost involved in this effort since this fact indi— cates that the probability of market penetration in the higher-order place is greatly reduced. In planning marketing strategy this insight into buyer behavior may be a crucial determinant of success. First of all, this result indicates 175 that the evaluation of possible market penetration is directly related to the actual or planned loca- tion of the decision-maker's facility. Should it be essential that a large proportion of a company's volume come from a high-order place, then it would probably be a mistake to locate a facility in a lower-order place. In other words, one Should not attempt to carry on a high-order function in a low-order place. An example of a failure to take this into account high- lights the relevance Of this conclusion. Several years ago a regional steel service center chain established a general-line service center in Dayton, Ohio, to serve the rich industrial crescent of the Miami River Valley running from Cincinnati through Hamilton, Middletown, Dayton, Springfield, and Columbus. Their major competitors, the national and regional chains, were all located in Cin— cinnati. Within a relatively short time the Dayton facility was closed and the Operation moved to Cincinnati, the highest- order place in the market area. Mistakes like this are costly. Had the management of this firm been aware of the fact that industrial buyers do "look up", this error might have been avoided. The first primary conclusion also provides insight for the development of marketing theory. 176 1.5. In the development of gravity models for the pre— diction of buyer behavior, such as the Huff model,3 or in theoretical constructs involving buyer behavior such as Devletoglou's,4 the hierarchy of place must be taken into account. With regard to the Huff model, the results of the research indicate that the effect of distance in determining the probability of purchase must be tempered by the effect of place hierarchy on that probability. Likewise in the Devletoglou construct, the "minimum sensible" distance may be dependent on the relative rank or size of the two places under consideration. The result, in fact, may be two hyperbolae which are not symmetrical or even similar. Market Structure. The second primary conclusion is that the spatial relationships of the three market structures are not consistent with central-place theory. In none of the three market structures studied were significant results obtained in the tests of pppp the hypotheses tested. The conclusion must, therefore, be that the spatial relationships of the three market structures are not consistent with central-place theory as formulated in the research design 3D. L. Huff and L. Blue, A Proqpammed Solution for Esti- mating Retail Sales Potential, op. cit. 4N. E. Devletoglou, Economica, May 1965, Op. cit. 177 and in terms of the hypothesis tested. The important question, then, is what do the results of this research on the market structures mean? First of all, does it mean that central-place theory does not provide a useful framework for the analysis of market structures? II.1. It is concluded that the failure of central-place theory to provide a description of spatial rela- tionships of market structures as they actually exist over space does ppp_mean that the theory has failed as a framework for analysis. In support of this conclusion stand the results of the research on buyer behavior which are in direct conflict with the market structures studied. On the buyer side of the market central-place theory provides a description of existing order. And yet, on the seller's side there is ppp corresponding evidence of order. For example some general line industrial distributors and some steel service centers in low-order places consider higher-order places to be a part of their market territories in spite of the evidence that buyers do not consider distri- butors in lower—order places than their own as sources of supply. Thus, the failure is not with the theory, but with the management of the distri- butors in not adjusting their marketing efforts to the behavior of the buyer. 4's" . IIII‘I ((( Ill: .. Ill. .. i I'll-all 178 The question, therefore, becomes: Why does this apparent lack of order in the market structures exist? The following corollary conclusions drawn from this research are presented in answer to this question. II.2. That for some functions there is in fact pp_hier- archy. The conclusion is that two of the types of distributors (market structures) studied are in this category, the general-line industrial distributors and the ball bearing specialists. The conclusion that there is no hierarchy of function among ball bearing specialist establish- ments is supported in the market structure data and in the buyer behavior data. Each establish- ment is about the same size and the market area for each one is restricted to its own place. Furthermore, the evidence cited in the previous section of this chapter concerning buyer behavior shows that the first preference for al$_ball bearing buyers is a local source. And, for over half of them one source is adequate. The conclusion regarding the general-line industrial distributors is not so clear-cut. The most telling argument here is that the largest Detroit distributor covers only the Greater Detroit area and the second largest distributor likewise restricts its market territory to the same area. This stands in conflict with some of those 179 distributors in other places who define their market territories to include great expanses of space as well as other central places of both lower and higher-order. It may well be that their effective market territories are not accu— rately described by the extent of their salesmen's coverage. A more effective measure of market territory might yield significantly different boundaries. Actual sales or percent of actual sales would certainly be a better measure; but it is doubtful that most could reference their sales geographically except by salesmen's territory, and even if they could it is doubtful that the infor— mation would be available to the researcher. Further substance to this conclusion is con- tained in the answers of distributors to the question: Whom do you consider to be your princi- pal competitor from out of town for your grinding wheel sales? Three of the eleven outside of Detroit (Port Huron, Flint, and Ann Arbor) indi- cated Detroit. But five of the remaining eight said they had pp outside competition. In contrast, all of the steel service centers indicated that they do have outside competition - for their local steel bar market. Thus, in spite of market terri- tory and size measurement to the contrary, there is evidence to support the conclusion that a 180 general-line industrial distributor function is a homogeneous function wherever it is located. The lack of a hierarchy of function has two important meanings for this research. First, the research design assumes a hierarchy of function. But, to attempt to force a ranking where one does not exist yields meaningless or spurious results. There is some indication that one should expect a negative correlation of market area ranking when there is no hierarchy of function since the estab- lishment in a higher-order place enjoys a more highly concentrated market than one in a lower- order place. Furthermore, market area may be correlated with proximity to the hinterland regard- less Of the hierarchy of place. Second, the converse of this conclusion is that in the steel distributors market structure there is a hierarchy of function. That there is a clear hierarchy of function among steel distri- butors is confirmed by significant results in the tests of both hypotheses, H1 and H2, when using the product-based measure of a hierarchy of function. This is significant in that previous work in central-place theory has assumed that a function ip_homogeneous. Of interest has been only whether or not a function was present, and sometimes additionally in how many establishments. II.3. 181 This research demonstrates the existence of a hierarchy within a function. While this one small bit of research does not add to the theory, it does provide a basis for further research. It is concluded that some amount of the lack of order in the results of the research is due to error in the measurements of the establishments for the tests of Hypothesis H1. The failure of the measure of the degree of specialization of product inventory was discussed in the preceding chapter. An alternate measure such as the product- based measure of steel distributors may have pro- vided an adequate measure of the other types of distributors. With regard to the size measures of warehouse size, number of trucks used, and number of outside salesmen, the principal distortions occur in measurements involving general-line industrial distributors. There is a very high correlation between the size measure of ball bearing special- ists and hierarchy of place (rS = .889). And when those steel distributors who are also general- line industrial distributors are excluded from the ranking of steel distributors there is an equally high correlation between the size measure and hierarchy of place (rS = .923; significant at = .01 for n = 8 when r .833). These two S 182 results are due to the fact that the same kind of establishment is being measured in each case. But in the case of the general-line industrial distributor the same kind of establishment was ppE_measured. Some are also steel distributors. Some are additionally hardware distributors, construction-product distributors, and even retail outlets. For example, the George Worthing- ton Company in Lansing with approximately 100,000 square feet of warehouse space is certainly not the same kind of establishment as the Grand Rapids Supply Co. with only 18,000 square feet. Even though both carry about the same industrial pro— duct lines, Grand Rapids Supply is strictly an industrial distributor whereas Worthington is also a hardware distributor. Thus, each distributor should have been measured on a common basis directly related to industrial products alone. The number of square feet devoted to industrial products probably would be appropriate but extremely difficult to determine. Since the trucks used deliver all products this is not an appropriate measure of the general-line industrial distributor. The number of outside salesmen is equally inappro- priate Since they too are involved with all or most product lines. 11.4. 183 However, even in light of this measurement error, there is such wide variation in the general— line industrial distributor market structure that other factors must be responsible for a considerable amount of the lack of order indicated by the research results. Two factors stand out as contributing to the lack of order in this particular market struc- ture. They are: (l) the lack of horizontal inte— gration, and (2) the existence of exclusive distributorships. These two factors are the basis for corollary conclusions. One further observation relative to the results of the measures of size is apprOpriate at this point. The fact stands that there is a very high correlation between size rank and place rank for both the ball bearing specialists and those steel distributors who are only steel distributors (not engaged in other business such as general-line industrial distribution). The conclusion is that one should expect a correlation between establishment size rank and place rank whether a hierarchy of function exists or not. It is concluded that the lack of horizontal inte— gration contributes to the lack of order in the market structure of general-line industrial dis- tributors. Here the term horizontal integration means ownership or contractual arrangement. In 184 other words, there is a lack of distributor "chains". The mass marketing of a large number of product lines with many items in each line through indepen— dent "stores" is an anachronism at this point in time. The day of the independent general store has long since passed in retail marketing; and yet, nine of the twelve general—line industrial distri- butors are independents. The other three may represent the beginning of an eventual chain. This conclusion is basically related to man- agement capability and resources. When a distri- butor is an independent he is an island unto himself. Decisions must be taken without the advantage of comparison to establishments in other places. Management resources are limited to those people within the establishment itself. Each must plan his own marketing strategy, decide what product lines to carry, what kind of promotion to employ, what his target markets are, what territory to cover, how many salesmen to use, and so on. In short, each decides what kind of business he is in. The result of independence is considerable varia- tion both in the management and the decisions taken. If order does exist it is probably a function of management ability and not of spatial relationship. Such a situation carries a high cost to the distributor. The independent distributor has no 185 way to consolidate inventory. Because of small size and lack of management resources, inventory control methods of independent distributors appear to be deficient at best. Centralized purchasing is not possible. Although this study did not consider the purchasing function, the thought that each must independently maintain accounts with each of his suppliers is, indeed, staggering. Viewing the market structure in the aggregate the duplication of inventories must be considerable. And, the amount of time and money wasted by the distributors and their suppliers in the sales/ purchase relationship suggests excessive costs in the total distribution process. The rigid price control by suppliers which appears to exist in this industry has provided a protective blanket for poor and archaic management. This industry has never really experienced compe- tition. This situation of dominance by indepen— dents is very similar to that which existed fifty years ago in retailing. Once the equivalent of Rexall, A&P, or Sears becomes a reality in this field the independents will either become limited- 1ine specialists or disappear entirely. It should be noted at this point that, while the steel distributors outside of Detroit are also all independents, there are three distinct l. I." ‘. 186 differences in their market structure that create an entirely different situation. First, this structure i§_integrated at the top. There are already national and regional chains which are continually expanding their scope of operations to include lower-order places. I would predict that before very long one or more of the national chains will establish operations in Grand Rapids. One major weakness of the national chains has been their inability to distinguish the hierarchy of functions. Unlike Sears with their "A", "B", and "C" sized stores they seem to limit their scope to A-size branches. This I feel is changing. Second, the nature of the operation of a steel distributor is quite unlike that of the industrial distributor with regard to product. The steel distributor is a specialist, his product line quite limited, even including nonferrous items. There is no comparison on the basis of number of items carried. The product is completely homoge- neous with no branding, and a very few suppliers (mills) can supply all of a distributor's needs. The steel distributor's management problems with purchasing and inventory control are practically nil when compared with those of the general-line industrial distributor. II.5. 187 Third, the fact that a hierarchy of function does exist is in itself integrative. The distributor in a lower-order place is related to the distri- butor in a higher-order place in that he realizes certain higher-order needs which he cannot afford to carry must be met by distributors in the higher- order place. Furthermore, the large distributors +~J in Detroit and Chicago are actually another source of supply for the distributors in lower order —V ‘m-Z‘, places. Hence, a hierarchy of function of itself imposes order on spatial relationships. Finally, it is concluded that the existence of exclusive distributorships for homogeneous products like grinding wheels tends to compound the prob- lems leading to a lack of order in the market structure of general-line industrial distributors. The point here is not to argue the merit or lack of merit in the existence of exclusive distri- butorships; it is, instead, to point out their effect on the distributors and the market structure. First, they provide additional protective shelter from competition for poor management. And, second, they tend to inhibit horizontal integration since entry by a chain into a new market would be pre- cluded if an exclusive distributor for their brand already existed in that market. It is interesting to note in this regard that the one developing 188 chain in this area carries Avco—Bay State grinding wheels, not Carborundum or Norton. Looking to past experience in retailing, it is expected that as chains develop there will be an emergence of private labels. The distributor will be attempting to convince his customer that he can serve his needs. He will be serving his pal customers rather than serving his supplier. i While this is a subtle distinction it will involve S‘ a shift in channel control from producer to distri- butor. The alternative for the producer is to undertake the distribution and selling functions himself. In the case of grinding wheels it seems unlikely that specialists like the ball bearing specialists who carry all brands will develop. Only a small percentage of total ball bearing sales are made through distributors so that the risk of market position is small. However, since the great majority of grinding wheel sales are made through distributors, market position is heavily dependent on the distributor. A change nevertheless seems inevitable, and the result should be more order in spatial relationships. Summarizing the foregoing conclusions on the market structure research, it is concluded that, even though central- place theory does not provide a description of spatial rela- tionships as they actually exist over space, it still can be 189 a useful basis for analysis by providing insights Of both practical and theoretical value. The insight provided by the market structure research are: 1. That there is an apparent conflict between what the seller perceives to be his market territory and actual buyer behavior. That for some functions there is np_ hierarchy within the function, while for others there is, indeed, a hierarchy of function. That the establishment is not a good measure of function, but that when the establishment is the equivalent of the function there is indication that ranking on the basis of establishment measure will correlate highly with place ranking whether or not a hierarchy of function exists. That the lack of order in the general-line industrial distributor structure is due in part to the lack of horizontal integration resulting in poor management and poor management decisions. That the existence of exclusive distributorships further contributes to the lack of order in this particular market structure. First, it acts as an obstacle to horizontal integration. And, second, the exclusive distributorship plus the rigid price maintenance by manufacturer provides a shelter from competition for poor management. Contributions of the Research By far the most important contribution of this research is the insight into industrial buyer behavior which it pro- vides. Even though the emphasis in the research design is heavily weighted on the side of market structure analysis, it is not unusual that the most significant research contri- bution results from secondary or subsidiary effort. The fact that the buyer does "look up" carries implications of 190 considerable importance for both the practitioner and academician in marketing. For the practitioner this aSpect of buyer behavior has a direct bearing on planning new facilities and marketing strategy, and on the evaluation of marketing effort. For the academician it adds a new dimen— sion to the study of industrial buyer behavior. The results concerning industrial buyer behavior serve as a supplement $" to Berry's work on consumer behavior.5 It is felt that this research offers four other contri- F butions. Two are specific and two are general. The two specific contributions are: l. The study highlights the area of industrial distri- bution, particularly the general-line industrial distributors, as a marketing anachronism. It is a paradox in this era of mass marketing and the mar- keting concept that market structure characterized by small independent entrepreneurs should still exist. In this area the general store is EEili a reality. This contributes to marketing by pointing out a fertile area for study and development. 2. The research makes a contribution to central-place theory by suggesting that there may be value in considering the concept of a hierarchy within a function as demonstrated in the case of the steel 5B. J. L. Berry, Geography of Market Centers and Retail Distribution, op. cit. 191 distributors. This could add a new dimension to future central-place studies regardless of the nature of inquiry, i.e., geography, sociology, business, etc. This research is concerned with high-order places. It is possible that this con- cept might be even more appropriate when consider- ing places on the low end of the hierarchical scale. The two general contributions are directly related to the goals of the research which were established at the conclusion of the initial chapter. 1. The attempt to search for order in spatial relation- ships in marketing is in itself a contribution to the marketing discipline. Space is a vital dimen- sion in the marketing effort but it has not yet received adequate attention from marketing scholars. The dramatic growth in the awareness of the impor- tance of physical distribution to the marketing effort is acting to rectify this situation. But the spatial dimension is equally as important in Obtaining demand as it is in servicing demand. Further work is needed. Similarly, focusing attention on the relatively unattended but vital area of industrial marketing is a contribution of significant value. While it involves a major portion of the total marketing effort in our economy, industrial marketing has received short shrift in the literature and aca- demic curricula. III I I) ‘II 1 Illllt...‘ 192 Limitations of the Research and Spggestions for Future Research By leaving these two aspects of the research until last, it seems natural that they be considered concurrently. Five areas are presented under this topic and in most there is a direct relation between the limitations of this study and the suggestions for future research. 1. The buyer behavior research is limited to a parti- cular class of buyer (metal working) as well as being limited to a size range for the buyer's firms. Furthermore, the data is minimal and lacking in depth. This only scratches the surface of what should prove to be a very fruitful research area. Before detailed analysis of observed behavior patterns can be made, questions of why, how much, and from whom must be answered. For example, it is speculated that the influence of Detroit in the grinding wheel patterns represents special require- ments met by specialists who are ppp_general-line distributors. The study of multiple source pur- chasing behavior seems to offer promise generally as well as specifically related to spatial patterns; this it seems is true both in industrial buyer be— havior and consumer buyer behavior. 2. 193 The attempt to study the spatial relationships of market structure was far from satisfactory. And yet, as we face the necessity of planning for all- new cities it is essential that we develop an understanding of the order that exists. Other frameworks for analysis must be developed and tested; and, better measures of central-place hypotheses also may prove fruitful. This study is limited to a consideration of indus- trial distributors. A great amount of industrial marketing does not involve distributors. Sales are made direct by producer to user. Spatial relationships are probably entirely different. This is a particularly important area where re- search is needed. Central-place theory may be of little value in this area; but, a construct like Huff's probability model might be extremely appropriate. Industrial distributors, and particularly the general-line industrial distributors were charac- terized earlier as a marketing anachronism Offering a promising area for academic inquiry. This study was focused only on spatial relationships. It is felt that the condition detected here warrants future research on other marketing dimensions such as product, price, promotion and channel structures as well as environmental factors. These 194 establishments are usually classified as merchant wholesalers. In fact, they are not; they are more like industrial retailers. A review of the classi- fication and functions of channel members in indus- trial marketing is called for. 5. Finally, this research is limited to only three types of industrial distributors in one restricted area of the United States. The same is true of the buyer study. Any generalization drawn is made in cognizance of this fact and should be interpreted as such. Summary This study has attempted to search for order in relation- ships as marketing is carried on over space. The focus was specifically on the marketing of industrial products through distributors. Central—place theory, from the field of geo- graphy, was used as the basis for analysis. While the study did yield noteworthy findings on the side of industrial buyer behavior, the study of the structure of the market place on the seller's side was less satisfactory. Central- place theory as interpreted in the research design did not provide a description of the spatial order of that structure. This result, however, does not invalidate the theory as a basis for analysis in that it does provide some insight into the causes for the lack of spatial order. LIST OF REFERENCES LIST OF REFERENCES Berry, B. J. L. Geoqraphy of Market Centers and Retail Distribution. EnglewoOd Cliffs, N.J.: Prentice—HaII, 1967. Bowersox, Donald J. "Evaluation of Alternative Solutions to the Food Distribution Center Location Problem." Unpublished Ph.D. dissertation, Michigan State University, 1960. ; Smykay, Edward W.; La Londe, Bernard J. Physical Distribution Management. New York: The Macmillan Company , 196 8 . Chisholm, Michael. "Johann Heinrich von Thunen." Readings in Economic Geography: The Location of Economic Activit . Editeduby Robert H. T. Smith, Edward J. Taaffe, and Leslie J. King. Chicago: Rand McNally and Co., 1968. Devletoglou, Nicos E. "A Dissenting View of Dquoly and Spatial Competition." Economica. (May, 1965). 140-160. Greenhut, Melvin L. Plant Location in Theory and Practice. Chapel Hill: The University of North Carolina Press, 1956. . Microeconomics and the Space Economy. Chicago: Scott, Foresman, 1963. Harris, Chauncey D. "The Market as a Factor in the Localization of Industry in the United States." Annals, Association of American Geographers, Vol. 44 (No. 4, 1954), 3154348. Hoover, Edgar M. Location of Economic Activity. New York: Hotelling, Harold. "Stability in Competition." The Economic Journal, XXXIX (1929), 41-57. Huff, David L.; Blue, Larry. A Programmep Solution for Estimating Retail Sales Potential. Lawrence, Kansas: University of Kansas, 1966. Isard, Walter. Location andngace Economy. New York: John Wiley and Sons, 1956. . Methods of Regional Analysis: An Introduction to RegiOnal Science. New York: John Wiley and Sons, 1960. 195 196 Losch, August. The Economics of Location. Translated from second revised editiOn by William H. Woglom.with the assistance of Wolfgang F. Stolper. New Haven: Yale Univer- sity Press, 1954. Reilly, William J. The Law of Retail Gravitation. New York: Reilly, 1931. Revzan, David A. A Geography of Marketing: Resource Bibliography. Berkeley: Institute of Business and Economic Research, University of California, 1968. Siegel, Sidney. Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill, 1956. Smithies, Arthur. "Optimum Location in Spatial Competition." Journal of Political Economics, XLIX (1941), Valvanis, Stephan. "Losch on Location." Readings in Economic Geography: The Location of Economic Activipy. Editediby RObert' . T. Smith, Edward J. Taaffe, and Leslie J. King. Chicago: Rand McNally and Co., 1968. Wolpert, Julian. "The Decision Process in a Spatial Context." Annals, Association of American Geographers, Vol. 54 (No. 4, 1964), 537-558. APPENDIX 197 EXHIBIT I GENERAL-LINE INDUSTRIAL DISTRIBUTOR QUESTIONNAIRE Name: Company: Address: Place: Section I For which brand, or brands, of gpinding wheels are you an authorized distributor? Which of the following sizes do you stock? Dia Thk Hole Type Grit Purpose 6 1 1-1/4 Straight Medium General 7 1/2 1-1/4 Straight For alloys Tool room 6 3/4 1-1/4 Straight Extra-coarse General 6 3/4 1-1/4 Straight Fine General 8 1 1-1/4 Straight Medium General 4"/3" 1-1/2 1-1/4 Flaring cup Carbide Roughing 10 1 1-1/4 Straight Extra-fine General 10 l 3 Straight For alloys Tool room 12 1-1/2 1-1/4 Straight Fine General 2"/l/4" 4 5/8"-ll thread, 1" deep Cone, Type 19 Snagging 198 EXHIBIT I (Cont.) Company: Section II How many square feet of warehouse space do you have? How many trucks do you use for delibery? How many outside salesmen do you have who are selling grinding wheels? Whom do you consider to be your principal compet- itors from out of town for your grinding wheel market here in (location of establishment)? Competitor Location If one of your customers had an immediate requirement for a size or quality of grinding wheel which is not available from any local source, where would you suggest he try (or, where would you try for him) to find it? Section III On the map provided, please indicate the extent of the terri- tory covered by your salesmen. If you have no outside salesmen please indicate What you consider to be the extent of your market for grinding wheels. 199 EXHIBIT II GENERAL-LINE INDUSTRIAL DISTRIBUTORS (GRINDING WHEELS) ESTABLISHMENT DATA Place Distributor Ann Arbor Royall Inc. Battle Creek Kendall Industrial Supplies Bay City Jennison Hardware Co. Detroit Chas. A. Strelinger Co. Flint Gransden—Hall & Co. Grand Rapids Grand Rapids Supply Co. Jackson C. E. Hamlin Co. Kalamazoo Kalamazoo Mill Supply Co. Lansing Geo. Worthington Co. Muskegon Muskegon Hardware Co. Port Huron Port Huron Equipment CO. Saginaw Morley Brothers Prod. Spec. Scale Sq. ft. the. 18,000 48,600 90,000 100,000 40,000 18,000 30,000 18,000 100,000 15,000 3,000 114,000 Trucks 13 Salesmen 19 200 EXHIBIT III GENERAL—LINE INDUSTRIAL DISTRIBUTORS (GRINDING WHEELS) Distributor's Place Ann Arbor Battle Creek Bay City Detroit Flint Grand Rapids Jackson Kalamazoo Lansing Muskegon Port Huron Saginaw MARKET TERRITORY DATA Square Miles of Market Territory 4.300 2.700 21.000 3.500 1.500 13.800 4.600 5.100 5.300 2.000 1.800 6.400 Rank 10 11 201 m .Hmm. I n ma room momam HoanumsocH poo xcmn coauoasmom consume “may coaumeuuoo mo ucmwoammwoo room cmeuommm ”ouoz .SMOHSOHS .mcfimcmq ummm .muwmum>fico mumum savanna: .cowumuumacfieom mmmcflmsm mo Hoosom monsoouo .nonmomom mo coflmw>ao .bosmomom oweocoom pom mmmcamsm mo smousm .moma .om sun .uomuumom Hoowumauoum cmmwnOHZ .cowumHsmom mo noncou .m.D coma so oomph chwuownoum coma .H poemsflumo I m “as mae.mm Ame mm loo asa.m~ o o amo.me~ smoaomm ANHV ooo.m Leas oo lune Hea.m we as mmm.moa oonom upon Ame mom.aa lac em Ame moa.m~ m m aom.mma ooooxnoz Ame mmo.em lee mos Lee Hmm.m~ m a mmo.oom anemone loo ooo.oa inc mma loo emo.m~ m a omo.ama commandos Ase mma.vs Ame 40H lone Hom.oa s 0H een.ems consume and ooe.em lac ode inc Noo.eo m N oao.m~m mesons compo Ame moo.- nos om “we mam.mo v m omm.aom ocean lac amm.aov lac om~.~ Ase mma.mmv a a ama.maa.v uaonuoo lees omo.m m imav mm fleas omo.m as «a Hmm.moa spec son hone moo.a m im.mc on “as aoe.ma on m o~o.mma xomno manomm Ame oem.m im.oc om loo smm.mm m m «mm.Ha~ neous one inomnamannmooonc i.eoso once no omv moamm mammoaonz unwazmfiaomumm moanedmsm xcmm oooam ucmnonoz mcfiusuommssmz mcwusuoomssmz HoHMHOOOSH Room coaumasmom woman Mzfim mofldm AdeBmDQZH >H BHmmem Formula: 202 EXHIBIT V CORRELATION COEFFICIENT but when ties are involved: r where 2x2 T - N - ZTx and Zyz 12 2x + 2y 1'5 xfizyT N3'- N 2 2 3 = N - N - ZTx 12 given rank. - Ed EXAMPLE OF CALCULATION OF SPEARMAN (r ) RANK 3 - t and t 5 the number of observations tied at a This example tests Hypothesis H1 for general-line industrial dis- tributors. Place Ann Arbor Battle Creek Bay City Detroit Flint Grand Rapids Jackson Kalamazoo Lansing Muskegon Port Huron Saginaw 3 2x2 - (12) 12 2 (12)3 ZY = 12- r _ 139.5 + 143 - 204.9 s ‘ 2] (139.5)(143) Hierarchy of Function (x) 000 WWW H WHO‘IbU'Il-‘KDO‘I—‘l-‘koo 0 Ln H O U1 U1 [(2 2: %) +(2)3 2 = 143 Hierarchy of Place (y) H Ia Oshac>¢>~lc>uoIOI4IOIOIn H d d2 4 16.00 0 0.00 -10.5 110.25 .5 .25 3.5 11.25 7 49.00 1.5 2.25 -2 4.00 0 0.00 1.5 2.25 .5 .25 3 .2199 282 = 204.90 2+K3) ' 3] = 139.5 2 Name: Section I 203 EXHIBIT VI STEEL BAR DISTRIBUTOR QUESTIONNAIRE Company: Address: Place: Which of the following sizes do you stock? 2 x 2 x 1/4" Angle _____ 1/4 x 2" Flat _____ 1-1/2 x 1-1/2 x 1/8" Angle _____ 3/4" Round _____ 3/4 x 3/4 x 1/8" Angle _____ 1/4 x 2-1/2" Flat _____ 1-1/2" Round _____ 2 x l x 1/8" Channel .____ 5/8 x 4-1/2" Flat _____ l x 3/8 x 1/8" Channel Which of the following products do you stock? Do you have a shear? Plates _____ If yes, max thickness? Wide Flange Beams _____ If yes, max size? Sheets: Hot rolled Cold rolled Galvanized Stainless Aluminum If yes, what size? 204 EXHIBIT VI (Cont.) Company: Section II How many square feet of warehouse space do you have? How many trucks do you use for delivery? How many outside salesmen do you have who are selling bars? Whom do you consider to be your principal competitors from out of town for your bar market here in (location of establishment)? Competitor Location If one of your customers had an immediate requirement for a Size or quality of bar which is not available from any local sources, where would you suggest he try (or, where would you try for him) to find it? Section III On the map provided, please indicate the extent of the territory covered by your salesmen. If you have no outside salesmen please indicate what you consider to be the extent of your market for bars. 205 .suoo Haom cosmoaom b 0 m.0H I0I m.0H H h v m.v m m h m.v o b m H AH N h m.0H N m.0H N cosmoaom xsnm moamuso 0 mm mH pomp mxosna mo nonesz deflo BZMZSmHAmdawm mmOQDmHMBmHD Mdm qmmsm .buop no>waoo mxosne 000.0v Coo-Hm 000.0H 000.0m 000.0N 000.0H 000.00H 0004m¢ 000.com 000.mh 000.vN 000.0H mmsozoumz mo poem whosom HH> BHmmem .mnononnnumno Honnnmoooa mannunmnoooo owns. v v oaoom avaswoomm acetone i mumbuonm moanoz zmcflmmm com o comsmnom .H comer unom Hmouw coomom comoxmsz Hooum Homno>as0 mcflmcmq .saomom Hans OONoeoaos oonEmHOM «Homum Guam :0mxomb Saddam o Hoonm meadow ozone panama ozone Hooum Haom ucwam comuomm .e smOmOO uaonuoo «onmzouom confiscon sync son Hooum ouad xoouu oauuom Hooum ocauo>aoz Moons one Sham woman STEEL BAR Ann Arbor Battle Creek Bay City Detroit Flint Grand Rapids Jackson Kalamazoo Lansing Muskegon Port Huron Saginaw 206 EXHIBIT VIII DISTRIBUTORS MARKET TERRITORY DATA Square Miles of Market Territory 4.300 2.000 21.000 40.000 11.500 16.000 11.300 5.100 25.000 2.200 1.700 6.400 Rank 11 10 12 207 EXHIBIT IX INDUSTRIAL BUYER QUESTIONNAIRE Name: Company: Address: Place: Section I 1. Product: Sources-- 1. 2. 3. 2. Product: Sources-- 1. 2. 3. 3. Product: Sources-- 1. 2. 3. Section II Grinding Wheels Ball Bearings 1. 1. 2. 2. 3. 3. Hot Rolled Steel Bars 1. 2. 3. 208 EXHIBIT X PRODUCT GROUPS SELECTED BY INDUSTRIAL BUYERS Aluminum Extrusions Brass and Copper (2) Bushings (2) Casters Castings Coolants Corrugated Paper--Packaging Materials (3) Die Mold Supplies (3) Die Wear—plates Drawing compounds Drills (13) Electric Plugs (1) Electric Switches (3) Expansion Joints Fasteners‘(7) Fittings (2) Fractional Horsepower Motors (3) Gloves Grating Hardware (3) Hand tools (9) Industrial Gas and Air Janitorial Supplies (2) Lubricants (6) Metal working machinery (2) Mill Supplies (4) Nuts and Bolts (4) Office Supplies (4) O-rings _ Portable grinders Power Transmission Equipment (belts. sprockets. drives) Punching Equipment Resistors Safety Equipment Solvents Tool Bits Tool Steel Welding Rods (2) (6) 209 EXHIBIT XI TEST OF HYPOTHESIS H3 Hierarchy Category 1 Category 2 of Place 5 Ann Arbor 2 l 9 Battle Creek 3 12 Bay City 3 1 Detroit 0 3 Flint 3 2 Grand Rapids 3 10 Jackson 1 l 7 Kalamazoo 2 4 Lansing 2 l 8 Muskegon 1 11 Port Huron 3 6 Saginaw 2 25 3 n = 28 degrees of freedom = 1 Critical value for x2 = 6.64 for x = .01 level. 2 2 x2 . 19;_-_§l) + (09 - En) El E2 2 a (25 - 14)2 + (3 - l4)2 14 14 = 17.285 X 210 EXHIBIT XII LOCATION OF INDUSTRIAL BUYERS' PREFERRED SOURCES Place Ann Arbor Battle Creek Bay City Detroit Flint Grand Rapids Jackson Kalamazoo Lansing Muskegon Port Huron Saginaw FOR GRINDING WHEELS Buyer WNI-‘WNI-‘UJNHWNHWNHWNHWNHWNHWNHWNHMNHWNH x - denotes own place - - denotes buyer does not buy lst x Detroit Detroit Kalamazoo X Kalamazoo X Ixxxxxxlxxx xxx x Detroit x xxxxxx x Detroit x x x Detroit 2nd Detroit Detroit x Kalamazoo lxxx xxxxxx x Detroit x x Detroit Detroit x x x 3rd Detroit Detroit x Detroit Saginaw Detroit Detroit Detroit xx 211 EXHIBIT XIII PURCHASING AGENTS' PREFERRED SOURCES FOR: GRINDING WHEELS media“ Po 3 I Buttes. $833. I G DOVE»? «£5 magnum Inna. awn Numbers inside circles indicate number of local sources. 212 EXHIBIT XIV LOCATION OF INDUSTRIAL BUYERS' PREFERRED FOR BALL BEARINGS Place Ann Arbor Battle Creek .Bay City Detroit Flint Grand Rapids Jackson Kalamazoo Lansing Muskegon Port Huron Saginaw Buyer WNHNNHDJNHWNHWNHWNHWNHWNHWNHWNHWNHWNH lst xxxxlxxx x xxxxxxxxxxxxxxxxxx x Detroit Detroit Detroit x x x 2nd Detroit Detroit Kalamazoo Saginaw Detroit Chicago x x Detroit Grand Rapids SOURCES 3rd Detroit Detroit Detroit Bend 213 EXHIBIT XV PURCHASING AGENTS' PREFERRED SOURCES FOR: BALL BEARINGS \NNWIM\ Q ay (Cwy‘g’ \ “I ‘\. I fluflo Mflhuou NV N 900] ‘ ® Pt. ”r ”b ‘1‘ K Gflho anhUS I / IR / «(79 enuuuu Ann ARBOR L mo Numbers inside circles indicate number Of local sources. 500:» amp 214 EXHIBIT XVI LOCATION OF INDUSTRIAL BUYERS' PREFERRED SOURCES FOR STEEL BARS Place Ann Arbor Creek Battle Bay City Detroit Flint Grand Rapids Jackson Kalamazoo Lansing Muskegon Port Huron Saginaw Buyer NNHWNHWNHWNHNNHWNHWNHWNl-‘WNHWNHwNHWNH lst Detroit x xxxxx x Detroit x xxxxx x Chicago x x x Chicago Grand Rapids Chicago x x x x Chicago x Detroit Detroit x x Detroit Flint 2nd Detroit Detroit Detroit x Detroit Chicago Detroit Detroit Detroit Chicago Chicago Detroit Chicago x Chicago Detroit Grand Rapids Detroit x x Grand Rapids Flint Detroit Flint Flint Flint 3rd Detroit Detroit x Lansing Chicago Detroit xxx Detroit x x Grand Rapids Detroit Detroit Chicago x Detroit Flint Grand Rapids Grand Rapids Detroit 215 EXHIBIT XVII PURCHASING AGENTS' PREFERRED SOURCES FOR: STEEL BARS PORT HURON tutu-s “:2? c’ I J «In... E11 Numbers inside circles indicate number of local sources. ”TITII‘MHIIIMI((111m)Infill)“ 03145 7900