>'K. Division of Agricultural Sciences * UNIVERSITY Of CALIFORNIA , ^f\ __* .:■:...: r ;. Structural Changes in the California Fluid Milk Industry their effects on competition and market performance Daniel I. Padberg and D. A. Clarke, Jr. mm CALIFORNIA AGRICULTURAL EXPERIMENT STATION BULLETIN 802 P» This bulletin investigates the postwar changes in the structure of Cali- fornia's fluid milk industry, particularly changes in the number and size distribution of buyers and sellers. It explores the effects of these changes on the economic performance of the industry, especially on competition within the industry and market results. n Table of contents Background and nature of the study 4 Organization of firms 4 y Development of the industry in California 4 Purpose of the study 6 Explanation of terms used in this analysis 6 Factors affecting structural development 9 Distributive channels 9 * Production and consumption trends 10 y Changes in techniques of processing and distribution 12 Changes in economies of scale 12 Changes in transportation cost 14 * Structure of market participants 14 The relevant market 14 ^ Product dimension 14-4 Geographic dimension 16 Recent changes in the structure of buyers 16" Retail buyers ]_7 Wholesale buyers Yl Recent changes in the structure of sellers 22 Turnover among sellers u 22 "** Changes in concentration of wholesale firms 24 ^ Changes in concentration of retail firms .... 30 Processor-retailer integration 34 Product differentiation 36 "" * [2] Conduct of market participants 39 Wholesale market participants 40 Retail market participants 40 Competition and performance 41 Changes in competition 41 The wholesale market 41 The retail market 42 The relation between competition and performance 43 Definition of performance goals 43 Performance of the industry 45 Appendix A : Estimation of concentration curves of California supermarkets, 1950—1958 52 Appendix B: Procedures for developing the population of entrant and exant fluid milk processing firms, 1950—1958 .... 53 Sources of published data 53 Sources of unpublished data 54 Compilation of the target population 55 Drawing the sample 66 ► The authors: Daniel I. Padberg, at the time of the study, was Junior Specialist in the Agricultural Experiment Station, Berkeley, and is now Assistant Professor of Agricultural Economics, Ohio State University. D. A. Clarke, Jr. is Professor of Agricultural Economics, Agricultural Economist in the Experi- ment Station and on the Giannini Foundation, Berkeley. JUNE, 1964 [3] Background and Nature of the Study 1 Organization of firms Three types of firms, or individual decision-making units, operate within the fluid milk industry — dairy farms, proc- essing plants, and retail food stores. All three have undergone major structural changes over the years. Dairy farms, relatively atomistic in structure, are ex- cluded from consideration in this anal- ysis which focuses attention on the proc- essing firms and the retail food stores. According to the California Crop and Livestock Reporting Service, 540 fluid milk processing firms were active in Cali- fornia in 1960. A large proportion of processor sales, however, are concen- trated within a few large firms. Some of these have developed multiplant process- ing and distribution systems, particularly in wholesale operations. Approximately 40 per cent of the fluid milk distributed in California reaches the consumer through about 10,000 retail food stores. Sales are concentrated in the largest few firms, however, because of the development of supermarket chains. Because of the importance of milk and dairy products in the human diet and because milk is readily susceptible to dis- ease-producing bacteria, the dairy indus- try is subject to more governmental regu- lation than most other food industries. Public controls include laws prohibiting fraud and adulteration and regulations designed to improve the sanitary quali- ties of milk and its products. In addition, the dairy industry is the only agricultural industry whose prices (for milk for fluid consumption) are established by the government. In California, the main governmental agencies regulating the fluid milk indus- try are the Bureau of Dairy Service and 1 Submitted for publication Oct. 10, 1963. the Bureau of Milk Stabilization, both part of the California Department of • Agriculture. The Bureau of Dairy Service, together with municipal health authorities, establishes and maintains sanitary specifications and inspections for milk-producing and milk-processing units. The Bureau of Milk Stabilization establishes minimum prices for milk for fluid purposes at all levels of distribution, ^ including producer, wholesale, and retail prices. The Bureau also enforces those sections of the Agricultural Code which prohibit the use of "unfair practices" by the firms operating in the fluid milk in- * dustry. 2 Unfair practices include various types of price and "nonprice" competitive * devices, such as the use of secret rebates, unearned discounts, and the provision of services, equipment, or financial assist* ance at prices which are below costs. 3 Development of the industry in California Historically, the organizational frame- work of the fluid milk industry was very simple and all firms operated in rather small units. Because of the bulk and perishability of the product and the lack u 2 The Milk Stabilization Laws, as they are * commonly known in California, appear in chapters 16 and 17 of the Agricultural Code of the State of California. Chapter 16 contains provisions designed to control and prohibit unfair trade practices; chapter 17 contains authorization to establish minimum producer prices, processor margins, and store margins. 3 For a discussion of the nature and opera- < tion of the California milk stabilization pro- gram, see J. M. Tinley, Public Regulation of Milk Marketing in Calif orina (Berkeley: Uni- versity of California Press, 1938), 213p. Also, D. A. Clarke, Jr., Fluid Milk Price Control in California: A Report to the Joint Legislative Committee on Agriculture and Livestock Prob- lems (Sacramento: State Printing Office, 1955), * 191p. [4] of modern transport facilities, production was located near consumption centers. The producer sold his output directly to final consumers, either delivered or on his premises. Milk was seldom pas- teurized or packaged in the early days, • and little, if any, processing was re- quired. Around the turn of the century, however, new processing functions were created by the establishment of sanitary standards and the development of new techniques. Standard pasteurizing pro- cedures, mechanical refrigeration, glass bottles, and sanitary standards for pro- ► duction were among the advances in- troduced. At about the same time, expan- sion of urban areas in the state made it feasible to separate the production and distribution segments of the industry. *- Specialization in both production and processing was greatly encouraged by these changes and by the potential scale economies in the performance of new processing functions which had entailed substantial investment in equipment. By 1920, fluid milk processing as a separate industry was generally established in the v urban areas of California. Although the growing cities pushed milk producers farther away from consumption centers, the level of development in handling and transporting techniques determined the organization of production into fairly tight "milksheds" around each city. Thus, in the early 1920's the fluid milk proc- t essing industry in California was typi- cally composed of single-plant firms which collected milk from nearby farms and sold it bottled and processed in a single town or urban area. ► In the decade between 1925 and 1935, a fundamental change took place in the type, number, and size of firms. By 1935, seven multiplant firms had been estab- lished, the largest two having more than ten plants each. This resulted from large manufactured-dairy-product firms enter- ing the fluid milk processing enterprise. By this time, fluid milk processing func- tions had been developed to such a degree that considerable investment in special- ized equipment was required. It is likely that the superior financial strength of the manufacturing firms, coupled with the opportunity to gain better utilization of their existing plant facilities, encouraged these multiplant companies to enter the fluid milk operations. The fluid milk industry grew rapidly during World War II. Between 1940 and 1942, four counties — Alameda, Los An- geles, San Diego, and San Francisco — from which a great part of the Pacific operations was supplied, increased their annual output of processed milk by more than 34 million gallons, or 28 per cent. 4 Although the dairy procession industry expanded throughout the state, these four counties gained in relative importance, moving from 68.8 per cent to 73.2 per cent of the state total. This expanded out- put resulted primarily from increased volume per plant, since plant numbers declined during the war. Following the war, other significant changes affected the processing industry. Annual government purchases of fluid milk in California dropped from the 1945 level of 9.8 million gallons to 2.5 million gallons in 1947. 5 Paper (fiber) con- tainers, which had been introduced in Los Angeles and San Francisco as early as 1938, had come into general use dur- ing the war. Large plants in the four counties mentioned had not previously shipped their output over very great dis- tances. Now such shipments were facili- tated by the wide use of these lightweight paper containers and by the extensive postwar road construction program, which increased the miles of freeway in California by 145 per cent, from 588.8 at 4 California Department of Agriculture, Sta- tistical Report of California Milk Products and Lists of California Milk Products Plants, 1942 (Sacramento, 1942), p. 15. 5 California Crop and Livestock Reporting Service, California Dairy Industry Statistics for 1954: Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacra- mento, 1954), p. 35. [5] the close of the war to 1,440.3 by De- cember, 1949. 6 Mergers and sale or dissolution of some business entities in the postwar period resulted in a new consolidation of the fluid milk processing industry, and some existing firms that were operating high-capacity urban plants were moti- vated to penetrate new areas to increase their plant volume and operating effi- ciency. The incentive to increase plant size was especially stimulated by the adoption of paper containers. The greater capacity of equipment adapted to paper containers gave large wholesale proc- essors a competitive advantage over small processors because this technique is asso- ciated with high cost for small-scale operations. Consequently, many local processors were acquired by larger firms and others were completely crowded out of the market. Thus the number of proc- essors declined while the market share of larger firms increased. Purpose of the study The postwar consolidation movement could have two possible serious reper- cussions: First, the welfare of many indi- viduals could be affected, since increased consolidation necessarily caused many firms to disappear. Second, greater con- centration could bring about possible monopolization and associated economic exploitation, for as numbers of firms de- clined, some existing processors greatly increased their market shares. Is it not likely that such increased economic power of the large companies might affect the welfare of all interested parties — in- cluding final consumers, producers, and competing distributors? In the light of these possibilities, it was determined that this study would ap- praise and evaluate the following ques- tions: (1) How and to what extent has the structure and institutional framework of the fluid milk processing industry changed during the past decade? Answers to this question required an in- vestigation of changes in the numbers and size distributions of both buyers and sellers of processed fluid milk in Cali- fornia and in the relationships between these buyers and sellers. • (2) What has been the effect of the observed structural changes upon com- petition within these industry segments during the postwar period? Answers to this question required an investigation of changes in the extent and degree of mar- ket power possessed by both buyers and sellers within the industry. \ (3) Finally, what has been the conse- quent effect of such changes upon market performance, that is, market results? While answers to this question cannot be precise, investigations were made into ^ various appropriate measures of market performance, including analyses of inter- market and interproduct price and product-service combinations. Explanation of terms as used in this analysis 7 To avoid ambiguity of meaning, this section explains several important con- cepts used in this report which are some- what imprecise and commonly used with varying interpretation : market structure, market conduct, market performance, competition, and the relevant market. , They are given more detailed treatment in later sections when specific application is made to the fluid industry. Those al- ready familiar with these concepts may prefer to proceed directly to page 9. Market structure Structural characteristics include the * number and size distribution of buyers 6 California Highivays and Public Works, vol. XXIX, nos. 3/4, March-April, 1950, p. 57. 7 Based, with some modification, on the classi- fication system in J. S. Bain, Industrial Or- ganization (New York: John Wiley & Sons, Inc., 1959), chap. II, pp. 19-43. [6] and sellers, entry conditions, and the ex- tent of product differentiation. The term "industry structure" is used when these measures are applied to a group of buyers or to a group of sellers; the term "market structure" when the attributes of both buyers and sellers are considered to- gether as a group. Market conduct "Market conduct" refers to the market practices and policies of both buyers and sellers. The specific price policies of indi- vidual firms provide examples of market conduct. Although the concept includes the practices of both sellers and buyers, it is often convenient to discuss the conduct of sellers or buyers separately. Market performance "Market performance" refers to the economic end results obtained from an industry, largely derived from its struc- ture and conduct. These results are re- flected in the amount and variety of goods and services produced per unit of re- source inputs. Important dimensions of market performance are technical (cost) efficiency, profit rates, and the amount of selling and promotion cost; also impor- tant, but varying in different industries, are progressiveness in adopting new and more efficient technology, product de- velopment, and the degree of variety and choice offered to consumers. Measurement of market performance would ideally constitute a comparison of observed results in a particular industry with some optimum. Each performance dimension would be weighted according to its importance. Such a ratio of ob- served results to optimum results would be useful for making comparisons with performance levels in similar industries in alternative areas and perhaps relating performance levels to conditions of market structure across many industries. Unfortunately, no optimum or "ideal" standard of market performace is avail- able; most dimensions of market per- formance, particularly technical effici- ency and product quality, are affected by the state of the arts and therefore change over time. Furthermore, no "standard" of progressiveness is available. While it is presently impossible to com- pare observed results with such ideal standards, it may be possible to note the way these results are affected by changes that have occurred over time. If such cause and effect relationships can be ob- served, market performance may be "improvable" by making appropriate changes within the industry. It is from this perspective that the term market per- formance is used in this analysis. 8 Competition The meaning of competition is typi- cally surrounded by considerable ambig- uity. This concept has been used in less than parallel manner by economists, law makers, and business people. Further, the term competition takes on different connotations when used in connection with the structure of markets, the con- duct of market participants, or the nature of economic results obtained in markets. Since competition can refer to different aspects of a market and has special meaning among the several groups interested in market activity, in its broad sense it is not sufficiently precise to be useful in empirical research. There- fore, for purposes of this study, it is es- sential to bring this broadly used concept into more specific focus. Competition may be regarded as a pattern of incentives which determines the way various individual decision- making units operating in this economic- system react to each other. The nature of these interrelationships and interdepen- dencies shapes the pattern of performance 8 It is in this sense that the criteria indicated here are similar to the concept of "workable" competition as developed by J. M. Clark. J. W. Markham, S. Sosnick, and others. [7] in specific industries. Our tradition of economic thought has seen competition in its "pure" form as a set of incentives under which firms struggle for survival by reducing the costs of operation. Indi- vidually, they can have no effect on the market (because of atomistic structure, homogeneous products, etc.). In such an idealistic economic system, market parti- cipants can in this sense affect their rivals only by lowering their costs. Economic results — and so, market per- formance — are optimal in this textbook type of competition in that firms are motivated to use the most efficient pro- duction operations, profits equal normal return on investment, selling costs are not incurred, and product development is consistent with consumer wants. 9 While this rather formal development probably dominates the concept of competition, our experience with antitrust policy con- cerning "restraint of trade" of various types has broadened the concept. A "les- sening of competition" has typically been defined as the imposition of some res- traint of trade. It is from this general background that purity of competition has come to imply a system of incentives in which rivalry of firms includes only the firm policy dimensions affecting costs. The core of this ideal concept is that no market participant has "market power," the ability to affect the demand facing rivals. Market power implies imperfec- tion — or a lesser degree of competition. It therefore follows that there are grada- tions in competition ranging from pure form to monopoly. In actual markets the pattern of incentives is not restricted to internal (firm) policy and therefore market power is typically present in some degree." Market power of sellers may be de- fined as the ability of an individual firm to affect the demand facing rival sellers. The concept may be extended to include interactions between buyers and sellers. In this sense, market power of sellers (buyers) becomes the ability to deny or restrict alternative sources of supply (market outlet) to buyers (sell- ers). The expansion of the use of this definition to express the interrelations be- tween alternative market participants is not inconsistent because the ability of sellers, for example, to deny or restrict alternative sources of supply to buyers is the same thing as reducing the demand facing rival sellers. The ability to obtain preferential reception from buyers by differentiating firm output is an example of market power which has the effect of both restricting alternatives to buyers and reducing the demand facing rival sellers. In this study efforts will be made to determine the effect that changes in the firms comprising the fluid milk in- dustry have had upon the pattern of in- centives to which market participants respond — or the nature of competition. The relevant market To be meaningful, the term relevant market must be defined in two dimen- sions: product dimension and geographic dimension. The basic determinant for the product dimension is substitution. Re- lated products are considered to be within the relevant market if they are substitut- able. In any practical case, few classes of commodities are perfectly substitutable. Generally, when any product may be used interchangeably with the product under consideration, the two products are con- sidered to be the same and so included within the relevant market. This inter- changeability is measured conceptually by substitution or cross elasticities. As re- liable substitution elasticities are difficult Development of this proposition is presented in Tibor Scitovsky, Welfare and Competition : the Economics of a Fully Employed Economy (Chicago: Richard D. Irwin, Inc., 1951) 457p. 1 Professor J. K. Galbraith in American Capitalism, the Concept of Countervailing Power (Boston: Houghton Mifflin Company, 1952) 217p., has given considerable illumination to the economic power implications of our concept of competition. [8] to obtain empirically, other indicators may be used to establish substitutability. u The "dividing line" that separates products is in a sense arbitrary and may differ for different classes of products. The geographic dimension of the rele- vant market must be defined in the con- text of the geographic area in which the major forces that affect the supply of or demand for a product are reasonably homogeneous. This implies the absence of institutional, geographic, or economic barriers to free flow and interchange within, though not between, the areas of competition so designated. An example from the dairy industry may serve to clarify this concept more fully. It can be argued that the relevant market for butter encompasses the nation in its en- tirety, since this is a homogeneous, un- differentiated product that can be and is shipped at low transfer cost per unit of value between widely separated geo- graphic areas. Further, wholesale sellers of butter — manufacturers and brokers — 11 A discussion of empirical evaluation of the product dimension of the relevant market may be found in E. T. Grether, "Economic Analysis in Antitrust Enforcement," The Antitrust Bul- letin, vol. I, no. 4, January-February, 1959, p. 59. are well informed with respect to geo- graphic price differences and act ration- ally on that knowledge in making deci- sions concerning the ultimate destination of shipments. For this reason, a firm that handles a relatively large proportion of the total butter sales within a given metropolitan area is powerless to influ- ence price, to practice discrimination, or to act in any fashion other than that of a "price taker." Conversely, either of two variants from this example would result in more limited areas of effective competition. One is the existence of institutional barriers to free access by certain sellers such as would result from health department licensing provisions which define a limited "eligi- ble supply." The other is the existence of high transport costs that essentially isolate and insulate a given geographic area from the effects of either supply or demand conditions in remote regions. The concept of the relevant market in the space dimension, then, includes any geo- graphic subdivision of the total supply of a product, the control of which — due to isolation or insulation from alternative supply — gives the seller a measure of market power. Factors Affecting Structural Development This section investigates the external environmental changes which would be expected to influence industry structure, but no attempt is made to explain these changes. For example, data concerning changes in processing and delivery tech- niques are presented not to show why they were adopted, but rather to investi- gate their impact. Distributive channels A distinction is made between whole- sale and retail channels of fluid milk dis- tribution based on the general differences in characteristics of buyers (customers of the processing firms) and in processing facilities required, as determined by type of container used. Generally, wholesale refers to sales to groceries, restaurants, and institutions; retail, to direct sales to households for final consumption. According to this classification system, the wholesale channel includes sales on wholesale routes and sales to the federal [9] Table 1 SALES OF FLUID MILK BY TYPE OF TRADE, CALIFORNIA, 1950-1962 Year Total sales Wholesale* Retailf gallons per cent 52.2 53.6 54.6 54.1 52.7 52.6 53.7 55.8 56.6 56.5 56.6 57.8 56.2 gallons 158,093,551 163,708,190 170,471,771 178,563,523 188,201,598 201,320,836 212,083,811 211,126,254 210,698,183 215,955,142 216,826,022 210,605,271 207,731,009 per cent 1950 330,739,647 352,819,374 375,563,056 389,027,284 398,068,816 424,620,600 458,345,993 477,661,208 485,274,351 496,516,709 499,799,389 499,284,205 474,100,813 172,646,096 189,111,184 205,091,285 210,463,761 209,867,218 223,299,764 246,262,182 266,534,954 274,576,168 280,561,567 282,973,367 288,678,934 266,369,804 47.8 1951 46.4 1952 45.4 1953 45.9 1954 1955 1956 47.3 47.4 46.3 1957 44.2 1958 1959 1960 43.4 43.5 43.4 1961 42.2 1962 43.8 * Includes wholesale sales ot fluid milk; sales of fluid milk to federal goverment; and, in 1959-1962, sales of concentrated fluid milk (whole milk equivalent). t Includes home-delivery sales, cash-and-carry sales, sales to peddlers, and sales unclassified by type of trade. Sources: 1950-1956: California Crop and Livestock Reporting Service, Dairy Information Bulletin (Sacramento, monthly issues). 1957-1962: Idem, California Dairy Industry Statistics for 1957: Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacramento, 1957, and subsequent annual issues). government. The retail channel includes sales on retail home-delivery routes, sales to peddlers (retail distributors who buy processed milk rather than operate a plant 12 ) , sales by cash-and-carry opera- tors, and sales of a small group of pro- ducer-distributors who are not required to classify sales by type of trade. (See table 14, page 32.) As shown in table 1, during the years 1950-1962 wholesale channels took slightly more than half of total California milk sales. There was a slight trend from retail to wholesale dur- ing the 13-year period. Production and consumption trends Table 2 shows the difference between the annual production of Grade A milk fat and annual consumption of Class I milk fat (primarily fluid milk and fluid cream.) 1 * The years for which data are available, 1944 to 1962, may be divided into three periods. During 1944, 1945, and 1946, the excess of production over consumption — 6.3, 3.4, and 4.7 per cent, respectively — was in each year lower than the excess normally required to allow for "necessary" surplus, standby. 14 During this three-year period the Cali- fornia Bureau of Dairy Service author- ized milk produced on ranches with facilities not meeting Grade A standards 12 The peddler category includes some fluid milk purchased from processors by a dealer and later resold at wholesale to food stores or restaurants. This very likely represents a small part of peddler sales and is not considered separately for lack of available data. 13 These data understate the actual surplus, particularly in more recent years, as explained in the following pages. 14 On the basis of variation of production and consumption within years and between years during the 1944-1961 period, it appears that 10 per cent annual excess is ample standby to provide an adequate milk supply during the months of lowest production. This figure, while arbitrary, closely corresponds to the "other than Class I" usage reported for 1947, the year which followed the revocation of the emergency permits. [10] Table 2 COMMERCIAL PRODUCTION AND USAGE OF GRADE A MILK FAT CALIFORNIA, 1944-1962 Year Commercial production Class 1 usage Other than Class 1 usage thousand pounds per cent of production 1944 110,729 117,825 128,580 136,289 135,274 139,485 142,842 149,547 153,077 164,871 171,621 179,351 191,244 210,691 215,679 229,511 230,133 236,510 240,104 103,731 113,825 122,569 122,565 123,173 124,853 128,640 134,880 141,869 145,893 148,494 157,435 168,108 174,148 175,739 179,253 180,422 179,544 181,913 6,998 4,000 6,011 13,724 12,101 14,632 14,202 14,667 11,208 18,978 23,127 21,916 23,136 36,543 39,940 50,258 49,711 56,966 58,191 6 3 1945 3 4 1946 4.7 1947 10 1 1948 8.9 1949 10.5 1950 9.9 1951 1952 9.8 7.3 1953 1954 11.5 13.5 1955 12.2 1956 12.1 1957... 17.3 1958 18.5 1959 * 21.9 1960 21.6 1961 24.1 1962 24.2 Sources: 1944-1960: California Crop and Livestock Reporting Service, California Dairy Industry Statistics for 1960: Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacramento, 1960), p. 37. 1961-1962: Idem, California Dairy Industry Statistics for 1962: Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacramento, 1962), p. 34. to be used as Class I milk if the bacteria count met Grade A standards. 15 By Au- gust, 1946, however, the Bureau con- cluded that "within the next few months there would be sufficient regularly qual- fied Grade A producers to meet all demands for market milk." 16 The war- emergency permits were revoked on De- cember 31, 1946. A general balance be- 15 "Emergency permits" for the use of such milk for fluid uses were authorized by Assembly Bill 934, which became effective in May, 1943. 16 Calif orina Bureau of Dairy Service, Circu- lar Letter No. 29 to Approved Milk Inspection Services, by O. A. Ghiggoile (Sacramento, August, 1946). 17 See California Crop and Livestock Re- porting Service, California Dairy Industry Statistics for 1962: Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacramento, 1963), p. 13. tween production and consumption of Grade A milk was maintained during the years 1947 through 1952. On the average, milk fat production during this period exceeded consumption by 9.4 per cent. Surplus Grade A production began to increase after 1952, and by 1957 was well in excess of standby requirements. In 1962, the amount of Grade A milk fat used for non-Class I uses had reached 24.3 per cent of total production. This particular estimate and probably the esti- mates in other recent years are believed to understate the actual surplus, since some milk that is otherwise eligible for Grade A use cannot find a market through a Grade A plant but is received instead at a manufacturing plant and is therefore reported as manufacturing rather than Grade A supply. 17 [in Changes in techniques of processing and distribution Until 1938, most fluid milk was proc- essed in the following manner : local proc- essing plants received milk in cans; pasteurized it in vat-type equipment (a batch process) ; bottled it in glass con- tainers; and distributed it to local ac- counts. The same equipment processed products for both wholesale and retail distribution. After the adoption of a series of new techniques, fluid milk processing became specialized. Milk for all uses was received in bulk tanks and plant operations were converted to a continuous-flow process. Milk for wholesale distribution was pack- aged in single-service paper containers and delivered to accounts which were sometimes located throughout several metropolitan areas. Retail distribution continued to be essentially a local opera- tion, but wholesale distribution tended toward centralized processing. The equipment used in this specialized wholesale processing operation generally required less labor but raised fixed plant costs. For this reason, the new operation was feasible only when plant volume was high. Paper containers were generally adopted in the early 1940's. Table 3 shows the trend of sales of packaged fluid milk since that time. 18 Changes in economies of scale Paper containers found quick accept- ance, particularly in groceries and restau- 18 It has been suggested that the adoption of paper containers has acted as a catalyst in the adoption of other techniques. See Hugh L. Cook, Paper Packaged Milk in Wisconsin: Its Part in Expanding Distribution Areas, Wiscon- sin Agricultural Experiment Station Research Bui. 179 (Madison, 1953), p. 33. The extent to which this causal relationship was important in California is unknown. The adoption of bulk farm collection methods, however, appears to have occurred at about the same time as the adoption of paper containers. See U. S. Eco- nomic Research Service, The Dairy Situation, June, 1957, p. 20. rants, because container weight was re- duced and the need for deposits and returns on glass bottles eliminated. The adoption of paper containers brought about further technological and, eventually, structural changes. No longer was milk for wholesale and retail cus- tomers processed identically. Wholesale operations, using paper containers, re- quired new and separate filling equip- ment and additional facilities for supply- ing containers to the filler. The initial cost of this new, complicated processing equipment was relatively high compared with the cost of glass bottling equipment, and the single-trip paper container itself was considerably more expensive than the returnable glass bottle. The cost dis- advantage of paper containers versus glass proved to be particularly severe for low-volume operations. A monthly vol- ume of about 250,000 gallons is normally sufficient to gain essentially all apparent available scale economies for a glass operation, while such scale economies for the paper operation extend beyond 600,- 000 gallons per month. Consequently, the adoption of the paper container created a considerable advantage for large-scale operations. A tendency developed toward centralized processing which resulted in a further tendency towards a smaller number of large-sized processing plants. Scale economies of some magnitude were also provided by the adoption of the high-temperature, short-time (HTST) pasteurizing techniques which made it possible to shift from a batch process to a continuous-flow operation. This not only eliminated some of the bottlenecks of the old procedure, but also favored the substitution of equipment for labor and thus provided substantial incentive for larger and larger volume to fluid milk processing plants. While small-volume operations can use HTST equipment, it is then usually operated well below capacity. These changes in technology were likely to influence the structure of the [12 rr> o r_l 2 < oo 2: O o 2: o — a- CO Q_ < ^ C3 O ^ OO CC ra £= +H- -r-t- -H- -t~f- -M- Ir^cnoocou-) 1 NdiNrv^-Nrvoirs^-rHM £ r^OOCDLOr^ OOOrHOIOOOlN^ftCOlDOO — 1 1 1 (O CO 't fl- 't 1 |LntOtDLnu->LntOtDtDtOCOtO O • 8 I 1 1 1 1 1 | 1 1 1 1 N C^ O DO ItOOOl — CVJ*d" 1 |OOCO-^-CD'^-CDl-OCr>COLOCDCNJ O OJ 1 — 00 in 1 — oocdi — lOirvcooj^LnMtDoo .SPra 1 I I 1 pi en t -ICvJOOCSICMOOvJ ra oj CT)>*rH<3-(Nooa)0^f cnoi — oocotocNjtor-.CNjtoto Q rHCMMiNiooro^r^ 1 1 (vj^rf* CO CO (O t ^ 1 ItOtOtOtDLOtDtOtDtOtOtOtO DO 00 I en 00 01 ^t *3-OOOC3CvJLO 1 ICsltOtOCVJOOLOCnOOCDi— it— it— 1 a> y— 1 too-) N'j-inmcoH'j-'j-cooN'* " COCO 1 I OOOOOOCXJOOOOOOOOOOOOCOOO a> E r— iM^(Osrs<*o)Hi — cxitOLnoco'3-tooocncD.— ilo 000 cNjrtm^j-^t"vi-Lnm | Ilololouo^j-^-^-^-'^-lololo ra XJ a> OOtOt— iLOOOCHt— (OLO I CN O 1 — OONNCOrHrsOOltD E r— it— it— ico"=*-coLT)Lor~ ooLor^-coococvjcocvjcoLOoo COCOrf"*t"*^-^-'* 1 ItOtDLOLOtOtOtOtOtOtOtOtO < ra >■ 4©0 Ot— iMro^-inuDNOOoOr- •CNJCO'^-Lntor^oocncDT— 1 cm ^T'^'^^'^'^'^'^'^'^'^LOU^LOLOLOLOLOLnLOLOtOtOtO cr cr cr 0- 0- cr a~ cr cr 0- cr cr cr cr cr cr cr cr cr cr cr cr l-ir> .0 a. >S JC >- 51? =5-2 s-S k» 3 TO ■O ra lb DO 3 v> E 5* "■oo v> >, .£«> >«j= CO ji a S-o .«= 2° a> 00 <- E*E £.<2 ■- — il 00 c a> oj S* F fc ra« TO O °-S. n. ra g; 0JO0 ot- ,0*3 £ v'. . . 0J . . >. , — .TJ -,t^ iavw Table 8 NUMBER OF PLANTS CLASSIFIED ACCORDING TO AVERAGE DAILY SALES OF FLUID MILK, CALIFORNIA, 1943-1960 Average daily sales of gallons of fluid milk Year Under 50 50 to 150 150 to 250 250 to 500 500 to 1,000 1,000 to 3,000 3,000 to 5,000 5,000 and over Total 1943 231 190 159 148 140 140 * 69 42 36 40 35 27 24 19 19 23 349 320 248 253 222 205 100 95 94 70 64 64 61 60 66 66 145 138 141 119 110 101 57 55 59 51 54 54 51 55 69 77 123 111 124 115 99 95 82 80 73 82 89 92 87 107 116 128 58 76 86 73 80 84 68 56 66 66 74 87 84 98 101 87 65 65 73 77 81 89 88 83 82 77 74 61 67 68 65 67 14 18 21 24 22 20 18 23 27 27 26 28 28 23 23 24 31 32 33 30 33 32 49 49 52 56 61 60 59 63 66 68 1,016 1944 950 1945 885 1946 839 1947 787 1948 766 1949 1950 . 1951 531 1952 483 1953 489 1954 469 1955 477 1956 473 1957 461 1958 493 1959 525 1960 540 * Dashes indicate data not available. Sources: 1943-1957: California Crop and Livestock Reporting Service, California Dairy Industry Statistics for 1943: Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacramento, 1943, and subsequent annual issues). 1958-1960: From special compilations. 40.0 per cent in 1958. Considering the "market" as that part of total food store sales which is not integrated with respect to fluid milk supply, the market share of chain-organized units increased from 29.5 per cent in 1950 to 45.1 per cent in 1954 and to 50.3 per cent in 1958. Clearly, conventional fluid milk proces- sors are now facing a more highly con- centrated group of buyers than in earlier years. Recent changes in the structure of sellers Fluid milk sellers may be defined in terms of their buyers: processing firms selling to wholesale buyers are classified as wholesale sellers, and those selling to retail buyers as retail sellers. Some proc- essing firms may handle both classes of sales. Since 1950 significant changes have occurred in the number and size distribution of all sellers, accompanied by a trend toward greater specialization. Other important changes include the in- crease in processor-retailer integration, the development and growth of "cash- and-carry" outlets by processing firms which sell directly to the consumer on their own premises, and the decline in importance of product differentiation. These structural changes are important because they explain changes in the com- petitive relationships among sellers and between buyers and sellers. Turnover among sellers The technological advances of the early 1940's had an influence on the structure of the group of sellers (processing firms) comprising the wholesale and the retail segments of the fluid milk industry. Al- [22 1000 - 900 800 O 700 CQ Z 600 500 \ I \ \ \ \ / *^ I I I I I I I I I I I I **Kr ^jty 1944 1948 1952 1958 Fig. 4. Number of fluid milk processing plants, California, 1943-1960. though wholesale and retail fluid milk operations constituted two separate seg- ments of the market during the 1950's, this distinction was not so clear before the introduction of paper containers and the subsequent trend toward specializa- tion and centralization of wholesale milk processing. The adjustment made by the aggregate of firms to changes in tech- nology, in buyer concentration, and in production-consumption balances is shown in table 8. These data indicate that during the period 1943-60 the number of small plants decreased while the num- ber of larger- volume plants (3,000 gal- lons daily and over) increased. The total number of plants decreased sharply be- tween 1943 and 1952, continued to de- cline gradually for a few years, and in recent years has been rising — mostly be- cause of the increasing number of proc- essing plants which carry on a cash-and- carry business (figure 4). Trends toward specialization are not reflected in these data but are indicated by the characteristics of firms which have entered and left the industry. The esti- mated gross and net changes in firm num- bers for the years 1950-1958 are shown in table 9. Samples of the populations of 381 entrants (firms that entered the in- dustry) and 493 exants (firms that left the industry) were investigated in order to estimate the general characteristics of each population. (See Appendix B for procedures.) The size distribution of the firms in the fluid milk processing industry indi- cates the existence of many small busi- ness units. When many firms go out of existence, a large number of entrant firms is required to replace them. This type of turnover among firms does not in itself suggest a structural change. However, if the exants are replaced by units of a different kind, this turnover might alter [23] Table 9 ESTIMATED NUMBER OF ENTRANT AND EXANT FLUID MILK PROCESSING FIRMS CALIFORNIA, 1950-1958 Year Entry of firms January 1 to December 31 Exit of firms January 1 to December 31 Net change for year Estimated number of processing firms, December 31* 1950 51 29 28 41 47 48 53 33 51 80 69 68 51 43 50 60 31 41 - 29 - 40 - 40 - 10 4 - 2 - 7 2 10 528 1951 488 1952 448 1953 438 1954 442 1955 440 1956 433 1957 . 435 1958 445 Total 381 493 -112 * The differences in number of firms between this and table 8 are caused by the existence of multiplant operations. Source: Data obtained from the California Crop and Livestock Reporting Service, the California Bureau of Milk Stabilization, and the California Bureau of Dairy Services on number of plants licensed and going out of business each year. the structural development of the indus- try. From 1950 to 1958 several differ- ences have been noted between the char- acteristics of entrants and exants. These characteristics are summarized in table 10. From these data, it appears that exants have generally been replaced by smaller and less diversified firms. While 38 per cent of the exants had participated in both wholesale and retail markets, only 21.2 per cent of the entrants did so. Al- though both entrants and exants were typically retail, a greater proportion of entrants specialized in cash-and-carry operations. Of the 381 processing firms entering the fluid milk business between 1950 and 1958, 55 per cent produced all of their raw product supply and 63 per cent were cash-and-carry operations. While these data do not tell the entire story, they suggest the following general structural adjustments. Firms handling from 1,500 to 24,000 gallons per month and serving both local wholesale and re- tail accounts, have been replaced in many instances by local retail firms, primarily of the cash-and-carry type. In the course of this adjustment, the total number of processing plants in the state decreased sharply until 1958, but since then trended upward (table 8) . Changes in concentration of wholesale firms As stated, specialization has tended to shift the emphasis of large firms toward wholesale operations and that of small firms toward retail operations. This in- fluence, along with other technological and economic changes, has affected firm growth patterns and thus the level of con- centration in the processing sector of the fluid milk industry. This section con- siders several alternative devices for measuring the degree of changing con- centration in the wholesale market. By treating the relative growth of firms as a random process, it is possible to use Markov processes for evaluating changes in seller concentration. 28 The purpose of this analytical technique is to investigate 28 For a discussion of Markov processes and their adaptation to the analysis of structural change in this industry, see Daniel I. Padberg, "The Use of Markov Processes in Measuring Changes in Market Structure," Journal of Farm Economics, vol. XLIV, no. 1, February, 1962, pp. 189-199. [24 Table 10 COMPARISON OF CHARACTERISTICS OF ENTRANT AND EXANT FLUID MILK PROCESSING FIRMS, CALIFORNIA, 1950-1958 Characteristic Per cent of exants Per cent of entrants Size (monthly gallonage) Under 1 , 500 10.7 13.1 32.1 25.0 19.1 35.7 38.0 72.6 21.4 17.5 1,501-6,500.. 32.5 6,501-10,500 20.0 10,501-24,000 12.5 24,001 and over 17.5 Produced all of raw product handled 55.0 Had both retail and wholesale accounts 21.2 Predominantly retail (cash-and-carry plu< cent of sales) ; home delivery equaled more than 75 per 86.3 Predominantly cash-and-carry (cash-and- carry over 50 per cent o f sales) 63.8 Source: Compiled especially for this study from information provided by the California Bureau of Milk Stabilization. the nature of the aggregate of forces affecting changes in size distribution and to represent graphically the tendencies of resulting firm growth. By observing the movement of firms between specified size categories over specified time periods, an equilibrium size distribution of firms can be generated. This equilibrium distribu- tion is that which would be expected to result if the initially observed activity continued through infinite time periods. The data used for this analysis were the wholesale sales volume for each wholesale milk processing firm for the months of January, 1950, 1955, and 1960. These data were obtained directly from the California Bureau of Milk Stabilization. Changes observed in the size distribution of firms are shown by the unbroken lines in figure 5 for each of these years. During the period between 1950 and 1955 considerable consolida- tion took place as the total number of firms in the wholesale industry dropped from 241 to 176. The number of larger firms (those controlling 50 per cent of the market) dropped from 6 to 5, while the number of firms controlling 80 per cent dropped from 25 to 17, and those controlling 95 per cent dropped from 90 to 56. If consolidation were to continue indefinitely at this intensity, the equilib- rium industry adjustment would place .03 firms in control of 50 per cent of the market, .23 firms in control of 30 per cent, and 6.45 firms in control of 95 per cent of the market. 29 The 1950-1955 equilibrium configuration of firms is shown by the broken line in figure 5. Between 1955 and 1960, the total num- ber of firms in the wholesale industry continued to decline — from 176 to 169. The number of firms controlling 50 per cent of the market remained 5, while those controlling 80 per cent dropped from 17 to 15, and those controlling 95 per cent dropped from 56 to 47. If con- solidation were to continue indefinitely at this intensity, the equilibrium industry would have 2.25 firms controlling 50 per cent of the market, 7.65 firms controlling 80 per cent of the market, and 30.53 29 This structural equilibrium is the type of industry size distribution which would result if the observed adjustment pattern were to continue through infinite time. Since the intensity of adjust- ment is related to many changing dimensions of the economic environment, a shift in intensity is likely to occur before the structural equilibrium is reached. For this reason, this analytical device is more useful as a reflection of intensity of adjustment than for purposes of forecasting future structure. [25] 1N3D d3d 'bhvhs i3»avw 1 *■ ■ST C $ it firms controlling 95 per cent of the market. The 1955-1960 equilibrium con- figuration of firms is also shown by broken lines in figure 5. This analysis shows that development of the size distribution of firms during the 1950-1955 period tended toward very high concentration. The equilibrium size distribution based on that period gives the largest firm more than 80 per cent of the market, while 95 per cent of the industry output is produced by only six firms. Activity during the 1955-1960 period also reflected a tendency toward consolidation, but to a lesser degree. Here, the equilibrium shows that the number of firms producing 95 per cent of industry output is reduced from the current (1960) 47 to 30 firms. When this type of analysis is appro- priate, it has an advantage over some other measures in that it takes into ac- count developments in the whole size dis- tribution instead of merely comparing one point on it at two different points in time. In evaluating the usefulness of this measure, it is helpful to compare it with some other measures that have been used for the same purpose in order to compare and contrast similarities and differences in results. 30 The measure of firm size dis- tributions most commonly used is the concentration ratio. This expresses the share of the industry output, or some other size dimension such as assets, con- trolled by a given number of firms. When an analysis of concentration of economic power is the objective, this measure may be useful. The concentration ratios for the largest, 4, 8, and 20 firms for 1950, 1955, and 1960 are shown in table 11. These measures also indicate consider- able consolidation in the 1950-1955 period, for the concentration ratios of the largest four and largest eight firms have increased greatly. The 1955-1960 period shows greater stability, the con- centration ratio of the largest four firms having decreased slightly from 45.1 to 44.9, while that of the largest eight in- creased from 67.0 to 68.7. The Orris Herfindahl concentration measure is designed to represent the de- gree of concentration throughout the in- dustry size-distribution. This measure consists of the sum of the squared market shares of all market participants, all measured as percentages of total industry size. If all firms are of the same size, the index is equal to the reciprocal of the number of firms. When only one firm constitutes the industry, the value of this index becomes unity. The Herfindahl index for California wholesale fluid milk processors is also shown in table 11. On the basis of these measures of structural development during the period 1950-1960 in the California wholesale fluid milk industry, the following ob- servations may be noted. In the 1950- 1955 period, by every measure, concen- tration increased. In the 1955-1960 period, concentration seemed to be gen- erally increasing. A slight decrease in concentration is observed in the "largest four" grouping — a category widely used ; but concentration increased in the other groups and by other measures. Compari- sons of the concentration ratios of the two time periods indicates that concen- tration of industry output among the largest few firms increased faster in the 30 Measures of firm configurations with respect to size distribution and the concentration of economic power have been discussed at considerable length. See: Gideon Rosenbluth, "Measures of Concentration," Business Concentration and Price Policy: A Conference of the Universities-National Bureau Committee for Economic Research ("National Bureau of Economic Research: Special Conference Series," No. 5; Princeton: Princeton University Press, 1955), pp. 57-99. M. A. Adelman, "The Measurement of Industrial Concentration," Readings in Industrial Or- ganization and Public Policy: Selected by a Committee of the American Economic Association ("Series of Republished Articles on Economics," vol. VIII; Homewood, Illinois: Richard D. Irwin Inc., 1958), pp. 3-45. [27] Table 11 MEASURES OF WHOLESALE SELLER CONCENTRATION, 1950, 1955, AND 1960, CALIFORNIA Concentration ratio, largest 4 firms* Concentration ratio, largest 8 Concentration ratio, largest 20 Herfindahl indexf 1950 37.5 59.3 77.1 ,051912 1955 45.1 67.0 83.2 068798 I960 44.9 68.7 85.0 071120 * Concentration ratio = the sum of the shares of wholesale sales volumes controlled by the given number of firms. t Herfindahl index = sum of squares of firm sizes, all measured as percentages of the total industry size. Source: Compiled especially for this study from information provided by the California Bureau of Milk Stabilization. early period. The Herfindahl index also indicates a stronger tendency toward in- creased concentration in the early period. Another device used to measure rela- tive firm size in an industry is the Lorenz Curve shown in figure 6. This presenta- tion of data measures the degree of equality — or inequality. In cases of a constant total number of firms it also measures concentration. In the present illustration, total firms decreased 29.9 per cent from 1950 to 1960, and inas- much as the number of firms producing 50 per cent of industry output remained almost constant, this measure shows less inequality in that part of the curve over time and more inequality in other parts of the curve over time. Because of the change in total firm numbers during this time period, the Lorenz Curves do not adequately measure changes in concen- tration within the industry. A further measure pertinent to this analysis concerns not what has happened to the large and "dominant" firms but rather to the group of larger inter- mediate-size firms who might "compete" more directly with the large firms in the market. For this purpose, records were obtained of firms with plants producing in excess of 3,000 gallons of milk daily, but excluding those operated by multi- plant national or regional firms. In 1950, the output of these plants amounted to 28.4 per cent of the total reported sales of fluid milk, including both wholesale and retail sales, in California. The per- centage held by these "major independ- ents" had increased to 32.4 per cent by 1952 and had risen to 43.2 per cent by 1957. By 1960 it had grown to 46.1 per cent. 31 The evidence that the number of wholesale fluid milk sellers in California has decreased and that a larger share of the total market is controlled by the largest few firms would suggest, all other things being equal, that competition among these sellers has been lessened. This conclusion is based on the belief that seller concentration in a market affects competition. As discussed earlier, market for fluid milk in that part of supply over which some con- trol must be obtained before sellers can effectively influence the demands facing rivals and so possess market power, in- cludes the entire state. However, for pur- poses of establishing minimum prices, the California Bureau of Milk Stabiliza- tion breaks down the state of California into various marketing areas. Twenty-six of these remained generally unchanged from January, 1950, to January, 1959, but several were combined during 1959. Useful information for 1960 is available for only 18 markets. The numbers of processors serving these marketing areas 31 These figures are not directly comparable to those presented in table 11 which related only to market sales of wholesale milk. the relevant California or [28] 40 60 PER CENT OF FIRMS Fig. 6. Lorenz curves of sellers in the California wholesale fluid milk market, 1950, 1954, and 1960. 100 are shown in table 12. In this tabulation, a seller was counted even though his ship- ment to a particular marketing area was quite small. The concept considered here is the total number of participating sup- pliers. 32 These data show that although the number of sellers in California has de- creased in total, those serving individual marketing areas have actually increased in some instances. Increases have oc- curred in 16 out of 26 cases and de- creases in 7 cases, while only 3 areas have remained unchanged. In consider- ing the averages of the first and last two years of the period, the number of sellers increased in 17 marketing areas, de- creased in 8, and remained constant in l. 33 32 It can, of course, further be argued that a seller may be a potential supplier in a market even though he makes no current sales. This would be true so long as there were no restric- tions — either institutional or economic — to entry to a given geographic area. For the present purpose, however, the more restrictive definition of a supplier will suffice. 33 Some of these marketing areas have changed from rural to urban use during this period — Orange County, for example. In this case the observed increase of wholesale sup- pliers from 5 to 17 reflects the change in urban development in addition to changes in com- petitive relationships. Across the several areas considered, however, these effects are most likely of minor importance. [29] Table 12 NUMBER OF WHOLESALE SELLERS, CALIFORNIA SELECTED AREAS, SELECTED YEARS Marketing area* Alameda-Contra Costa. . . Monterey-Santa Cruz — Marin Napa-Sonoma Solano Mendocino-Lake Los Angeles Orange San Bernardino-Riverside San Diego Imperial Ventura Santa Barbara San Louis Obispo Fresno Kings Tulare Kern Madera-Merced Stanislaus Sacramento Sutter-Yuba Shasta-Tehama Del Norte-Humboldt Siskiyou Inyo-Mono 1950 1952 1954 1956 18 9 8 16 8 6 45 11 25 14 6 6 5 9 16 9 12 11 11 10 18 6 4 5 4 1957 1958 21 7 10 17 9 4 36 15 27 13 7 7 6 6 15 8 16 13 12 10 11 8 7 6 5 1959 1960 * All marketing areas having changes in boundaries during 1950-1959 are excluded. t Dashes indicate data not comparable with previous years due to changes in marketing area boundaries. Source: Compiled especially for this study from information provided by the California Bureau of Milk Stabilization. Changes in concentration of retail firms An investigation analogous to that de- scribed above for the wholesale market is not possible for the retail market. While wholesale market participants are typically large specialized firms with well-defined characteristics for which rather complete data are available, retail distribution by processing firms is done through two major and some minor sub- channels, with typically small firms for which available data are incomplete. Since 1957, these subchannels have been classified as home delivery, cash- and-carry, peddler, and nonreporting. Home delivery is self-explanatory — sales by processors directly to consumers' homes. Cash-and-carry represents sales to consumers on the premises of the processing firm. Before 1957, these sales, were included in the category "peddlers and platform" and were indistinguish- able from the peddler group — distribu- tors who do not themselves process any milk. The nonreporting category includes processing firms who produce all of their raw product and are not required by law to file monthly reports. Sales volumes for this group are estimated from an un- classified source. Retail fluid milk sales volumes, by type of seller, are shown in table 13 for [30] Table 13 RETAIL FLUID MILK DISTRIBUTION, BY TYPE OF SELLER CALIFORNIA, 1950-1962 Year Total Home delivery Peddler Cash-and-carry Non-reporting thousand gallons 1950 158,178 163,541 170,472 178,711 188,202 201,321 212,084 211,376 210,698 215,955 216,826 210,605 207,731 112,836 112,615 111,277 114,429 119,530 127,859 132,812 130,091 124,178 126,664 125,780 119,647 115,734 * — 1951 _ 1952... 46,380 50,635 55,244 62,069 67,356 12,815 1953 13,647 1954 13,428 1955... 11,393 1956 11,916 1957 42,717 42,363 43,687 43,676 45,385 47,284 25,552 30,337 36,715 37,033 45,573 44,713 13,016 1958 13,820 1959 8,889 1960 10,337 1961... 1962 * Dashes indicate data not available. Sources: 1950-1956: California Crop and Livestock Reporting Service, Dairy Information Bulletin (Sacramento, monthly issues). 1957-1962: Idem, California Dairy Industry Statistics for 1957: Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacramento, 1957, and subsequent annual issues). the period 1950-1962. The major sub- channel, home delivery, decreased from 71.3 per cent of total retail sales in 1950, to 56.0 per cent in 1962. This decline in relative importance of home delivery sales by processing firms has been brought about by the rapid increase in cash-and-carry sales since 1957. Since the relevant market for home- delivered fluid milk is more restricted geographically than that for wholesale operations, data about seller concentra- tion must relate to a more local environ- ment. The marketing areas established by the California Bureau of Milk Stabili- zation are suitable for this purpose. 34 However, changes in the boundaries of some of these marketing areas limit their use for measuring concentration changes over time. Therefore, only the 3i Some of these established marketing areas probably correspond better with the concept of relevant market than others. This difference, however, is not sufficiently large to affect this analysis. San Diego, Kern County, Fresno, and Del Norte-Humboldt marketing areas have been chosen to represent retail markets because their boundaries have not been shifted between 1950 and 1960, and the markets are self-contained, that is, supplied essentially by local proces- sors. It has been assumed that buyers within these areas have no potential sup- pliers from other sources, since this is generally true for retail home-delivery. Changes in retail concentration in these four areas between 1950 and 1960 are shown in table 14. In general, the market share of the largest firms was large but firm growth and changes in seller concentration varied both within and between the four areas. Two of these showed a consistent trend over the total 10-year period: Kern County toward greater concentration, and Fresno toward lesser concentration. Del Norte-Humboldt tended toward higher concentration in the first five-year period (1950-1955), while in the latter period (1955-1960) [31] w =3 V O c5 o m oo oo -C to o ^t cn oo 00 o cd r-» oo a> _SS CO CO E o O o to CD c a> o 5 C3 >=s- csi r~. CD CD *d" CNJ CXI a) 1— ~ OO ^3" CO a. CO __l Kg ^J- 1 — CD CD 3 ljz o Z co CD 1—1 Q a> 3 O < CO *" o ro rv n- - 00 cz> co CD r- LO GO O CD 1 — CD CO cn CO 1-1 CO —1 E CD LO CD o o $ h-~ LO CD CO o t OJ LO LO UJ CD LO cx> r>» i — i£ CD 00 cn rs - Od CO ,/. LxJ > 1*1 OO I — O IS Zi UJ O LlJ ^ 3 o CD CO O CO LO CO "t c/> cnj r--. r-» o i oo CD OO OO CD < co I— <5 1 UJ E CC LU o o O ■g js h- LO CD CD ^t tO CO CM Z ■— • ,_ CO t cn ih cn CD OO CO CO CO LO — 0_ CO Z —1 o h- CO +_, -a o -Q E 3 X CD 2. DO CD o o ■ — eIN ern. resn an C a iC U. - a £ zr. — Oi < — BE n ■x. 1 ~ - — o w £ - u s S ^ < 2 an U Tf O 3 lO 2 2 (7> ~* •£ a - s- 'S ■2 § n *- S & a, 4 - 1 /- — C\J u - - •r* v o> fc .* — »5 <_ W .fl u S-c -3 W c £ t> c o A - If) <7> S* (00L=6£-ZC6l) X3QNI Appendix Al Estimation of concentration curves of California Supermarkets, 1950-1958 A concentration curve shows graphically the relationship between number of firms and cumulative market shares of firms ranked from largest to smallest. The shape of these concentration curves indicates the size distribution of firms within an industry. For example, when all firms within an industry are of equal size, the concentration curve is a 45-degree line running through the origin. An indication of changes in the size distribution of firms may be obtained by comparing the concentration curves of an industry over time. The information given in tables 5 through 8 gives directly three points on these concentration curves: First, they must start through the origin, because zero firms can contribute only zero share of the market. Second, the total number of firms in the industry must contribute 100 per cent of the industry output. An additional point is obtained by using the concentration ratio for the largest 10 firms (table 8) . This information adequately describes changes that have occurred in the size distribution of a very small number of the larger firms. It is of interest for our problem also to investigate the nature of growth patterns of the group of smaller chains, since the larger firms are essentially all supplied by captives. Details concerning the number of chain store firms and the number of stores they operate, by years, are given in table 14. No information concerning the market share of these smaller chains is available, however. It is known that the average of annual sales of all stores is much below the average per chain store outlets for which data are available. For example, in 1958 the average of annual sales of all food stores in California was $424,000, while the average of sales per unit of the largest 10 chains was $1,375,169, or more than three times greater. It seems reasonable to assume that the average of annual sales of all chain store units is nearer to the average for the largest 10 than to the average of all stores. On a na- tional basis in 1958, the annual sales per store of all chain store outlets were 83.6 per cent as large as the annual sales per outlet of the largest 10 chains. 64 By applying this weight of 0.836 to the average of annual sales per outlet of the largest 10 firms in order to estimate the average of sales of all chains, an additional point on the concentration curves may be obtained. Computation of this data for 1950, 1954, and 1958 is shown in table A-l. It is possible that the use of the 1958 national weight for other years may bias this estimation procedure, for casual ob- servation suggests that some of the smaller chains of recent origin have higher an- nual sales per store than the largest chains. If the annual sales per store of all chains have in fact been gaining over the past few years in relation to the annual sales per store of the largest 10, this estimating procedure understates the increase in market share of all chains from 39.0 per cent to 59.5 per cent as shown in table A-l. 64 This Week Magazine, footnote table 7, p. 53. [52] Table A-l ESTIMATED MARKET SHARE OF ALL CHAIN STORE OUTLETS CALIFORNIA, 1950, 1954, AND 1958 Year Annual sales per outlet, largest 10 chains Estimated annual sales per outlet, all chains Number of chain store outlets Number of chain store firms Total grocery store sales Market share of all chains • 2 3 4 5 6 dollars 1,849 1,982 2,291 232 311 356 million dollars 2,557 3,292 4,430 per cent 1950 1954 1958 645,549 1,099,267 1,375,169 539,679 918,987 1,149,641 39 55.3 59.5 Sources: Col. 1: Table 7. Col. 2: Column 1 weighted by 0.836. Cols. 3, 4, and 5: Table 5. Col. 6: Column 2 times column 3 divided by column 5. Appendix B: Procedures for developing the population of entrant and exant fluid milk processing firms, 1950-1958 All firms which conducted bottling operations for raw or pasteurized fluid milk in the state of California during some part of the period January 1, 1950, to December 31, 1958, and which commenced or terminated processing operations in that time period were included in the population of fluid milk processors. In order to identify this population, a survey of available data was conducted. Sources of published data No published data are available concerning fluid milk processing firm numbers or entry and exit of such firms over time. The following sources, however, contain some information concerning the number of fluid milk dealers and fluid milk process- ing plants, by years. 1. Stewart Johnson, Dairy Marketing (Storrs: University of Connecticut, Depart- ment of Agricultural Economics and Farm Management, 1956) . The number of licensed fluid milk dealers in California, including peddlers as well as processors, is given in this publication for the years 1950-1956. Each plant of a multiplant firm was counted as a separate entity. This series is shown in table B-l. 2. California Crop and Livestock Reporting Service, California Dairy Industry Statistics . . . : Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacramento, annual issues). This publication contains a series of numbers of fluid milk distributors in California for the years 1951-1957. The definition of these distributors indicates that only those with processing plants are included. [53] Table B-l NUMBERS OF FLUID MILK DEALERS AND PROCESSING PLANTS CALIFORNIA, 1950-1960 Year Licensed dealers Net change from previous year Processing plants Net change from previous year 1 2 3 4 1950 2,181 1,701 1,752 1,651 1,599 1,713 1,598 * -480 51 -101 - 52 114 -115 531 483 489 469 477 473 461 493 525 540 1951 _ 1952 -48 1953 6 1954 -20 1955 8 1956 - 4 1957 -12 1958 +32 1959 +32 1960 +15 * Dashes indicate data not available. Sources: Col. 1: This series consists of the number of licenses issued for each year. This information was obtained directly from the California Bureau of Milk Stabilization, Licensing Section, and was published in Stewart Johnson, Dairy Marketing (Storrs: Connecticut Agricultural Extension Service, 1956). Col. 3: 1951-1957: California Crop and Livestock Reporting Service, California Dairy Industry Statistics for 1951: Manufactured Dairy Products, Milk Production, Utilization, and Prices (Sacramento, 1951, and subsequent annual issues). 1958-1960: From special compilations by the California Crop and Livestock Reporting Service. Again, each plant is counted as a separate distributor. This series also appears in table B-l. These data show the net changes in numbers of dealers and processing plants, by years. No further published data are available concerning the breakdown of this net change in number or identity of the firms leaving or entering the industry. In view of this lack of published data, unpublished sources of information were con- sulted. Sources of unpublished data For several decades each fluid milk distributor in the state of California has been required to file reports with the California Crop and Livestock Reporting Service concerning receipts and utilization of milk. The California Department of Agri- culture in turn published annually the Official List of Milk Products Plants and Fluid Milk Distributors. Publication of this list was discontinued in 1947, and a coding system of identification was later devised. Code numbers were assigned to all dis- tributors named in the final issue of the list and have been assigned to all distributors who have entered the industry since then. Receipt of each report is recorded on a card which is kept current for each distributor, indicating whether the distribution unit operated a plant, conducted retail home-delivery operations, etc. Upon termina- tion of operations by a distributor, this card is placed in an "out of business" file. As new distributors enter the industry, new code numbers are assigned and the detail of their reports is entered on new cards. When ownership of a distribution unit is transferred between firms or individuals, a new card is prepared for the new firm, but the original code number is retained. The old card is then removed from the active file and placed in the out-of-business file, which thus reflects changes in ownership. [54] The Licensing Section of the California Bureau of Milk Stabilization also main- tains a current file of licensed distributors completely separate from the records kept by the California Crop and Livestock Reporting Service. Here, too, when distributors terminate operations their cards are placed in an out-of-business file. Although the coding of the Licensing Section is different from that of the California Crop and Livestock Reporting Service, identity may usually be established by firm name and address. This file provides important supplementary details concerning the functions performed by the distributors (processing, production, etc.). In addition to these data, still further information may be obtained from the California Bureau of Dairy Service, which maintains a current file of all fluid milk processing plants and has available an out-of-business file for a part of the 1950-1958 period. Compilation of the target population It was decided to combine the available information into a yearly series of firms, maintaining the identity of all firms entering and leaving the industry. The list of code numbers and the out-of-business files were the primary sources of information used in this compilation. Between 1947, when the coding system was initiated, and December 31, 1958, 1,682 code numbers had been assigned. In addition, 73 entrants had begun operating under the code number of the previous owner. To obtain all possible information concerning the target population, an effort was made to determine the current status of each of these 1,755 plants. Among these plants were many that did not process fluid milk. Also, code numbers had been assigned to each operating unit of multi- plant firms. The first step in determining the current status of these 1,755 plants was to ascertain which of them continued in operation after December 31, 1958. These accounted for 445 of the plants. The remaining 1,310 distribution units were then examined in an effort to determine which belonged in the target population. The out-of-business files provided a classification of functions performed and the date of termination of operations. Using this information, it was possible to eliminate from the target population distributors in the following categories: 1. Distributors that did not process fluid milk (peddlers) . 2. Distributors that terminated operations prior to 1950. 3. Plants that were a part of multiplant firms. (These multiplant firms, just as single-plant firms, were counted only once.) 65 Fluid milk processing firms that terminated processing operations between January 1, 1950, and December 31, 1958, were classified as elements of the exant population, while those that commenced processing between these dates were classified as elements of the entrant population. In a few of these cases, there was no information as to whether the processing operation was performed. These were nevertheless counted in order to avoid the possibility of underestimating the target population. Given the above information, it was possible to eliminate or classify all but 140 plants. Letters were sent to the appropriate local municipal and county health authorities and the regional offices of the California Bureau of Milk Stabilization asking for information regarding these plants. With the details obtained from these 65 This procedure in developing the target population serves to explain the variance occurring between the series shown in table 18 and the published statistics regarding numbers of milk processing plants. [55] sources, 95 of the 140 units were classified. No further information could be obtained concerning the remaining 45 distributors. The data used in estimating the target population also served as the basis of a series of firm numbers with entries and exits by years for 1950-1958. Given the estimated number of firms operating at the end of 1958 (445) and the net change for each previous year, the number of firms operating at the end of each year was developed. Drawing the sample 66 From the population of 381 entrants and 493 exants, a sample was desired which would ensure a standard error of estimate of the mean of not more than 10 per cent. Size of sample influences the standard error of estimate. To simplify the discussion, assume that the size of the population, N, is 500 and that the population parameter to be estimated is the proportion of the population having some particular attribute. The standard error of estimate also depends on the variability that exists in the population. A proportion of 0.5 corresponds to the largest variability of the popula- tion when sampling attributes. When sampling from a very large population, the standard error of estimate is equal to ,— where a 2 is the population variance. For a population proportion, v, of 0.5, o- = V 7r ' (l _7r ) ~ V-25 = 0.5. Suppose that the objective is to estimate the population proportion it — 0.5, using a method that will 0.5 ensure an average error of 10 per cent. This implies that /'— = 0.05 and that n should be 100. In other words, using unrestricted random sampling from a large population having a proportion of 0.5 possessing some attribute, a sample of 100 elements is needed to ensure a standard error of estimate of 10 per cent. Actually, the accuracy is better than this, since the population is not large relative to the sample size. Under these conditions, the standard error is not /— but /— ^4 . In other words, the v n V n N " 1 standard error in the instance cited would not be 0.05 but (0.05) — , or approxi- mately .04. In the particular case of the population of 493 exants, it may be observed that the sample size of 84 achieves the desired result of a coefficient of variation of less than 0.1: 0. 5 493 - 84 I n n/l r- i 0.045 ^ r\ i y'ir —*%- + 0045 and irr < - 1 - In the case of the population of 381 entrants, a sample of 80 achieves the desired result : 0. 5 . 381 - 80 r\ r\A a j 0. 044 ^ r\ t y/W "lib— = 0.044 and -^y- <0.1. The samples were drawn using random numbers. An additional 40 alternatives were drawn from among the exants, while 20 alternative entrants were drawn. This num- ber of alternatives was more than adequate in both cases. 00 The authors are indebted to Professor J. N. Boles for his assistance in drawing the sample for this phase of the analysis. 5m-6,'64(E4015)J.F. [56