SOME MEASUREMENTS OF CONSUMER‘ DEMAND FOR MEATS Thesis for the Degree of Ph. D. MICHIGAN STATE COLLEGE Harold M. R.i!ey 1954 ‘Hfibl‘d III III II II II III III II III IIII III III III III IIIII 91 This is to certify that the thesis entitled . Some Measurements of Consumer Demand for Meats presented. by Harold M. Riley has been accepted towards fulfillment of the requirements for Doctor of Philosopm degree inAgnicnlLural Economics WW — Date M 0469 r g1? . .7 m . «7f ' -- , ~- --; 1 1- '1 ~-1 1er v ’ SWafiidASfihflnLTSCM COuofllu DEIAHD FOR MEATS by Harold M. Riley AN ABSTRACT Submitted to the School of Graduate Studies of Kichigan State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1954 Approved J/Wfl' $294M¢ 1 , t__ Harold M. Riley ABSTRACT This study was an attempt to measure consumer responses to changes in prices for different kinds of meat. Previous demand studies of this nature have been based almost entirely upon annual average prices and quantities for broad groups of meats for the entire United States. In most cases the period for these studies has been the interval between World Wars I and II. It was believed that demand relationships based on more recent observations and for periods of time shorter than one year would be a useful supplement to these earlier studies. The basic data for this study were the weekly food pur- chase records of the H.S.C. Consumer Panel. This panel is composed of approximately 250 families selected so as to be representative of the city of Lansing, Lichigan. Weekly aver— age prices and quantities purcaased per family were available for a two year period, July 1951-June 1955. Fortunately this was a period of substantial price changes for both beef and pork. Single equation demand models were fitted to the data using least Squares regression techniques. The basic equa- tions expressed the quantity purchased of one kind of meat as a function of the price of that meat gr up, the prices of competing meats, and a temperature variable. It was found that the price elasticity of demand for both beef and pork were near unity at their respective mean values. Beef prices seemed to have a significant influence on pork purchases, however, pork prices had a somewhat weaker influence on beef purchases for the period studied. The prices of sausage, poultry or fish did not have a significant influence on either beef or pork purchases. The price elas- ticity of demand for sausage meats was not significantly dif- ferent from zero while poultry and fish appeared to have elastic demands. The price elasticity of demand for all meat was about -.7 at the mean value of price and quantity. Temperature was significantly related to meat consump- tion during the warm season of the year. An increase of 10 degrees in the weekly average of mean daily temperatures de- pressed purchases of pork, beef, and all meat by approximately 8 percent. A preliminary analysis of demand for retail cuts of meats indicated that the price elasticity of demand for beef steak was highly elastic. The price elasticity of demand for beef roasts, ham and pork chops were slightly elastic while the demand for ground beef and bacon was slightly inelastic. SOLE LEASURELEI‘ITS OE? COIISUESSR DEMAND FOR LEATS by Harold h. Riley ATHESIS .. __ .—_—' Submitted to the School of Graduate Studies of Michigan State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics \ 1954 (4/57/5‘ 7 érl/ef/ék AC I’ll-TOY. HEDGELISZITS The author wishes to express a special debt of gratitude to Dr. G. G. Quackenbush for his helpful and patient counsel in supervising this investigation and in the preparation of this manuscript. Special thanks are also extended to Dr.J.D. Shaffer who provided invaluable assistance in the planning of IBM tabulations and for his constructive criticism of results at various stages in the study. Acknowledgement for advice on the more technical aspects of this analysis is due Dr.L.L. Boger. Appreciation is also expressed to Dr. R. C. Kramer and Dr. B. C. French for helpful suggestions as the study prOgressed. Sincere thanks are offered to the clerical staff in the Department of Agricultural Economics who carried on the laborious task of statistical analysis. EA? Till-i I . CII.’ . DTJR II. T TABLE O? CO TSITS INTRODUCTION ooooooooooooooooooeooo The I.i.S.C. Consumer Panel ................. Previous Studies........................... Objectives of the Study ................... Usefulness of Results ..................... LORY ATD LLASURBLLNT OF DSLAND i I .0 NT LRRLLATLD CCIfiOII )ITILS....... h—l IntI’OdLlCthD coo-coo.oooooooooooooooeoeoooo Basic Concepts of Demand Theory ........... measurement or Defiando00.000000000000000... OLA? TEE III. iALvICLS STUDILS OF DLZIID FOh HEAT Studies Based on harket Data .............. Cross-Sectional Budget Studies............. S‘LLZL-lary 0.0ICCOOOOOOOOOOO0.00000000000IDOOO C-:APTQ:I F]. T1:.1: ~"\‘/".,I..I.C.I.1J fl ID .LIxilU J12“; CF DAonooeoo The Operation of the Consumer Panel........ The Characteristics of Lansing ............ Representativeness of the Panel Over Time . Preliminary Processing of the Data ........ Limitations of the Data ................... Level and Pattern of heat Consumption ..... Lansing Prices Compared to Detroit BLS Prices. CKAPT‘I V. SITSLL EQUATICI ELAND MODELS FOR GROUI‘PS .11?I.[IAE'OOOOCOCCOOOOCOOCCOICOOO IntrOdUCtiQn OOOOOOIOOOO0.0000000IOOOOOOOOOIOO Evaluation of Leekly Observations .............. Principal Variables to be Included in the Lodel. Other Variables Affecting Heekly Heat Purchases. The Form of the hathematical Function........... The Basic Demand Equations ..................... memw H CO OKOCD PO (fl 0'1 PIN (fl {\3 O] O O 65 69 77 O U 84 92 C EEAPTL“ R Int Var lies 3113.1: The Demand for Selected Retail Cuts CIZAP. TL‘ Int Variations in Prices and ‘uantities KL lIes The CTAPT LR VIII. TIE DLLAHD FOII oA'LAJL ALL, 1Q--.ooooooooo.ooo Int The The The CHART LR v.-. VI. I... 33mm 1303 a-.. rOduCtion .OOOOOOOOOOOIOOIOOCOOOOO iations in Prices and Quantities ults of Reéression Analyses lary 0.0.0.0...OOOOOOOOOIOOCOOOOOO V’II. TILED D2: JJ’L' ID FUR POILI‘: .oooooooo rOdllCtiOn .OOOOOOOIOOOOOOCOOOOOOOO ults of Regression Analyses roduction .................. Demand for La‘”a Leats .. Demand for PoultrJ heats .. Deaand for Fish ........... IX. T13 DELAJD FOR ALL LSAT C::APT LR X. SUZILJLRY AID CO ICLUS IO S BIBLIO GR APPLIIDH A.L::f .OIOOOCOOOOOOOOOOOIOCOO D )4“ Demand for Selected Retail Cuts 120 120 123 129 144 I45 155 155 f- .3 UK.) 164 I75 173 173 179 185 190 Table Table Table Table #3 able Table Table Table Table Table 10. Table 11. l. 4. 5. 9. LIST OF TABL S Price elasticity as related to total I’GVblx-JC on00000000000000.0000...-coo-coco. Elasticity of demand for meat, Sweden, lObl“39 00.00.000.000...oooooooooooooooooo Income- -con31L1ption elasticities for selected meat items, urban fa.nilies United States, Sprilid 1948 00.00-00.00... Si5nificance of differences in per capita consumption of selected meat items when related to Specified socio—econonic characteristics, Lansin5, sprin5, 1050.... Income—quantity elasticities for meat items by size of family, Lansin5, sprin5 lGSOOOOOOOOIOCOOOQOOOOOOOOOOOOOOOOOOOOOOOOO Characteristics of 1050 censas the Lansin5 pepulation, Production and consumption of meats in 1410111me 00000.00...-coo-000.000.000.00... Characteristics of the average fa;ily in tne n.S.C. Consumer Panel at different time periods compared to the avera5e family in the 1950 census................. Frequency distribution of families parti- cipatin5 coztinu ously in the L.S.C. Con— sumer Panel for varying lengths of time... Quantities of meats for tile year July Conswolel‘, Pallel 0.00.0.0...OOOOOOOOOOOOOOOO purchased by families heat consumption by I.S.C. Consumer Panel compared to U.S. avera5e, July lO51-June 1902 0.0.0.0000...OOOOOOOOOOOOOOOOOOOOOOOO 051 to June 1052, L.S.C. 64 75 C") CO Table Table Table Table Table Table Table Table Table 13. 14. 15. 22. Consumption pattern for red meats in Lan- sin5 conpared to patterns for the North Central re5ion and the United States........ Average purchases of meat durin5 holiday weeks compared with avera5e weekly pur- chases for the remainder of the year, h.S.C. Consumer Panel, 1052................. flecks of the calendar year durin5 which holidays occurred and the day of the week upon which the holiday fell, July 051- r, (1.11.110 190L500.00......OOOOOOOOOOOOOOCOOOOOO... H C p Avera5e weekly purchases of meat by h.s.o. Consumer Panel Families, July 1951- J1melg'53 000.000...OOOOOOOOOOOOOOO00.00.... Meat consumption per person by quarter—years U.S. avera5e, 1050 to mid—1955 ............. Simple correlations between pairs of vari- ables, beef equation, July lQSl-December 1‘35?! OI.0....OOOOOOOOOOOOOOOOOOOOOI0.0.0.... Summary of re5ression results, beef equation JUly lQSl-DGCGMber 1952.00.00...coo-0000000. Simple correlations between pairs of vari- ables, beef equation, July 1952-June 1955... Summary of re5ression results, beef equation Jilly lQSZ-J‘dlflle 1953 OOOOOOOOOOOOOOOOOOOOOOOO Relative importance of selected retail cuts of beef, M.S.C. Consumer Panel, July 1052- Jl/LTIB 1953 OOOOOOOOOOOOOOOOOOOOOOOIOOOOOOOOOO Comparison of purchases of selected retail cuts of beef, 2d quarter of 1952 and 2d qfllarter Of 1955 00.0.0.0...OOOOOOOOOOOOIOOOC Changes in relative quantities and expendi- tures for ground beef, roasts, and steaks, h.S.C. Consumer Panel, 2d quarter 1952 com— pared to the 2d quarter of 1955 ............ C) m 102 105 120 25 146 148 150 Table Table Table Table Table Table Table 24. 29. 50. -‘ Relative importance of different retail pork cuts, E.S.C. Consumer Panel, July lDSZ-Jlllqe1955.00.00.00000000.000.00.000. Summary of re5ression results, pork equation, uly lQSl-December 1952........ Summary of regression results, pork equations, July 1952-June 1955........... .1) .L __. _. __ . .n 1 3 no A, - J. .. nelative importance oi aiiielent sausa5e items, K.S. . Consumer Panel, July 052- JLUIS1955......OOOOOOOOOOOOOOOOO000...... Relative importance of different kinds of poultry meat, M.S.C. Consumer Panel, July 19:32-‘3-1338 1.95:7)...OOOOOOOOOOOOOOOO... Summary of re5ression results, sausa5e, poultry and fish equations, July 1951- JGCGMbCr 1952.00.00.00000000000coco-coco. Summary of re5ression results, equation (6.3), all meat, July 1951-June 1955..... 156 167 171 179 186 195 204 VIJ Indifference maps for related comma di ties. Inco e a1d substitrtion effects of a price 0118.4"er for related CO lOLllthS............ i et\.een family lACOMO and censumotion, urL an housekeepin5 fami- " States, 35. rin5, 1943 ........ Avera5e of weekly ince e reported by h.S,C. Consune r Panel families, July 1051-June /.r"rv \1:"~ UUO...................................... '3 1- . .° .. (1 Coupar ison o1 ha1s1n5, 1.5.3. Co1s1uer Panel pri N: s and Detroit LLS prices for selected 0 ate of beef, July 1951-June 1955 Conparisou of Lansin5, L.S. C Panel prices and Detroit BLS pr ”F selected cuts of porlc, July 13 Q (‘1 O (W 1‘ Compar’son of La:1sin5, 3.5. v. Consumer Panel prices and Detroit BLS prices for ham and frankfurters, July 1951- June 1955. E.ee1w1 avera5e purchases of dif' lzinds of nests by fazilies, 1. suuer Panel, July 1LC1- -June 19 bee {1y ave1a5e purchases of different kinds of meats by fanilics, fi.S.C. Con- suuer Panel, July 1952-Juue 1255.......... Heekly avera5e purC1ase s 01 all meats by families, L.S.C. Consuuer Panel, July lQSl-JLIJB1952...ooooooooooooooooooooooooo Weekly ave1a5e purchases of all meats by fauilies, L.S.C. Consumer Panel, July lSSB-June 1952.00.00.00.0.000.000.0000.o.o (9 C21 (’3 (11 r) C) Figure XII. Variations hlpriee elasticity alon5 a demand function which is linear in arithmetic values and the effect of a parallel shift in the dem1‘ 1d iunction...... 117 Fi5ure XIII. COHparison of eta il beef prices in Lansin5 titl w:olesal e prices for selected 5rades of ‘ “ “ eee1 at oaica 50, four week avera5 es ; July 1951- Ju11e IO 5'}. 1955.... l Fi5ure XIV. {elat101snip eet11een wee11;1j avera5e pur- el1ases aad :riees 01 0031, L.S.3.5010rzer Panel, Jul~ lSSl-June 1355, holiea~s 051t- deOIOOOOOOOOOOOOOOOOOOOI...OOOCOOOOOIOO... 1‘4 Fi5ure X7. Residuals fro; beef equation (1.1) cogpared to the 1.eeklj avera5e of mean daily su;11er Lo1>blltulCS 111T&-181115.................... 13:5 Ei5ure XVI. Lelati ans Aips betwc eon teel {ly avera5e pur- chases and prices for selected reta1l cuts of beef, J1lj 952- June 1955, holidays onit- ted...O...I...O00......OOOOOOOOOOOOOOOOOOCO 155 Fi5ure XVII. Co. a11so of retail pork prices in Lansin5 with prices Ior selected wholesale cuts in Chiea5o, four w ek avcra 5es, July 1951- JL‘L‘TC195.35.................................. lUO 115Lu%3XVIII.ifielationsni oetw een wee kly average pure Q A chases and prices 01 pork, L.S. C.C ensumer Panel, July l951-June 1955, solida"s oait- tedOO:OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.. 162 Fi5ure XIX. Residuals froxz pork equation (2.1) compared to wee} :13 ave ra e of mean daily summer tea— pGI‘atUI’eSlu' (1.10.1.--0 0.00.0000.0.00.0.00000 ‘30 P15ure XX. fielatior.snip between VJGOli lv average Pur- chases and prices of saus a 58, ;.S.C.Cons1mr er Panel, July l95l—June 1955 ............. 130 Figure Figure Figure 4 P1.) H- mfxr JUL . JuxII o XXIII. XXIV. 'r~ «7 AA rjrrr x V 0 your I I . KKVIII. Residuals from sausa;e ‘uat-J (3.1) co :pared to xec‘l, average of LHGfll daily iincr te1ger aturcs in la‘. siLg... Relationship between weekly avera Q0 purchases and prices of poultry, L..:.C. Consumer Tanel, July CSl-June 1355, , J L lidays omitted fr Pesiduals p J ultry equ atic q(4.l).. Relationship between weekly average purchases, and prices of fish, L.3.C. Consumer Panel, July 1951- June 19 55, holidays omitted ...................... Residuals fron fish equation (5-1)-°'°° Relationsliip between x.*eeklj avera e purcllases and prices of all me cat, UL.S.3 Consu';1erB—a11el,July l‘._)5l-June 1:23.55, holidays 0 ".itted ..................... duals from equation (6.1) for all , las‘ half of 1951 and last half ~52; comparisons wit1 summer tem- ra'ures '3 I"; " S v n. AQA. 09 1 Hc+H° H) (D O ch ()0) o p O 00.0.00...OOOIOOOOOOOOOOOOOOO llesiduals from o (6.1) for all meats, iirst lialf of l952 and first half of 1955; connarisons with s111or telperatures H u q H (L) H 194 198 CHAPTER I INTRODUCTION The H.S.C. Consumer Panel This dissertation is a report on an analysis of con- sumer demand for meats. The principal source of data was detailed food purchase records provided by some 250 families in Lansing, Lichigan. This group of families will herein- after be referred to as the "i. 8.0. Consumer Panel." Each week these families record their food purchases in a diary provided by the Department of Agricultural Economics of hichigan State College.1 (See appendix). Information re- ported includes the price, quantity and total expenditure for each food item purchased. Additional information is also reported on current income, size of family and number of meals served. The M.S.C. Consumer Panel has Operated since Harch 1951. The research project that supports the panel was originally set up to continue over a period of ten years. The data l The organization and operation of the H.S.C. Con- sumer Panel is under the direction of Dr. G. G. Quackenbush and Dr. J. D. Shaffer. available for this study covers 104 weeks beginning in July 1951 and ending in June 1955. During this period substantial fluctuations occurred in retail prices of beef and pork, thus making it feasible to study consumer adjustments to price changes. The panel is unique in that it provides a complete rec- ord of each individual family's food purchases on a weekly basis over an extended period. host of the previous demand studies have been limited to the use of two principal types of data. One type is annual time series of price and quanti- ty estimates for broad groups of commodities for the entire United States. The other main source of data has been a series of cross-sectional studies where food purchase data for a given week are obtained from a sample of families re- Siding in selected localities. Previous Studies Most of the empirical studies designed to measure price and cross elasticities of demand for meats have been based on annual time series data for the interval between World Wars I and II. In general, the results of these studies have failed to provide a reliable basis for forecasting price-quantity relationships during the post World War II period . This difficulty can be attributed, in part, to the rapid changes 5 in the economic and social environment during the past thirty years. At present the number of annual observations are too few to support a rigorous demand study limited to the post- war period. The usefulness of demand elasticities derived from annual observations is also limited by the high degree of aggregation which goes into the raw data. The demand characteristics for more narrowly defined commodity groups and for periods shorter than one year should be of greater usefulness to food merchandisers.’ Cross-sectional studies have provided useful informa- tion relating meat consumption to various social and economic characteristics of the families interviewed. Difficulties have been encountered, however, in attempting to predict the effects of changing income levels on meat consumption based on results of these studies. The data from the 1.8.0. Consumer Panel can be analyzed both as a time series and on a cross-sectional basis. Due to its flexibility, both as to time periods studied and degree of aggregation of commodities, the panel data should yield some worthwhile measurements of demand. These measurements will supplement those already available from the studies based on annual time series or cross-sectional survey data. Objectives of the Study The primary objective of this study was to obtain some useful measurements of changes in consumer meat purchases as- sociated with changes in retail prices. In more traditional terminology the objectives were to obtain empirical measure- ments of the price and cross elasticities of demand for dif- ferent kinds of meat. Emphasis was placed upon the analysis of demand for beef and pork, however, demands for sausage meats, poultry and fish were also studied. Some preliminary analyses of demand for retail cuts of beef and pork were made during the later stages of the investigation. A secondary objective of this study was to develop pro- cedures for analyzing panel data. Since this was one of the first demand studies based on information from the M.S.C. Consumer Panel much was to be learned about the peculiarities of processing this type of data. Due to the large number of available observations, extensive use of IhM equipment was necessary. This was followed by graphical examination of the data to determine the nature of the relationships as well as some of the disturbances present. Several single equation demand models were formulated and fitted mathematically using the traditional least squares regression procedures. Usefulness of Results Heat is one of the most important food items produced and consumed in this country. During 1953 farmers obtained 29 percent of their total cash receipts from the sale of meat animals.2 The processing and wholesale distribution of meat is the principal activity for some 1160 commercial meat packers.5 In addition, there are approximately 2000 small slaughterers and a large number of Specialized meat wholesalers who derive a major portion of their income from the handling of meats. Between the packer and the consumer there are 562,000 4 retail food stores in which the meat de- partment accounts for 25 to 50 percent of total store sales. Restaurants and institutions are also important users of meat. During 1953 consumers spent approximately 26 percent of their disposable income for food.5 Heat purchases made up approxi- mately one-fourth of the total food bill. 2 Agr. Mkt. Ser., U.S. Dept. of Agr., The Farm In- come Situation, March 1954, p.9. 5 Eur. of Agr. Econ., U.S.Dept. of Agr., The Live- stock and Heat Situation, September,1950, p. . 4 The Progressive Grocer, March, 1954, p.46. 5 Agr. hkt. Ser., U.S. Dept. of Agr., The Karketinc and Transportation Situation, February, 1951, p.40. The pricing and merchandiSing of meat is a complex pro- cedure. heat supplies fluctuate from.week to week, seasonally and cyclically. A large proportion of the meat is sold as "fresh" meat. Due to its perishability,it is extremely im- portant that meat prices, at all stages of distribution, are adjusted to facilitate the smooth and rapid flow of the product into the hands of consumers. Merchandisers must also consider changes in consumer demand due to weather, holidays or shifts in purchasing power. On the less perishable meat items merchandisers must also make decisions with regard to storage policy. It is hOped that this study will provide information that will be useful to the meat trade in their pricing and merchandising operations. A comprehensive knowledge of the demand characteristics for different kinds and cuts of meat appears to be essential if merchandisers are to price effi- ciently and profitably. Some merchandisers may gain suffi- cient knowledge through experience to do an effective pricing job. However, it is believed that there are many others who have an inadequate knowledge of the basic demand relationships and therefore they must depend upon crude rules of thumb or over-simplified tables in setting their prices.6 6 Meat Kerchandising, Inc., Master Meat Pricer, 105 South Ninth Street, St.Louis 2, hissourf. ‘1a4a Information on price and cross elasticities of demand for meats should be of use to those charged with public re- sponsibility for the formulation and administration of agri- cultural programs. Reliable elasticities are relevant to considerations such as how best to carry on a government purchase program to support the price of beef or pork. Ques- tions may also arise with regard to appraising the effects of policies which would encourage or discourage livestock production. Policy decisions with regard to import and ex— port restrictions on meats might also be affected by informa- tion on elasticities of demand. Trade groups organized to promote the sale of meat may find the results of this study useful in planning their pro- grams. During 1955, the National Livestock and heat Board spent over $500,000 in promoting the sale of meat. Funds were provided by assessments paid by farmers and marketing agencies. Promotional programs of a similar nature are now being initiated in individual states. heasurements of con- sumer demand for meats as an aggregate, as well as for dif- ferent kinds and cuts of meats, should be relevant in deciding what meat items to promote and also in appraising the effec- tiveness of the promotion programs. CHAPTER II THEORY AND hBASUREJENT OF DELAND FOR INTERRELATED COMIODITIES Introduction Most empirical investigations are guided by a body of theoretical concepts which influence the researcher to se- lect certain hypotheses for testing. In this study the theory of consumer demand for interrelated commodities appeared to be relevant. Unfortunately, many of the demand concepts are expressed in terms of "marginal utilities" and "marginal rates of substitution." These concepts provide powerful tools for a subjective analysis of demand, but their empiri- l cal measurement has proven to be most difficult. Perhaps an even more serious criticism of existing demand theories is the inadequate development of concepts which explain consumer behavior under non-static and imperfectly competitive condi- tions. 1 For a recent statement on the problem of measuring demand, see the article by Irving korrissett, "Some Recent Uses of the Elasticity of Substitution--A Survey," Econometrica, 21:41-62, January 1953. See also Frederick hosteller and Philip Nogee, "An Experimental Measurement of Utility," Journal of Pol. Econ., 59:571-404, 1951 3 Stephen W. Rousseas and A. G. Hart, "Experimental Veri- fication of a Composite Indifference Map," Journal of Pol. Econ., 59:288-318, 1951. This chapter is divided into two parts. The first is a brief statement of some of the theoretical demand concepts relevant to this study. Little or no attempt was made to develop new theories since the primary purpose of this dis- sertation was that of measurement. The second part of the chapter will deal with some of the measurement problems. Basic Concepts of Demand Theory Underlying assumptions. Most of our demand theories .2 'have been developed within a framework of a perfectly com- petitive, static system. The principal assumptions of this system, which are most directly related to a study of demand, are as follows: 1) Individuals possess perfect knowledge. 2) Preference patterns are fixed. 5) Commodities are perfectly homogeneous. 4) Individuals are motivated to maximize their satisfactions within the limitations of their real incomes (rational behavior). (5) The distribution of real income is fixed. (6) No individual seller is large enough to appre- ciably affect the price of a commodity. (7) Population is fixed. (8) Technology is fixed. (9) Government and other institutions are fixed so as to permit individuals freedom of choice. Definition of demand. Before progressing further with a statement of demand theory, the term "demand" should be de- fined. As used in this study, demand will be considered to be a schedule of the quantities of a commodity that an indivi- 10 dual (or group of individuals) is willilg to buy at all pos- sible prices, other things remaining the same.2 This definition of demand is essentially the one devel- oped by harsha11.5 Considerable debate has taken place over the interpretation of the "other things remaining the same" clause in the definition.4 This is often referred to as the ceterus paribus condition. In this study the "other things" considered to be held constant are as follows: (1) tastes and preferences of the group of purchasers considered; (2) their real income; (5) the price of every other "related" commodity. ' Law of demand. Traditional demand theory usually begins with individual demand and proceeds, through an aggregation process, to market demand.‘ The inverse relationship between price and quantity, which is typical of most individual demand schedules, is rationalized in terms of diminishing marginal utility for additional units of a commodity. Iarshall formal- ized this relationship in his classic "law" of demand which states, “the greater the amount to be sold, the smaller the 2 For alternative definitions of demand see Victor E. Smith, "The Classicists Use of Demand," Jour. of Pol.Econ., 59:242-57, 1951. 5 Alfred Marshall, Principles of Economics, 8th ed. Macmillan, London, p.190 and p.109. 4 Milton Friedman, "The Iarshallian Demand Curve," Jour. of Pol. Econ., 57:465-495, 1949. 11 price at which it is offered in order that it may find pur- chasers" or, in other words, the amount demanded increases with a fall in price and diminishes with a rise in price.5 Exceptions to this law have been recognized in the cases of "inferior" goods and prestige items. Equilibrium conditions. It is assumed that the indivi- duals can maximize their total satisfactions by adjusting their holdings of consumer goods until the ratio of the mar- ginal utility to the price for each good is equal to similar ratios for all other commodities.6 For the individual purchaser, prices are assumed as fixed. A market demand schedule represents an aggregation of the demand schedules for all individuals in the market. A market equilibrium exists when all individuals have adjusted their holdings of commodities so as to fulfill the equilibrium condition stated above and when prices have adjusted so that the sum of the quantities, which all individuals wish to hold, is just equal to total stocks. In simpler terminology, sup- ply is equal to demand.7 5 harshall, op.cit., p.96. fl 6 Knut Wicksell, Lectures on Political Economy, Routledge & Kegan Paul Ltd., London, 1934, pp.47:48. 7 Ibid., p.53 l2 ‘ Demand for relatengooas. Classical demand theory has been extended by Pareto, Hicks and others to provide a use- ful explanation of demand for related commodities.8 A system of indifference curves was used as a geometric illustration of the theoretical relationships (Figure I). Figure I shows the general case with an indifference map for two related commodities, X and Y. Each indifference curve shows all the combinations of X and Y to which the individual is indifferent. Startilg at the origin, 0, moving upward and to the right, each indifference curve represents a higher level of total satisfaction. According to Pareto it is not essential to be able to attach a cardinal measurement to each curve.9 It is sufficient merely to know the one curve repre- sents a higher or lower level of total satisfaction as com- pared to another curve. In arriving at an equilibrium position the individual adjusts his holdingsof X and Y until he reaches the highest indifference curve attainable with his limited income. As- sume that Y is money and an individual has 00 units. The market rate of exchange (price) is such that CO units of money are equal in value to OI units of X. An equilibrium is reached by exchanging AC units of money for CG units of commodity X. 8 J.R.Hicks, Value and Capital, 2nd ed. Oxford Univ. Press, London, 1946, Part I. 9 Marshall's development of individual demand implies measureability of utility. 13 Fig. Ib Fig. 10 Figure I. Indifference maps for related commodities. 14 Point R represents the highest level of satisfaction attain- able from income 00. Any other combination of X and Y will place the individual on a lower indifference curve. If his income should rise to OD, the new equilibrium position is represented by point S. Pareto distinguished between two limiting cases of re- lated commodities. Figure Ib represents the indifference map for two commodities which are perfect substitutes for one another. In this case the exchange ratio between the two commodities must remain equal to the slope of the indifference curve. Any deviation in the exchange ratio will result in a complete shift to the purchase of the cheaper commodity. Figure Ic illustrates an indifference map for two com- modities which are perfect complements. Shifts in the ratio of exchange between X and Y will not change the relative quan- tities purchased. This means that a rise in the price of X relative to Y will not alter the proportions in which the two commodities are purchased. Figure Ia illustrates a more general case with two com- modities which are related but not as perfect substitutes or complements. According to Pareto's definition, Y is competitive with X (or is a substitute for X) if an increase in the supply of X (Y constant) lowers the marginal utility of Y. Y is a com- plementary with X if an increase in the supply of X (Y constant) 15 raises the marginal utility of Y.10 Hicks .objected strongly to Pareto's definitions of complementarity and substitutability on the grounds that they have no preciseness unless utility can be measured in cardinal terms. Hicks suggests that one way to avoid this difficulty is to abandon the marginal utility concept and replace it with a new concept, "marginal rate of substitution," hereinafter referred to as KRS. By definition the KRS of X for Y is the quantity of Y which will just compensate the con- sumer for the loss of a marginal unit of X.11 Geometrically the MRS is represented by the slope of the indifference curve. At the equilibrium point B, in Figure Ia, the slope of the in- difference curve is equal to the slope of the price line, CI. This equilibrium is also described by the condition that the LES of Y for X is equal to the ratio of the price of X to the price of Y. hicks' definitions of substitutability and complementarity are intuitively quite precise. The definitions are as follows?2 "Y is a substitute for X if the HRS of Y for money is diminished when X is substituted for money in such a way as to leave the consumer no better off than before." 10 Ibid. p.43 11 Ibid. p.20 12 Ibid. p.44 16 "Y is complementary with X if the LES of Y for money is increased when X is substituted for money in such a way as to leave the consumer no better off than before.“ These definitions are quite similar to those stated by Pareto with one major exception. Hicks specifies that the consumer be "left no better off than before." While intuitively clear this condition has rendered the definitions empirically un- workable. Income effect. Perhaps one of Hicks' more useful contri- butions was his careful separation of the effects of a com- modity price change into an "income effect" and a "substitu- tion effect."13 If the price of a commodity X falls in relation to Y, the adjustment to the new equilibrium occurs as shown in Figure II. The original equilibrium was at point R. When the price of X declines, as shown by the new price line AI, the new equilibrium is at point T. The adjustment to this equilibrium was the combination of movement along an income- consumption path from R to S and then down the indifference curve to point T. The first portion of this adjustment was the effect of a change in income brought about by the reduc- tion in the price of X. The substitution effect was the , . l4 nwvement down the indifference curve from S to T. For __ 13 Ibid. p.29. 14 Marshall's demand function represents only the substitution effects of price changes. The income ef- fects are omitted by assuming a constant marginal utility for money. Harshall argues that for most consumer goods the income effect is too small to be of importance. 17 Figure II. Income and substitution effects of a price change for related commotidies. 18 inferior goods the income effect will be in the opposite di- rection from the substitution effect. For superior goods the income effect reinforces the substitution effect. This differentiation of the effects of a price change appears to be relevant to a study of demand for meats where certain items apparently are inferior goods for a large seg- ment of population. Assuming substitution is the dominant relationship between different kinds and cuts of meat, what is the effect of a fall in the price of one item, e.g. beef, on the demand for a competitive item, pork? If the two items are mildly substitutable a fall in the price of beef would have a verv slight effect on the demand for pork since the income and substitution effects tend to cancel out. If pork were an inferior good, demand would be likely to contract. If pork and beef were highly substitutable, a decline in the price of beef would depress the demand for pork due to the dominance of the substitution effect over the income effect. Recent theoretical concepts. Thus far the demand theories discussed have been limited to those formulated under the assumptions of a perfectly competitive, static system. These theories provide a useful framework for some mpirical demand studies. Nevertheless, it is recognized that the underlying assumptions do not adequately represent reality. There have been several attempts to develop new 19 theories based on more realistic assumptions. These contri- butions have not been well integrated. Consequently, only a brief recognition of some of the principal ideas will be pre- sented in this dissertation. Norris has develOped one of the more significant con- tributions in attempting to present a theory of consumer's demand based on conditions of imperfect competition.15 Product differentiation and non-price competition are recog- nized as conditions essential to a realistic explanation of the activities in the consumer market. It is assumed that most goods are presented to the consumer in clusters of com- peting substitutes due to the existence of brands and grades. The dominant role of the seller in influencing shifts in de- mand is also pointed out. An argument is advanced that the process of comparing prices and weighing them against expec- ted marginal satisfactions is a disagreeable process. Con- sequently consumers may give little consideration to the purchase of "petty goods" which are inexpensive in relation to total consumption expenditures. Norris cautioned against the view that the consumer at any time actually brings all of his consumption pattern into any kind of equilibrium. 15 Ruby Turner Norris, The Theory of Consumer's De- mand, 2d ed. Yale Univ. Press, New Haven, 1952. {\3 0 Some recent publications by Katonalo, Bilkeyl7 and others suggest that the psychologists have much to contri- bute toward a more realistic and comprehensive understanding of consumer buying behavior. This approach to demand analy- sis recognizes the dynamic nature of the decision making process involved in consumer purchasing. The underlyilg motives and attitudes of consumers are studied and related to buying behavior. Changes in behavior are explained in terms of the learning process through which consumers ac- quire new attitudes and motives. In this study of demand for meats traditional theories have been used to provide the basic framework for analysis. Hevertheless, consideration of some of the newer demand con- cepts have influenced the choice of relevant variables and the interpretation of statistical results. Measurement of Demand '5‘ General. hmpirical studies of demand have centered around the estimation of functional relationships between 16 George Katona, Psychological Analysis of Economic Behavior, thraw-Hill, New‘Ybrk, 1951 17 Warren Bilkey, "The Vector Hypothesis of Consumer Behavior," Jour. of th., 16: 137-151, 1951. prices, incomes and quantities purchased of different com- modities. Due to limitations of available data and the inadequacies of statistical procedures the accuracy of some of the estimated demand parameters have been subject to criticism. As improved data and statistical procedures are developed more reliable measurements of demand can be ex- pected. In this section the various elasticity concepts will be defined and th methods of measurements will be discussed briefly as a background for the empirical work which makes up the main body of this dissertation. Price elasticity. Price elasticity of demand is a term used to express the functional relationship between the prices and quantities purchased of a given commodity. "The elasticity (or responsiveness) of demand in a market is great or small according as the amount demanded increases much or little for a given fall in price, and diminishes much or little for a given rise in price."18 The mathematical definition 01 price elasticity of de- mand is as follows: E q E : : dq . P dp at a P where p is the price and q is the quantity of the studied commodity. Price elasticity is a measurement of the per- 18 Marshall, op.cit., pp.102-105. 22 centage change in quantity purchased associated with a one percent change in price. Price elasticity will ordinarily be negative for most commodities due to the inverse relationship between prices and quantities purchased. A useful classification of price elasticities is based on the changes in total revenue as the price of a commodity moves up or down (Table l). Table 1. PRICE ELASTICITY AS RELATE TO TOTAL REVEKUEIQ W W Elasticity of Demand Effect on Total Revenue Price Rise Price Decline Inelastic, < 1 TR rises TR declines Unit elasticity, 1‘ TR unchanged TR unchanged Elastic, ) 1 TR declines TR rises Cross elasticity. Cross elasticity of demand measures the percentage change in quantity purchased of one commodity associated with a one percent change in the price of a second commodity. When studying the demand for interrelated items, such as different kinds of meats, his demand measurement be- comes important. The mathematical formula for a cross elas- A 19 George J. Stigler, The Theory of Price, rev.ed., Macmillan, New York, 1952, p.37. 25 ticity is as follows: dql q E: ____;___ gdq1°§g dpg de ql p2 In this equation ql is the quantity of the first commodity and p2 is the price of a related commodity. When two goods are competitive the cross elasticity will be positive. Con- versely, if the goods are complementary the cross elasticity will be negative. In this study the regression coefficients will indicate the relationships between different kinds of meat. Income elasticities. The term "income elasticities" refers to two separate groups of empirical estimates. The first group of elasticities is derived from market data with observations extending over a period of time. Based on this type of data there are three different kinds of income elas- ticity. Income-quantity elasticity is a measure of the percen- tage change in quantities purchased associated with a one percent change in income. mathematically this is equivalent to this expression-- pj ll ”L3“ F1Q H u as 0 QIH 24 where q is the quantity purchased and I is a measure of income. Income-expenditure elasticity is a measure of the per— centage change in expenditures for a commodity associated with a one percent change in income. Expressed mathematically this becomes: tI‘ gl _. :4 ‘J d? .4 O I Hle H Elk! where E is the expenditure for the commodity and I is income. Income-price elasticity can be defined as the percentage change in the price of a commodity associated with a one per- cent change in income. [11 I I *6 If; II Q.» "d o Hle H p: H Due to differences in procedures used in adjusting time series data there are wide variations in estimates of income- elasticities. host of the difficulty centers around the pro- cedures used for deflating price and income data so as to differentiate between the effect of changes in “real income“ as compared to "money income.n20 20 Elmer J. Working, "Appraising the Demand for Agri- cultural Output During Rearmament," Jour.Parm Econ., Vol.54, 1952, p.215. 25 A second group of income elasticities have been derived from cross-sectional data. The definitions of elasticities stated above are adaptable to these data. However, he inter- pretation of the results is somewhat different. When using cross-sectional data the income-quantity and income-expendi- ture elasticities represent the differences in purchasing patterns associated with different levels of family income measured at a point in time. Due to difficulties in measur- ing the "net" relationships between income and purchases of food items, attempts to reconcile income elasticities based on cross-sectional data with those derived from time series have been relatively unsuccessful.21 Apparently there are many interrelated factors that affect differences in family food purchases, with income being only one of them. Income-price elasticity is relatively unimportant in cross-sectional analysis since it represents differences in the "quality" of commodities purchased by families with dif- ferent income levels. Regression analysis. The various methods for measuring demand elasticities range from the computation of simple arc elasticities to the fitting of highly complex mathematical 21 Karl A. Fox, "Factors Affecting Farm Income, Farm Prices and Food Consumption." Agr. Econ. Res., Vol. 3, 1951. pp.79-81. 26 models. Probably the most widely used procedure has been traditional least squares regression. In recent years con- siderable controversy has arisen over the applicability of 22 the single equation methods of estimating demand parameters. A system of equations approach is being developed to handle some of the estimation problems which are not adaptable to single equation methods. Disagreements still exist, however, with regard to the kinds cf problems that can be handled satisfactorily with single equation methods. A complete analysis of this question is beyond the scope of this dis- sertation.25 A combination of circumstances made it desirable to use the more flexible and less expensive single equation regres- 24 sion procedures in this study. One reason was that the 22 h.A.Girshick and Trygve Haavelmo, "Statistical Analysis of the Demand for Food: Examples of Simultane- ous nstimation of Structural Equations," Econometrica, 15:79-110, 1947. 23 For consideration of this problem see Richard J. Foote and Karl A. Fox, Analytical Tools for heasuring Demand, Agr.1kt.Ser., U.S.Lept.A¢r., Agr.handbodk £0.64, I954. See also, Herman Hold and Lars Jureen, Demand Analysis, John Riley, New York, 1955, Chap.II. 24 In an unpublished Ph.D. thesis entitled, An Econo- metric Analysis of Demand for Eggs, Iowa State College, 1952, George G. Judge, concludes as follows: "Computations with the simultaneous equation method are quite complex and time consuming. Unless the investigator possesses a thorough knowledge of the simultaneous equations proce- dure and has a large amount of resources available (both monetary and physical), he will probably find a more effi- cient use of research resources could be made with the alternative methods even though in some cases the accuracy of the results may be questionable. Resources spent on including more variables in a single equation may, in some instances, yield more information." (p.215). 27 large mass of data available from the h.S.C. Consumer Panel was being processed and analyzed for the first time. It was expected that errors might be uncovered during and after the equations had been fitted. Recomputing a complicated mathe- matical solution to a system of equations would have been ex- pensive and time consuming. Another reason for favoring the use of single equation models was that the relationships ex— A isting in the panel data and the peculiarities of handling weekly time series observations were not well known. It was reasoned that even if it were known to be desirable to use a system of equations procedure, such an analysis should be preceded by a rather thorough examination of the data using simpler methods. Some of the limitations of the single equation multiple regression methods of analysis should be recognized. The as- sumptions of this approach are as follows:25 (1) The observations of the explanatory variables are not subject to errors of definition or measurement. It follows that the unexplained part of the variance of the dependent variable is due either to errors in that variable or to the influence of omitted variables. 25 (a, -—. o o 1 o A. R. Prest, "some mxperiments in Demand AnalySis," Review of Econ. Statistics, 51:35-49. 1349. 28 (2) The residual errors are not autocorrelated, being drawn independently in each time period from a stable, normal population. (5) The equation does not form part of a set of simul- taneous equations or, at least if it does, the influence of changes in these equations can be neglected. 1 The extent to whicigthe equations and the data used in tnis study fulfilled the above assumptions can be commented on only briefly at this point. Hith regard to the first assumption, it is almost a certainty that some errors exist in the ex- planatory variables and, therefore, some of the parameters will be biased to some extent.26 The amount of such bias is not easily determined. However, if the errors are randomly distributed the regression coefficients should be unbiased. Tests for autocorrelation of residuals can be made. If the amount of autocorrelation is large there may be some advan- tage to using first-differences instead of actual observations.2 The third assumption is not likely to be completely fulfilled by any single equation designed to explain economic behavior. In this study there was reason to believe that simultaneous relationships could be neglected, at least during the early 25 Ibid. p.37 ‘l 27 R. L. Anderson, "The Problem of Autocorrelation in Regression Analysis," Jour.Amer.Stat.Assoc., 49:115-129, 1954. 29 stages of analysis. Evidence supporting this vieWpoint in- cludes an observation that for any given week the supply of meat in the Lansing retail markets can be considered as pre- determined. In addition, it is believed that most meat items are purchased on a current week—to-week basis and, therefore, storage demand or demand for non-food uses can be largely ignored. However, this does not preclude the possibility that simultaneous relationships may be uncovered as the analysis proceeds. If this occurs appropriate changes will be made in the procedures used for analysis. Additional consideration of some of the other problems of single equation regression analysis will be taken up in Chapter V. CH AFTER III PREVIOUS INVESTIGATIONS OP DELAJD FOR KEATS Based on “a arliet Data During the 1920's there were attempts to apply regres- sion analysis to the problem of forecasting livestock prices.1 Although these efforts were not designed primarily to measure demand elasticities, they represent some of the first appli- cations of re dression analysis to price-quantity data for livestock products. Probably the first concerted effort to obtain empirical measures of demand elasticities for meats was included in Henry Schultz's "The Theory and heasurement of Demand."2 In this monumental treatise Schultz discussed the theory of related demands and tested some of his hypothe- ses with examples of demand equations for beef, pork and mut- ton. Using a single equation multiple regression technique, the average annual per capita consumption of beef, pork, and mutton were expressed as functions of their retail prices and 1 G. C. Haas and Iordecai Ezekiel, Factors Affecting the Price of Regs, U.S.Dept. of Agr., Bul.1446, 1926. Lordecai Ezekiel, "A Statistical Examination of Factors Relating to Lamb Prices," Jour. Pol. Bcon., 35:233-60, 1927. 2 Henry Schultz, The Theory and Eeasurewent of Demand, University of Chicago Press, 1938, p.641. 31 per capita income. Equations (a), (b), and (c) summarize some of the results obtained.:5 (.07) (.11) (.21) (.21) (b) 0p : 1.19 + .19 Pb - .70 P - .004?m + .54 I (.05) (.05) (.10) (.10) m ,(.07) (.12) (.22) ‘ (.22) The price, cross, and income elasticities can be read directly from the equation since the data were fitted in log- arithmic form. The figures in parentheses are the standard errors of the regression coefficients. The price elasticity for beef was -.86 as compared to -.70 for pork and -l.80 for mutton. The cross elasticity between the price of pork and the quantity of beef was non-significant while the price of beef apparently had a significant effect on pork consumption. The consumption of beef and pork shows little relationship to the price of mutton. The income-consumption elasticities for beef and pork are both close to ¢.50. In appraising these results it is essential to understand the nature of the data used. Annual average observations were used for the period 1922-1955. The quantity data represent carcass weights with beef and veal being combined. Lamb and mutton were likewise 5 Ibid., p.659. combined. Prices used were Bureau of Labor Statistics average annual retail prices. These prices were deflated by the B.L.S. cost of living index as was the income variable. The quantity and income data were both placed on a per capita basis. In 1955 Shepherd4 made some estimates of the price elas- ticity of demand for pork, based on annual time series data. Using multiple graphic regression with the retail price of pork as the dependent variable and consumption of pork and consumer income as explanatory variables, he tentatively con- cluded that the price elasticity for pork was about -l.O for the period 1921-54. Using alternative measures of consumer income in the same type of analysis gave price elasticities of -.55 and -.72. These latter elasticities were considered to be less reliable than the -l.O due to the differences in the data used to measure income. here recently Shepherd has made an analysis of changes in demand for meat and dairy products.5 In his analysis of change in demand for meat he used a multiple correlation of four variables: (1) U.S. average retail price of meat in cents per pound (2) Per capita disposable income (index) (3) Per capita consumption of meat (pounds of red meat) ( ) T ime 4 Geoffrey Shepherd, "The Incidence of the Processing Tax on Hogs," Jour.Farm Econ., 17:521-59, 1955. 5 Geoffrey Shepherd, Changes in Demand For Ieat and Dairy_Products in the United States since 1910, Iowa Agr. Exp. Sta., Ames, Iowa, Res. Bul. 568, 1949, p.381. 55 The data are annual observations for the period 1920-41. A single equation least squares fit of the relationship, with retail meat prices as the dependent variable, gave an R of .97 with all of the regression coefficients being significant at the one percent level. The price elasticity of demand turned out to be -.75. Shepherd concluded that this "appears reasonable in comparison with the elasticity of demand for pork, which is slightly higher than unity. One would expect the demand for meat to be less elastic than the demand for any one meat."6 The income elasticity turned out to be 0.75. Evidence indicated that the demand for meat had declined slightly in relation to income between 1910 and 1346, but only as part of the general decline in expenditures for food as a whole. 7 has In a detailed analysis of demand for meat, Working obtained several measurements of price and income elasticities. Using annual data for the years 1922-41 he has made some 22 different analyses using a single equation least squares method. One of the basic equations included these variables expressed . . , 8 in lOgaritnm8: 5 Ibid., p.587. 7 Elmer J. Working, "Studies in the Eeasurement of Demand With Special Reference to the Demand for heat. Unpublished Ph.D. thesis, Harvard University, 1952. 8 Ibid., p.115 '2 b) 1) Average retail price of meat (2) heat consumption per capita (5) Deflated disposable personal income per capita (4) Consumers price index. Four regression equations were fitted to the data using each variable alternately as the dependent variable. Eorking then takes the geometric mean of the four regression coefficients for each pair of variables as the best estimate of the true structural parameters of the relations betW'en the variables. In the analysis mentioned above the weighted regression co- 1 efficients indicated that a change of one percent in consump- tion is associated with a 1.45 percent change in retail price in the opposite direction and that a one percent change in "real" income is associated with a .75 change in price, also in the same direction. From this and several other analyses using different forms of soue of the same basic data Korhing concludes that, "on the basis of the correlations reviewed it would seem most likely that the coefficient of price flexi- bility (the reciprocal of the coefficients of the elasticity of demand) is somewhere in the vicinity of-l.55 and -l.50"9 The corresponding price elasticity range would be -.67 to ‘0740 9 Ibid., p.125. 55 " of the In an attempt to explore some of the "dynamics demand for meat, Working made a regression analysis which in- cluded two consumption variables: the five-year average, and the current year as a percent of the five-year average. His conclusion was that, "in the long run the demand for meat is less inelastic than in the short run."10 Working also cites evidence from his analyses that the demand for meat changes with changes in the general price level quite apart from the effect of the change in real income. This would partly explain the increase in demand for meat in the post World War II period when the general price level was rising rapidly. Fox used the single equation method and annual data for the period 1922-41 to analyze the demand for a number of farm commodities, including meats.ll Linear relationships were assumed and data were expressed as first-differences of logarithms. When per capita consumption of all meat was ex- pressed as a function of the retail price of meat and diSposable income per person, a price elasticity of -.64 and an income elasticity of .56 were obtained. Price elasticities for both beef and pork were reported to be about -.8 based on a similar lO Elmer J. Working, "Agricultural Demand During Rearmament," Jour. Farm. boon., 54:218, 1952. 11 Karl A. Fox, "Factors Affecting Farm Income, Farm Prices, and Food Consumption," Agr. Econ. Res., 5:65-111, 1951. 36 analysis. Income elasticities for beef and pork were about .7. Fox also found a competitive relationship between beef and other red meats. His analysis indicated that a ten per- cent increase in the supply of other red meats depressed the average annual price of beef by 5.2 percent.12 French13 and Viahby14 have applied some of the more recent statistical techniques to the problem of measuring demand for meats. Results were compared with those obtained using the more traditional single equation least squares methods. French used a system of nine equations to explain the relationships in the market for meat. (heat included all red meats, poultry and fish). The demand equation was solved by he maximum likelihood method and then compared to results obtained by ordinary least squares regression. The variables in the demand equation were annual observations for the U.S. for the period 1919-41. Y1 - per capita consumption of meat Y2 : retail price of meat Y5 = price of other food 12 Karl A. Fox, The Analysis of Demand for Farm Products, U.S. Dept. of Agr., Tech. Sal. 1081, 1955, p.45. 15 Burton L. French, Application of Simultaneous Equa- tions to the Analysis of the Demand for n8&t9 Unpublished M.S. thesis, Iowa State College, 1949. 14 Omar hahby, Econometric Analysis of the Demands for Beef, Pork and Poultry Products, Unpublished Ph.D. thesis Iowa State College, 1951. 57 K1 price of non-food 3 4 Y5 a disposable income 2 = time The results obtained by the two methods are summarized in the following equations: (1) Limited information maximum likelihood method Y1 I '0045 Y2 " 1.14 Y3 - 2.05 Y4 + 0017 Y5 " 1.17 Z1 + 450 (2) Ordinary regression minimizing on quantity. Y1 = ‘0081 Y 0.78 Z " 1.14 Y - 2.05 Y * 0.18 Y '- 2 5 + 504 5 4 l (3) Ordinary regression minimizing on price and normal- izing on quantity. Y " 1.31 Y Q 0.19 Y '- = -1027 Y2 " 0.58 Y 5 l 0.44 Z1 + 580. 3 4 The signs associated with the coefficients were consistent for all equations, but the magnitudes of the coefficients are quite different, particularly with regard to the relationship between price and quantity of meat. The price and income elasticities computed at the mean values of the variables were: Equation Price Elasticity Income Elasticity (l) ‘0024 .50 (2) “004:5 .55 (5) '0071 .58 Wahby set up a more detailed model of the meat market with separate demand equations for beef, pork and poultry 15 The complete model included 12 stochastic equa- products 0 tions with the relationships assumed to be linear in logarithms. The variables included in the demand relationships are annual observations for the U.S., 1921-41. K: I 0" Ib ()3 l--' O3 0'! tb CR (0 CD <1 H II quantity of pork consumed per capita 1 m quantity of beef consumed per capita i4 quantity of poultry consumed per capita retail price of pork retail price of beef 2 retail price of poultry products retail price of dairy products +4 K H1 *4 l< retail price of oleomargarine N a time retail price of other food disposable income per capita NNN moving average of Z4 for the preceding 5 years As originally set up, the equations included were over-identi- fied. Consequently, the limited information method of estima- tion was followed in solving for the relationships. The re- sults obtained were unsatisfactory when measured against a priori knowledge. Therefore, the model was altered slightly 15 Ibid., p.14. J. um so as to make possible the use of reduced form solutions. The results are shown below: (1) Pork equation Y = "0091 Y 1 - 0.05 2 r'.‘ , _, '7 _ C 2 + 0.00 Y3 + 0.07 Y4 1.20 Y5 0.Jl Y6 l + 0.16 25 + 0.76 24 + 0.29 Z5 + 2.70 (2) Beef equation Y7 = '0077 Y3 4 0.52 Y2 ‘- 0.67 Y4 "’ 0.22 Y5 - 1.09 Y6 - 0.02 z . 0.29 z + 0.65 z 1 5 4 - 0012 (.1 ‘9 0006 5 (5) Poultry products equation - _ O . Y8 - 0.68 Y4 4 .12 Y2 4 0.28 Y3 0 0.22 Y5 + 0.31 Y6 + 0.001 21 + 0.56 25 + 0.55 Z4 + 0.28 Z5 - 0.42 The elasticities can be read directly from the equations since the variables were expressed in logarithms. In a recent study of demand for meat in Canada, the price elasticity of demand at the wholesale level was estimated to be -065. 16 Annual data summarized on a September-August market- ing year basis was used for the period 1926 to 1942. The variables used in the single equation regression analysis were as follows: X weighted average wholesale price of all red meat 1 divided by the general wholesale price index. X2 average domestic consumption per capita for all meat. lb F.K.Schrader, The Demand for Feat in Canada, Econo- mics Division, Canadian Department of Agriculture, Ottawa, 1955. 40 X3 - the index of inrustrial production per capita reflecting consumers' ability to purchase goods and services. The regression equation was: - 1.9453 - 1.5500 103 x 1 _ + 1.0118 103 X 103 X 2 5 The price elasticity is the reciprocal of -1.5390 or -.65. Similarly, the income elasticity is estimated to be .99. Shrader estimates that the price elasticity of demand for meat at the retail level is -l.05 and about -.40 at the farm level. Studies made by Jureen indicate that price and income elasticities of demand for meats were lower in Sweden than in the U.S. Lultiple regression analyses of annual data for the period 1921-1959 were used in deriving the elasticities (Table 2). Table 2 ELASTICITY cw DELAHD FCR HEAT, SWEDEN, 1921-1959“ Price Elasticity ._ find of Heat; . - p . Income Elasticity Separate Price Preportional Changes Price Changes All meat --- .28 .28 Beef .50 .22 .50 Pork .45 .31 .53 % Herman 2010 and’Lars Jureen, Demand Analysis. John Kiley, flew York, 1955 p.282. 41 The quantity data were estimated consumption per person for the entire country. The prices were takes at the retail level. Another approach, which differs from the analyses dis- cussed above, has been used to examine the competitive rela- tionship between different kinds of meats. Using this method the price ratios for two different items is related to the quantity ratios existing in different time periods. Clawson 17 In one made several graphical comparisons of this type. chart, he plotted the cattle/hog price ratio against the beef/pork consumption ratio for the years 1899 to 1939. By inspection he draws two lines of relationship between the two ratios: (A) for the years 1899-1914 and 1954-1959; (5) for 1915-1955. Both lines slope downward to the right and are slightly convex to the origin. Little attention was given to an explanation as to why there are two different lines of relationship. The correlation between the price and quantity ratios was .87 and .97 for the two groups of years. The elasticity of the lines, measured roughly at dif- ferent points on the curve, varies from -.55 on the A curve to well over -l.0 on the B curve. .18 . . . Szatrowski used a multiple regre381on analy31s which 17 Marion Clawson, "Demand Interrelations for Selected Agricultural Products," Quar.Jour.Econ., 57(2):265-502, 1945. 18 Zenon Szatrowski, "Time Series Correlated with the Beef-Pork Consumption Ratio," Econometrica, 16:60-78, 1945. 42 expressed the annual beef-pork consumption ratio as a func- tion of the following variables: X1 = time X2 = beef-pork price ratio (actually cattle-hog price ratio) X5 = consumption ratio of preceding year X4 = cattle-hog population ratio corn yield of previous year in bushels per acre X6 income in dollars per capita Results indicate a negative relationship between the price and quantity ratios with a regression coefficient of about -.5. Both Clawson's and Szatrowski's price ratios are based on live animal prices and therefore do not represent the price to which consumers react. Variations in by-product values on cattle, lard values on hogs and marketing costs cause the re- lationship between live animal and retail prices to change. Woodlam19 has also made a regression analysis of the price and quantity ratios for beef and pork using Canadian data for the period 1928- 951 omitting the war years. He used the wholesale price of carcass beef on the Toronto market and estimates of consumption based on production data that were adjusted for exports, imports and changes in storage stocks. 19 . n . . . . G. E. heedlam, Tne Influence of Prices on the Relative Consumption of'Beef and Pork, Aconomics Div., Canadian Dept. of Agr., 1953. In his analysis he finds that -- low X1 = 4.075 - 1.029 log X2 where X1 is the consumption of beef expressed as a percentage of the consumption of pork and X2 is ratio of prices in terms of percent. The interpretation of the results of these studies based on price and quantity ratios is not clear. Presumably the high correlation between the two ratios and the signifi- cant regression relationship is offered as evidence that beef and pork are competitive items. This may be subject to ques- tion since it can be demonstrated that the price and quantity ratios are highly correlated for two independent goods with price elasticities of approximately -l.0. The slepe of the regression function between the price and quantity ratios might provide some evidence on the competitive relationship between two goods providing the price elasticities were known in advance. Cross-Sectional Budget Studies Several cross-sectional surveys have been conducted to obtain information on the relationship between certain socio— economic factors and the demand for meat. In 1954, a survey 9 o to g 20 of 2200 families was made in Linneapolis. One of the prin- 20 W. .Waite and R.Y.Cox, A Study of the_Consumption of heats in Kinneapolis, 1954, Linn. Asp. pxp, Sta., Bul. 521, 195”. U 44 cipal conclusions was that income exercises a predominant 1 influence on meat consumption with the 1igher income fami- lies Spending nearly three t mes as much per person for meat as compared to families in the low income groups. The study also indicated that size and composition of families had only a minor influence upon per capita meat consumption. It should be noted that this survey was made at a time when unemployment was large and incomes generally depressed. During 1940, tne Bureau of human nutrition and Home Economics of he U.S. Department of Agriculture made a sur- vey of some 1600 urban families to study their food consump- Q Q ntion patterns varied between tion habits. Although consum, cities, the relationship between incone and per capita neat consumption is summarized in an illustration taken from the tudygl (Figure III). 03 publication reporting the results of the It can be seen that the quantity of beef increases with rising income, while pork consumption remains about constant. However, the consumption of particular cuts of meat show mixed patterns. For example, ground beef consumption increases as incomes in- crease up to $4,000 per family. Beyond $4,000 ground beef consumption declines. A similar pattern exists for pork chops, bolo¢na and other cold meats. Salt pork is the only item that ZlBur. of Human Nutr. and Home Econ., U.S.Dept of Agr., heat Selections of City Families, Commodity Summary No.1, 1949. OONSUIP‘FION PER PERSON II A WEEK (POUNDS! 45 8 q .4 q d .1 ‘ [rIIT1IIIII‘ IIIIII I [111111111111 PORK l A \ALL PORK UIIIII 111 l l l 8 388388 ”011.1110. sumac. sour run PORK .0. 11111 1111111111411 11111 1111111111111 Lullllrl IIIIITIIITII IIIIIII I IITIIIIIIITIII '00 _ OTHER MEATS FISH, POULTRY, EGGS .oo :- 2 Z .70 '- —I I— .n h- -1 1-— .OO " - "' .40 _ IOLOONA, 071101.. d - .30 - _. .— .20 '- '— - VARIETY HEAT! - '- d '0 — — nuv’ P- *LAII m 1111 l llllll1lllll 111111 11111111111111 coo |.000 2.000 5,000 IODOO 500 I,000 2.000 5.000 I0,000 I947 INCOME AFTER FEDERAL INCOME TAXMOIION) SOURCE: Bur. of Human Nutr. and Home Econ., U.S.Dept. of Agr., Meat Selections of City Families, Commodity Summary No. 1, 1945, p.7. Figure III. Relationship between family income and meat consumption, urban housekeeping families, United States, spring, 1948. consistently behaved as an "inferior product." The study also showed that total meat consumption increased with in- come, but leveled off at family incomes of $4,000 and over. As incomes increased meat expenditures increased more rapid- ly than quantities consumed due to the fact that the higher income families purchased higher priced cuts of meats. A series of income-consumption elasticities were com- puted by Iaite and Trelosan from the 1948 survey data men- 1 tioned above. These are summarized in Table 3. Table 3 ITCW -CF1SEL‘31LJ noAS110191- TOR SELECTED IEAi Im‘IS thAU EinILIns, ITino STA1LS, SPRING, 1948 Kind and Cut of Keat Elasticity Pork chops -0.21 Ground beef -0.15 Bacon 0.10 Beef, boiling and stewing 0.36 Beefsteak, round 0.55 Poultry (total) 0.64 Smoked ham (cooked) 0.82 Beefsteak, other than round 1.04 J’ ' 1.0J1aite an H.C. Trelogan, Agricultural Karket Prices, John $41193], 2d 6d. 1951, p.41. The above elasticities are simple arc elasticities using he second income group, wi'h an average family income of $1,555, and the sixth income group, with an average income of 85,861. Average family consumption was the quantity variable. The existence of wide variations in meat purchases among fami- 0 n 1 o o 2 2 - 1 1 0 lies 01 tne same income grouping sun“ests that tne reliabil- QC) ity of these estimates of income elasticity may be rather low. A cross-sectional survey of 726 families was made in Syracuse, How York, in the Spring of 1948.25 These same fami- lies had been surveyed in 1942. During the intervening six- ' year period the income position of these families and their meat consumption had shifted considerably. One group of families with a slight decrease in income reported a decrease of 11 percent in per capita meat consumption. These families had below medium income in 1842. Another group of families with the largest increase in incomes between 1842 and 1948 (114) percent) reported an average increase of 65 percent in the quantity of meat purchased per capita. Families were classified on the basis of per capita income rather than family income as in the 1948 surveys made by the BHNHB. The Syracuse study also analyzed the shift in consumption of dif- ferent kinds of meats among the different families. Not only did the families with the greatest rise in income increase 22 n . v w Bur. of human sutr. and home Econ., U.S.Dept. of Agr., heat: Variations in Consumption and Interrelation- ships with Other Foods, Commodity Summary No.11,1951,p.5. 25 Will I. Simmons, Consumer Heat Purchases in Syracuse, New York, 1948 and Comparison with 1942, Cornell Univ.Agr. Exp. Sta., Ithaca, New York, Bul. 809, 1951. their meat purchases, but there was a noticeable shift to- ward the higher valued, "more desirable” cuts. Thus, the results are relatively consistent with the relationships found in the cross-sectional analysis of the iHHHE survey data discussed earlier. In the spring of 1950, a sample survey of 1385 families was made in Lansing, hichigan. Family characteristics and food purchases for a period of one week were obtained. Zioss24 analyzed these data to determine some of the relationships of socio-economic factors, such as family income and size of family, to per capita food consumption and expenditures. Cross-sectional tables were used and the difference between means tested by analysis of variance procedures. Family in- come and size of family appeared to be most important factors affecting per capita meat consumption (Table 4). Since these two factors exhibited a ositive relationship it was desirable Fr to test their relationship with per capita meat consumption while holding one or the other constant. When size of family was controlled, family income was not significantly related to per capita consumption of pork, lamb and mutton, poultry, fish, or seafood (Table 4). Certain irregularities can be noted. For example, consumption of red meats rises as incomes 24: 1 1" .r v 0 1 0 .. 0 Thomas h. moss, Some fielationsnips of Selected S0010- Economic Factors to Food Consuhption and Expenditures, Lansing, Spring, 1950. Unpublished Ph.D. thesis, hichigan State College, 1952. 49 meow -qe eHHoo epepc QeLHHs.e M2 ICOU @002. On. thHOGHH 02.2.;OGOr ;...,OHOC.O,_ CepooHo .HO mam ecoHHMHem ezom ammo; .m meioflb \M . mQSOfim I ewe aeQQSOb en» :H mebHaemzofl gasp mmeH bemeSOASb eae 903 maeeb on ae>o me>HBemsom \e . .9302: ezoodH dzH: 202 meCHHoev degp 6:2 I esopm efioocH EZHpeE Op 30H Eonm memeeaomH 4eeo 352023 Ho SCHmcizmsoo epHceo gem \b .teHeHeH MHHQeeHHHCG Hm mep.pHUCeChe epHneo Lem I .bepeHeh prQeOHHHcth um: mepSp-_tce cxe epHceo hem \9 .::H: was Sste: Ceeapep no: 259 .mgsoaw e:oocH 2 :HH tne 30H 3G2 SfiHdeE 3:2 30H See32en eoCepe.HHHs HCeOHHHGJHmI\e . pcepmcoo bHHfieb go eNHm A+v A+V pce eEooGH :HHB HHHQ on web on 0: cm on em . web @002 beeea Ho eNHw . I pnepmzoo bHHfiem Ho emHm tne eioosH :uHs on o: 02 on o: o: o: o: eMHsemsoa Ho GOHHeOStJ I. pmemeOQ bHHEeH \o \m go eNHm tne eEoenH oc web 0: on ea 0: 0: web 32H; .emHBemsoa Ho em¢ HIV \m. HIV HIV HIV usepmdoo eEoosH web on web on on web web web 32H; bHHEem Ho eNHm A+V \U A+V \w. usepmzoo bHHEeM Mo eNHm web web on oc on 0: web on SHHB ezoocH bHHEeb Cepuzm xeepm ween boomeem 2 QEeH £20m Heem muse; OHHmHeepoeaeso tweem bfisoaw 2 Qlo bpszom HH< HH< HHw pet HH< OHSocoomuoHoom COHHmfiswdoo eHHmee peg :pHs emHzmcoHpeHea Ho eoanHchmHm HomoH§ H22. .22Hc222 o OHzola2 equo Hose 21H H31.a 022:52212 2,HLD¢:nfivJ fimHadU fifib 7H mHUmHmHLBHQ PO HofifionHfime P 4144., NJ? .0114. Alfiflqgilg. 1.11.1 [)4‘ .I. ..I.1 ..1 Frukrrr ....\ I .i-EH E Q. I .. .. .,.rLr\.. .y F. ..m 7.. my w fldmdB 50 rise up to the medium income level and then levels off. Ground beef consumption increases from low to medium income group, but then declines from medium to high incom group. Age and education of the housewife had little influence on meat consumption after sorting by size of family and family income. As the size of the week's food bill increased, beef purchases rose significantly. It has been pointed out that fanily income and size of family were positively related and both were siynificantly related to the consumption of several different meats. For practical application it would be helpful to know more about the direction and magnitude of these relationships. Using data in Loss's thesis a series of simple arc elas- ticities were calculated. (See Table 5.) It was found that the income-consumption elasticities varied considerably by size of family and that elasticities are much higqer between the low and medium income groups than between the medium and high income group. In fact, a negative elasticity exists for ground beef when.comparing the medium and high income group. Ttese elasticities point up the importance of adjusting for size of family before calculating an income elasticity. In the elasticities computed by Waite and Trelogan (Table 5), the influence of size of family distorts the "true" relation- ship between fauily income and consumption. 51 TABLE 5 INCOME-QUANTITY ELASTICITIES FOR MEAT ITEMS BY SIZE OF FAMILY - LANSING, SPRING, 1950a b Kind of SiZe of Elasticity Between Income Groups Meat Family Low to Medium Medium to High All meat 1-2 .67 .06 5-4 .27 .04 5 or more .05 .12 Beef 1-2 .68 .20 5-4 .41 .16 5 or more .28 .06 Pork 1‘2 .48 ’006 5'4 .08 .OO 5 or more -.16 .15 Ground beef 1-2 .76 -.72 3-4 .40 -.10 5 or more .24 -.26 Beef steak 1-2 .81 .52 3-4 .52 .50 6 or more .80 .45 a Data taken from Moss's Ph.D. thesis. Formula used to calculate elasticities was . 2 _ 1 C11 4' qz I2 ’ II 11 ‘ I2 b Low income group includes families with incomes under $5,000; mean of group - $1,900. Medium income group includes families with incomes of $3,000 to $4,499; mean of group - $3,542. high income group includes families with incomes of $4,500 and over; mean of group - $6,074. 52 Q It is sometimes difficult to reconcile income elastici- ties derived from time series data with those derived from budget studies. In addition to the disturbing influence of 0 factors such as size of family, there is also a question as to how readily families take on the consuuption habits of a higher income group as their incomes rise relative to other families. This thesis will not develop these aspects of in- come elasticity, however, the empirical relationships pre- sented in this section will assist in the interpretation of the results of analyses to be presented in subsequent chap- ters. Summary host of the demand studies using market observations are based on annual a “regative data for the period between World Wars I and II. In most cases a single equation least squares technique has been used with minor variations in the procedures for adjusting the data. Therefore, it is not surprising that there should be rather close agreement in the results obtained from these analyses. The estimates of price elasticity of demand for all meats ranges from -.64 to -.75. The price elasticity estimates for beef and pork are less in- elastic than for all meat with a range from -.70 up to -1.0. Income elasticities range from about .50 up to .75, with the 55 lower estimate being associated more closely with real in- come where the price and income series have been deflated by a consumer price index. Where the income data were ex? pressed in current dollars the income elasticities approach the upper end of the range. The only major deviation in elasticities resulted from an attempt to use the limited information method of estima- tion. The price elasticity for all meat, using this method, turned out to be rather low as compared to a least squares fit of some of the same data. Some rather important questions have been raised as to the usefulness of elasticity estimates based on studies of inter-war period conditions as a basis for predicting econo- mic behavior in the postwar years. Kuznets‘g5 criticizes the assumption underlying these studies that certain "social variables" remain constant or change slowly and smoothly over time. Some of these factors which are usually assumed away under the ceteris paribus of accepted demand theory include (1) changes in the distribution of real disposable income, (2) lags in consumer adjustment to rapidly changing prices and income situations, (5) changes in eating habits and (4) changes in the composition and distribution of the population. 25G.M. Kuznets, "Measurement of Market Del nand with Particular Reference to Consumer Demand for Food. " Jour. Farm Econ., 55:878-895, 1955. 54 During the past thirty years there have been significant and in some cases rapid changes in some of these social variables. Between 1941 and 1944 there was a marked im- provement in the relative status of the lower income seg- ments of the population, particularly for families of two or more.26 This gain in income position was further accentu- ated by Sharp increases in income taxes which occurred during this same period. Available evidence indicated that changes in occupations, the trend toward urbanization, improvements in refrigeration, plus many other factors, have contributed to changes in demand for food items.27 It seems questionable that an elasticity representing the entire range of price, quantity, income variations during the interwar period can be eXpected to perform with a high.degree of predictive ac- curacy under the present economic and social environment. '5 priori reasoning suggests that price and income elastici- ties for different food items might be one thing during a period of depressed economic activity and quite another dur- 26 Council of Economic Advisors, The Economic Situa- tion at Midyear, 1951, A Report to the President, U.S. Gov't. Printing Office, pp. 90 and 96. 27 Bur. of Agr. Econ., U.S.Dept. of Agr., Agricul- tural Outlook Charts451954, p.243 Marguerite Burk, ‘Dhanges in Demand for Food from 1941 to 1950," Jour. Farm Econ., 55:281-298, 1951; Earl E. Miller, "Changes in Demand for Pork Products," The Livestock and Meat Situation, Bur. of Agr. Econ., U.S.Dept. of Agr., May-July, 1955, pp.14-l9. 55 ing a period of full employment and high incomes. Similarly, it might be argued that price elasticities are likely to be different when supplies are large as compared to when supplies are average or small. However, in many empirical demand studies a demand curve of constant elasticity has been fitted to the data. In some cases this may be justified and certainly it has some advantages of simplicity in computation. This sug- gests that the usefulness of elasticity estimates may be hampered by a tendency toward over-simplification. Instead of a single elasticity for a commodity there might well be whole family of elasticities. Detailed demand studies at ~both the micro and macro levels and for time periods of dif- ferent lengths are essential to the development of a more useful set of functional relationships. Another shortcoming of the interwar elasticity estimates is that they apply to large groups of food items and may rep- resent such a high degree of aggregation so as to have limited use in everyday problem solving. For example "meat" may in- clude everything from fish to beef, and within sub-groups, such as beef, there are many different grades and retail cuts. In any particular time period prices and supplies of these sub-groupings may be moving in different directions, and over a period of time the relative proportions of each group changes. This makes it difficult to construct a representa- 55 tive price and quantity series of data over time. Fox28 found that a simple average price in some cases moved in the opposite direction from a weighted average price. The consumption data used in all of the studies re- ported above were based on carcass weights at the wholesale level whereas the price series are based on the retail prices collected by the Bureau of Labor Statistics in 55 large cities throughout the United States. Using the car- cass weight data on beef, pork, veal and lamb, it is im- possible to treat separately the processed sausage items which may have somewhat different consumer demand charac- teristics than fresh beef or fresh pork. The price series are based on a few retail cuts of_a specific grade of meat and therefore cannot adequately reflect price changes when the price spreads between grades fluctuates. Although numerous criticisms have been directed toward inter-war period elasticity studies based on market data, it should be recognized that these studies have been limited by the availability of adequate data. The results obtained, have provided an essential background for the development of more detailed demand studies as suitable data becomes available. 28 Karl A. Fox, The Analysis of Demand for Farm Products, U.S.Dept. of Agr., Tech. Bul. 1081, 1955, p.26. CHAPTER IV THE SOURCE AND NATURE OF DATA The Operation of the Consumer Panel The M.S.C. Consumer Panel is a group of 200 to 250 families residing in Lansing, Michigan, who keep detailed records of their food purchases. Diaries are filled out each week showing the price, quantity, and total expendi- ture for each food item purchased. (See Appendix for copy of diary). These diaries are then mailed to the Department of Agricultural Economics, Michigan State College, where the data are transposed onto IBh cards. The principal source of data for this dissertation was these food purchase records. The first diaries from the panel were received in Feb- rruary, 1951; however, it was late summer of that year before as many as 200 families were reporting regularly. Since that time the number of panel members has risen to about 250. The project, which supports the panel, was approved in late 1948 and was designed to run for ten years. The objectives of the original project were as follows: "The first is to determine the effect of price changes (both real and money) upon the quantities of food purchased, and the associated time-lag in adjustment. The second objective is to determine the effect of a change in income (both real and 58 money) upon the quantity purchased and the expendi- ture for various food products, and the associated time-lag. The third objective is to measure the effect of price changes and income changes upon substitution among different products. In a sense, therefore, the objectives are to determine price elasticity, income elasticity and cross elasticity of demand."1 The leadership for the organization and operation of the panel has been the responsibility of Dr. Gerald G. Quackenbush and Dr. James D. Shaffer. Dr. Shaffer's doc- toral dissertation dealt with the methodological problems of organizing and operating the panel.2 This dissertation will touch briefly upon some of the characteristics of the sample as they affect the representa- tiveness of the data that were used in the analysis of de- mand for meats. A more detailed discussion of the sampling plan can be found in Dr. Shaffer's doctoral thesis and in a recent journal article.3 The first step toward obtaining a representative sample of families was to conduct a sample census of the Lansing 1 Gerald G. Quackenbush, "Demand Analysis from the M.S.C. Consumer Panel," Jour. Art. No. 1594 of the Michigan Agr. Exp. Sta. A paper delivered at joint meeting of the Amer. Stat. Assoc., and the Amer.Farm Econ. Assoc., Washington, D.C., Dec. 50, 1955. 2 James D. Shaffer, Methodological Bases for the Qpera- tion of a Consumer Purchase Panel, Unpublished Ph.D. thesis, Michigan State College, 1952. 5 James D. Shaffer, "A Plan for Sampling a Changing Population Over Time," Jour.Farm Econ., 56:155-65, 1954. 59 population to learn more about its characteristics. A ran- dom sample of 2000families was systematically selected by taking every nth residential address from the complete list of addresses in the Lansing City Directory published by R.L. Polk and Company. The sampling rate was approximately seven percent. The same area, as defined in the city directory, included corporate Lansing plus the highly urbanized fringe but excluded East Lansing. A total of 1885 interviews were completed in the spring of 1950. From this group a sub- sample of 500 families was drawn, stratified on the basis of income of the household, number in the household, age of the housewife, and education of the housewife. A plan was set up whereby panel members received about fifty cents (per week for keeping the food diary. As would be expected, not all families would agree to be cooperators. The families least likely to cooperate were those in low or high income groups, those where the housewife had an 8th grade or less education, those with broken homes, those where the housewife was elderly, and those where both the husband and wife were employed.4 As panel members drop out, new members are recruited from the list of families catalogued in the sample census. 4 Ibid.,p.156. 60 The new family is selected so as to be as much like the old family as possible. When families move out of the city, an attempt is made to replace them with new families moving in to the city. Provisions are also made for picking up a pro- portionate number of newly formed families so as to maintain a representative sample over time.5 A second sample census was made in 1954 as a basis for revising the sample and to provide a new pool of potential members. The Characteristics of Lansing 1950 census. Some knowledge of the basic characteris- tics of the Lansing population is essential to an appraisal of the research findings based upon data from the M.S.C. Consumer Panel. One of the best standards of comparison now available is the published results of the 1950 census of population. Table 6 summarized this information for Lansing, the state of Michigan, and the United States. These statistics indicate that Lansing is a city with a fairly high level of income. The median family income in 1949 was $4,097. This is one-third above the average for the Unitedetat s and is 19 percent above the average income of urban families in the United States. This higher than average income level is also evidenced by the relatively small percent of families with incomes under $2,000 per year and the higher than average proportion with incomes over $6,000. 5 Ibid.,p.159. TABLE 6 61 CHARACTERISTICS OF THE LAN§ING POPULATION, 1950 CENSUS" *— t United States Characteristic Urban Lansing Michigan Michigan Urban Total Percent 65 years old . or over 8.0 7.2 6.6 8.1 8.1 Percent non-white 3.3 7.1 9.5 10.0 10.4 Persons per household 3.16 3.42 3.39 3.24 3.38 Percent of males 14 years old and over in labor force 81.5 80.1 81.9 76.1 76.4 Percent of females 14 years old and over in labor force 36.3 27.3 30.2 42.5 36.7 Percent labor force unemployed“ 4.8 5.4 5.8 5.6 4.3 Percent employed in manufacturing 33.8 40.9 44.3 29.4 25.9 Kedian income;fimfilies $4097 $3519 $38l5 83431 83073 Percent of families with incomes less than $2,000 20.7 28.4 24.4 32.6 38.6 Percent of families with incomes over 21.6, 15.7 18.6 15.3 12.3 $6,000 - *SOURCE: U.S.Bur. of Census, 1950 Census of P0pulation, Vol.II, Characteristics of the PopuIation. 62 Other population characteristics which might be of interest in studying demand for food, are theILow percent- age of non-whites in the population, the smaller than aver- age size of households, and the high propertion of persons employed in manufacturing compared with the average for the United States. The proportion of persons engaged in manu- facturing, however, is low compared to the average for urban areas in Michigan. This is probably due to the fact that Lansing, being the state capital, has a sizeable number of persons employed in public administration positions. As is true in many other cities in Michigan, the manufacturing industry is dominated by firms producing motor vehicles and motor vehicle parts. Other characteristics of Lansing, which might be of use in appraising the demand for meat, include location with respect to livestock production and slaughter, the kinds and sizes of retail outlets, and the amount and kind of meat advertising. Deficit areas in meat production. Michigan is a deficit area from the standpoint of livestock production as related to meat consumption (Table 7). Liveweight farm pro- -duction of meat animals is equivalent to about 44 percent of total meat consumption in the state. Due to substantial in- shipments of live animals to slaughterers, dressed meat pro- duction is 76 percent of total consumption. The largest 63 deficit is in pork products. A high proportion of the beef produced on michigan farms comes as a by-product of the dairy industry. Therefore, a higher percentage of the locally produced beef would grade U.S. Commercial and lower. Veal pro- duction is relatively high in relation to consumption; whereas lamband mutton production probably is about equal to or slight- ly less than consumption. Meat retailing in Lansing. The organization and opera- tion of retail stores handling meat in Lansing is not unusual for a city of this size. Three large national chains operate stores in or near Lansing. These firms are The Atlantic and Pacific Tea Company, National Tea Company, and The Kroger Company. A local chain operates three supermarkets, and un- til 1953 another local chain operated six supermarkets. The second local chain has since been bought out by National Tea. These firms operate a total of 19 supermarkets in or near 'Lansing, 17 of which carry a complete line of self-service meats. In addition to these larger stores, there are several individually owned superettes and a large number of small neighborhood groceries which carry meats. Up until February 1953, when price controls were lifted, most of the larger supermarkets featured U.S. Choice grade beef. Since that time, about one-third of the stores have dropped down to U.S. Good grade or its equivalent in packer 64 TABLE 7 ° PRODUCTION AND CONSUKPTION OF MEATS IN KICHIGADa Percent of U.S. Total v, N Liveweight Dressed heat Meat hind or meat Farm Production by Consumption Production Slaughterers Jan.-har. 1947 1947 1944 b Beef 2.05 3.55 -- Veal -— b 5.92 _- .Lamb and mutton 1.63 1.51 -- Pork 1.39 2.31 —- All Meat 1.72 2.94 3.89 a Grover J. Sims and Lucile Johnson, "Geography of meat Animal Production and Meat Consumption," Livestock and Meat Situation, Bur.of Agr.Econ.,U.S.Dept.of Agr., August 1948, pp.17-23. b Beef cattle and calves are combined in the total for beef. 65 brands.'7 Competition is fairly keen among the larger stores. Each week these stores run large ads in the food section of the Lansing State Journal which is the only local daily newspaper. It seems probable that this may affect the rela- tive quantities of the different cuts of meat purchased in Lansing in any given week, particularly when the same cut, or same kind of meat, is featured as a special by more than. one chain. Representativeness of the Panel Over Time In appraising the reliability of the data from the M.S.C. Consumer Panel, it is important to know something of the representativeness of the sample over time. It has al- ready been pointed out that some difficulty was encountered in establishing the original sample so as to be completely representative of the population as classified in the sample census. Perhaps, even more important is the stability of the sample over a period of time. When characteristics of the sample are compared to those of the sample census, it appears that the panel has 7 While O.P.S. controls were in effect all beef was government graded. With the suspension of controls grading was no longer compulsory. 66 remained relatively stable over the two year period 1951 to 1953 (Table 8). The average age and education of the house- wife has tended to be slightly above the levels found in the sample census. The number of persons per family has been main- tained near the sample census average of 3.28 with deviations within a range of one-tenth person above and below the average. In appraising the stability of the sample with regard to level of income, it must be remembered that families are clas- sified on the basis of last year's income. The basis for calculating the average income of panel members changes each January 1. In comparing the income levels in Table 8 it is necessary to take into account the general upward trend in income payments. Last year's income for panel members reporting during the week beginning July 1, 1951, was $4,463. With few ex— ceptions average family income fluctuated between $4,000 and $4,100 during 1951 (based on 1950 income). This is approxi- mately $300, or eight percent, above the sample census average of $3,738 (based on 1949 income). Since national average dis- posable personal income increased by eight percent from 1949 to 1950, the panel appears to have been fairly representative as to level of income during most of 1951. If family incomes in Lansing moved parallel to national disposable income per person, the average level of income for panel members in 1952 (based on 1951 realized income) should 67 TABLE 8 CHARACTERISTICS OF THE AVERAGE FAMILY IN THE M.S.C. .CONSUMER PANEL AT DIFFERENT TIME PERIODS COMPARED TO THE AVERAGE FAMILY IN THE 1950 SAMPLE CENSUS W a:— '14.; 1.1;.“ Family Characteristic Sample July 1 Jan. 1 July 1 Jan.;l June 50 Censusa 1951 1952 1952 1955 1955 Average age of Average education of housewife (years) 10.9 11.2 11.1 10.9 11.2 11.3 Average family in- come last year b d (dollars) 3758 4465C 4184 4055d 44059 45846 Average number of persons perfamily 3.28 3.28 5.59 5.25 5.29 5.18f Number families reporting 1885 179 207 223 242 230 a Personal interview survey of 1885 Lansing families. b 1949 income after taxes. c 1950 income after taxes d 1951 income after taxes e 1952 income after taxes. f A sample census of 1,110 families in the spring of 1954 revealed an average size of family of 3.18. 68 have been 84,359. Oh this basis the actual level appeared to be less than desired for an optimum sample; however, this seemed to have been corrected as the panel moved into 1953. The panel average of 84,584 for June 30, 1953 was about 22 percent above the 1949 level of $3,738. This com- pared with an overall increase in national disposable income of 19 percent for the corresponding period. Local income data on gross_week1y earnings of manufac- turing workers in Ingham County, where Lansing is located, showed an increase of 21 percent from the last half of 1951 to the first half of 1953. Weekly average income of panel members reported on a current basis rose 16 percent during the same period. The difference in rate of increase could be due to the lag in wage increases received by non-manufac- turing workers and to the increase in overtime pay for manu- facturing workers. Families with fixed incomes also affected the panel average. In spite of minor fluctuations in the characteristics of the panel, as measured by averages on control factors, such as age and education of housewife, size of family, and family in- come, it appears that the panel group has been reasonably rep- resentative over time. Another aspect of the sampling problem is the continuity of participation of panel members. Table 9 shows a frequency distribution of families that have participated in the panel 69 for varying lengths of time. This indicates a considerable stability of the panel since it began in the spring of 1951. TABLE 9 FREQUENCY DISTRIBUTION OF FAMILIES PARTICIPATING CONTINUOUSLY IN THE A. S. C. COHSUEER P REL FOR VARY EG LENGTHS OF TIRE“ ' Cumulative Length of Time ' Number or Number of Participating Families Families More than 2% years 50 50 2 to 2% years 38 83 1% to 2 years 57 125 l to 1% years 27 152 * Based on families in panel for week ending October 24,1953. Continuous participation is defined as families not missing more than two weeks diaries in the time period. Preliminary Processing of the Data IBM analysis. When this study of demand for meat began in November 1952, the procedures had been set up for coding the data from the panel diaries and punching them on IBM cards. 8 A listing of the IBM cards by families was being used to check against the original food purchase diaries for errors or omis- sions. Each week's purchase records requires approximately 6,300 IBM cards. This does not include the different sets of summary cards which are punched as the analysis proceeds. 8 These procedures for IBM analysis were developed and carried on under the supervision of Dr.G.G.Quackenbush and Dr. J. D. Shaffer. ‘70 When this meat study began approximately 530,000 IBM cards were available from more than one year's Operation of the M.S.C. Consumer Panel. As the study progressed IBM cards became available for another year of panel Operation. The basic IBM cards were sorted into three income groups and within each income group the cards were serialized by product number. The income groups were set up so as to divide the panel families into three groups each containing about an equal number of families. The income measure used as a basis of classification was last year's annual income after federal income taxes. The product numbers were those listed in the food purchase diary. (See Appendix). The IBM cards were then summarized and tabulated so as to yield the following information on a weekly basis for each of the major meat groupings, such as beef, pork, veal, etc.: (1) Total quantity purchased by all families by income groups. (2) Total expenditures by all families by income groups. (5) Average price by income groups arrived at by dividing total expenditure by total quantity. (4) Average quantity purchased per family by in- come groups. (5) Average quantity purchased per capita by in- come groups . ' (6) Percent of families buying. 71 At a later date a similar analysis was made for retail cuts of meat. However, this was confined to a summary for the entire panel disregarding income groupings of families. Adjusting the data. The weekly observations on quantity per family, average price, and percent buying were plotted graphically for each of the major meat groups (beef, veal, lamb, pork, other meats, poultry, and fish). It was apparent from these graphs that fluctuations in average prices and average quantities, based on the panel data, were more erratic than could reasonably be expected from similar observations for the entire population. This brought about a careful re- checking of the data where several processing errors were discovered and corrected. This, however, did not remove all of the seemingly erratic observations. Further checking disclosed that locker purchases, gifts, and game were causing substantial disturbances in the average price and average quantity series. It was reasoned that the sample of families in the panel was too small to provide reliable estimates of weekly locker pur- chases occurring in the total population. For example, in a week when one family purchased a 500 pound side of beef the average weekly purchases for all panel families would be increased by more than one pound per family. With typical purchases averaging near 2% to 5 pounds per family per week, 72 the occurrence of a locker purchase represents a greater shift in purchases than was likely to be true for the total popula- tion. The average price was also biased downward since locker purchases are usually reported on the basis of a wholesale price. Therefore, it was decided that locker purchases should be adjusted out of the data for this study of consumer response to price changes. It was recognized that such an adjustment would bias meat consumption measurements downward, particularly on beef and to a lesser extent on pork (Table 10). The inclusion of deer, game birds and fish in the quan- tity series also caused substantial disturbances during the seasons when they were important. Since no expenditures were listed for these items the average price series for other meats, poultry, and fish would fluctuate with changing propor- tions of game. It was decided that all game items should be adjusted out of the data for this study.9 For similar reasons gifts were subtracted, but this was a minor item. The relative importance of locker purchases, gifts, and game, as related to annual meat consumption, is summarized in Table 10. In working with the data, it became obvious that veal and lamb were purchased by a relatively small segment of the population. In any given week, only about three percent of the families were buying lamb or mutton and abOut eight percent 9 Hereinafter the meat group referred teas "other meats" will be labeled "sausage" since the adjustments leaves most- ly franks, weiners, and assorted varieties of cold meats. 75 TABLE 10 QUANTITIEU OF KEATS PURCHASED BY FAMILIES FOR THE YEAR JULY 1951 TO JUNE 1952, M. S. C. CONSUMER PANEL Pounds per Family hetail Locker Purchases Purchasesa Kind of Meat Otherb Total Beef 125.34 9.46 1.54 136.34 Pork 135.08 4.63 1.53 141.24 Veal 10.74 -- -— 10.74 Lamb or mutton . 4.76 -— -- 4.76 Other meats . 52.19 -- 5.02 57.21 Total red meats 328.11 14.09 8.09 350.29 Poultry 60.83 .93 4.02 65.78 Fish 23.18 -- 4.13 27.31 Total all meats 412.12 15.02 16.24 443.38 a In February, 1954, 13 percent of the panel members either rented a locker or owned a home freezer. The quantity of meat purchased for locker storage is probably underesti- mated since it included only "large" individual purchases. b Includes gifts, home grown and game. A substantial portion of deer is reported under other meats. Game birds are part of the fish total. 74 were buying veal. Under these conditions, the average price and quantity series were extremely erratic, reflecting week to week changes in the composition of these commodities. Because of this instability in the data and the relative un- importance of these items, veal and lamb were not included in subsequent analyses (Table 10). Another problem that became apparent was that the classification of families on the basis of last year's income did not bridge smoothly from one year to the next. For ex- ample, on January 1, 1952, all families were reclassified into income groups based on 1951 income. In this process the dividing points were selected so as to shift a sizable number of families from the low to the medium income group and from the medium to the high income group. This did not become known until after the data for the first half of 1952 had been processed. Because of this difficulty, it was decided to consolidate the observations for the three income groups into overall averages for the entire panel. This process of aggregation also had the advantage of giving greater stability to some of the price series, particularly for the less im- portant meat items. I Future studies should give attention to differences in income level as related to meat purchase patterns over time. An inspection of meat purchases by income groups indicated 75 mm .32 case n Ema 32. £335“ Hausa headmcoo .06.: .8. doggone.” ascend Adamo: Ho cwdhmbd .pH madman nmm. “—o x003 Eww 2 Em. we x003 ENN .. 9.025 mm mm mm 39.26 9.32: x33 m. Ill. 2:35 22.50 Ill NN Om Ow ON Om 0m 00. 0.. ON. 96:00 76 that medium and high income families purchased more beef than low-income families. However, part of the difference may have been due to differences in size of family. The graphs also showed the average price paid for pork was slightly lower for low income group as compared to the medium and high income groups. A summary of "current" incomes of panel families showed wide variations from week to week (Figure IV). Each family reports its total income payment actually received during the diary week. (See page 15 of diary in Appendix). Part of the families are paid on a weekly basis while others are paid bi- weekly, monthly, or at irregular periods. The problem was to develop a time series of weekly observations that would re- flect short term changes in current income. A simple average of weekly incomes reported by all panel families sometimes fluctuated from $65 to over $100 within a period of a month. It seemed unlikely that such an income series would be satis- factory as a variable in a multiple regression analysis designed to explain weekly meat purchases. However, it is recognized that meat purchases by individual families may be affected by the timing of pay periods. In an effort to smooth the income data, a four week moving average was computed using the current week's income and the incomes of the previous three weeks. This appeared to be much more satisfactory than the unadjusted series; 77 however, some irregularities still existed when monthly pay days fell outside the four week period. Consequently, it was decided that a 15 week moving average should be computed using the current week's income and incomes for the previous 12 weeks (Figure IV). The use of a lagged moving average, instead of a centered moving average, implied that the in- come already received had more effect on meat purchases than the anticipation of future income. In some instances this may be questionable, but for the majority of the cases the lagged relationship seems more apprOpriate since a relatively small prOportion of the meat is sold on credit basis.lo Limitations of the Data In addition to the problems mentioned above, certain limitations of the data should be recognized. One of the more obvious limitations was that the retail prices of beef were under the control of the Office of Price Stabilization from Kay 1951 until February 1955. During the later portion of 1951 and the first half of 1952, the prices of most retail beef cuts hovered near ceiling levels. Later in 1952, prices declined below the ceiling for many beef items. There was 10 A survey of 1,351 retail meat stores in the North Central Region in May 1955, revealed that 24.5 percent of the total meat sales were on a credit basis. 78 little to indicate that any real "shortage" of beef occurred in retail stores in Lansing during the price control period. (See Chapter VI for a more complete statement on the effect of CPS controls on beef). Another limitation of the panel data was that the segre- gation of beef purchases by grades was impossible. There was little that could be done about this since it is believed that most housewives are unable to identify and report beef pur- chases by grade. Furthermore, with the elimination of com- pulsory government grading, a substantial portion of the beef carries one of many packer grades or no grade identification at all. Even in stores carrying graded beef, the identifica- tion stamp is frequently removed in the process of breaking wholesale cuts into retail cuts. The non-homogeneity of products from week to week probab- ly leads to some "false" changes in the prices of these items due to method of computing the average price series. Neverthe- less, there are practical limitations on the amount of detail that can be gotten on food items purchased for home consumption. Undoubtedly there are errors in the reporting of purchases. Since meat is a major item in the food budget, it is believed that errors due to emission of purchases are relatively unim- portant. There is some reason to believe that some of the veal purchases are being reported as beef and that hams and picnic hams are sometimes confused by panel members. Page 79 lacking in numbering only. IIVE’ZSITY MICROFIIMS 80 Level and Pattern of Meat Consumption Per capita purchases of all meats for home consumption by the M.S.C. Consumer Panel were 15 percent less than the per capita meat consumption estimated for the United States for the year July 1951 to June 1952 (Table 11). Red meat purchases by panel members werell.4 percent below the U.S. average. The differences between panel purChases and U.S. average consumption estimates might be attributed to some of the fol- lowing reasons: (1) the panel quantity series do not include meat eaten away from home as snacks or as part of meals eaten out; (2) the estimates for U.S. average consumption may be in error or may not be comparable in terms of retail weight equivalents. It is believed that the first reason is the most impor- tant. An examination of the data reveals that panel families spent 12% percent of their total food bill during 1952 for meals eaten out. This does not include snacks, such as ham- burgers and hot-dogs. A recent study in Kinneapolis, Kinne- sota, indicated that about 16 percent of all food (valued at retail prices) was passing through public and private eating 11 places and institutions. "A slightly higher proportion (22 percent) of the total value of meat, sea food, and poultry was 11 Lester C. Sartorius and Marguerite Burk, Eating Places as Marketers of Food Products, Bur.Agr.Econ., U.S.Dept. of’Agr. in cooperation with the Univ. of Minn., Marketing Research Report No. 5, 1952, p.89. 81 marketed through eating places, apparently because of the greater_emphasis on meats in eating places, particularly in the higher pricedcnuns"12 Generalizing from the above observations it seems reasonable to conclude that total meat consumption by M.S.C. Panel families was greater than the U.S. average consumption when proper allowance is made for meat eaten away from home. It is doubtful that the panel estimates of meat con- sumption are biased downward appreciably. The extremely high income segment of the population is not well repre- sented in the panel, but on the other hand, neither are the extremely low income families. There is reason to believe that the average consumption estimates are fairly representa- tive of the true population parameters.13 There is some reason to believe that the U.S. average consumption figures for meats may be biased upward. The original estimates were in terms of wholesale carcass weight equivalents. These were converted to retail weights by using the following conversion factors:14 Beef ............... .79 Pork, excluding lard .95 Vea100000000000000000091 Lamb and mutton..... .89 12 Ibid., p.88. 13 This was developed on page 65. 14 Bur. of Agr.Econ., U.S.Dept. of Agr., Consumption of Food in the United Statesl Agricultural Handbook N6. 62, 1955, p.155. 82 Tkmnli MEAT CONSUMPTION BY M.S.C. CONSUMER PANEL COMPARED TO U.S. AVERAGE, JUEY 1951-JUNE 1952 Pounds Per Capita Panel as United States M.S.C.PaneI Percent of Kind of Meat U.S.Average Carcass Retai Retail Weighta Weight Weightc Beef 56.5 44.6 41.6 95.5 Veal 6.5 5.9 5.5 55.9 (Beef and veal) (65.0) (50.5) (44.9) 88.9 Lamb and mutton 5.7 5.5 . 56.7 Pork (excludingikufi) 71.7 66.7 45.1 64.6 Sausage -- -- 17.4 -- Total red meats 158.4 120.5 106.8 88.6 Poultry 27.2 27.2 20.1 75.9 Fish 11.1 11.1 8.5 74.8 Total all meats 176.7 158.8 155.2 85.1 a Based on quarterly estimates of red meat consumption as reported in The Livestock and heat Situation, U.S.Dept. of Agr., Agr.fikt.Ser., January 1954, p.7. Based on conversion factors used in Consumption of Food in the United States, 1909-52. U.S.Dept.of Agr., Bur. of Agr.Econ Agr. Handbook No.62, 1955, p.155. Based on total quantities reported by M.S.C.Panel mem- bers. The "sausage" grouping includes all luncheon meats, franks and processed meats where beef, pork, etc., lose their identity. Therefore, the Lansing retail weights are not directly comparable to the U.S. averages.‘ 'L» n cab? .E‘;-T-,£;;.'T.’.’*-§73733‘: mm .. 2‘1 L'1'2V‘LV.- .-T '- jfiflu'u( .1. . lied. I‘;Jsr aasrixt ”Jv 1:} : 718' girl-‘5'." 3.6? 8.1: “.b: 3.06 8.86. .‘-:.~”:. \ .r‘. a .e , (7.qu vj" .93) (.1.‘ ‘1) (0.52.8 , (Inc? has I $.56 :01. ' .Iv‘? v0 7 ‘ 5033” .1 Ni (7.5?" 1.3:, 73‘): $.12? (21113111513131! ' J -_ ;,\j —— -~ ~CM' ,_ , 7 , . - -; 'vkbvj“ a.“ (,1; - 6.",SI 9mm siaam 50': n7, , 1.05: - 89/3 8.735: 8.9? 5.8 .ZI‘ 'I.££ 4A4; L38 $.31: _s.s,d£ new: 3"”:Lu5W .11 2 3 81-999.: 803338 103 Vtwg .em .91.a:(109m.131 anal IMJJJI rd homage-z "33¢ng “do” )7“? so Wm: :14. “buxom parquet; can-nu" our, :1 :‘7 .1556: .10“! "on: null 5793.10.61; “has ' L " gnaw We“ .noteqem «new: and: o .8. am at 016818111500 tuao115 son on ”W 85 The conversion factors for beef and pork may be appropriate for portions sold as fresh cuts, but it is doubtful that ample allowance has been made for weight losses for cured and processed items. Weight losses, due to shrinkage and deterioration in the retail store, probably amounts to about 5 percent of Wholesale weights.15 A tentative conclusion is that the U.S. average consumption figures used in this study probably overestimate actual weights purchased by consumers. A general conclusion at this point is that total meat consumption by the U.S.C. Consumer Panel families is greater than the U. 8. average. This would be expected since the in- come level of Lansing families was about one-third above the ‘U.S. average based on the 1950 Census. (See Table 6). A re- cent study of regional variation in red meat consumption showed that the North Central Region was 6 percent above the U.S. average based on 1944 data compiled from records of the Office of Price Administration.16 The 1948 Food Consumption Surveys, made by the BENRE of the U.S. Department of Agricul- ture, indicated that consumption of red meats by urban persons exceed the U.S. average by 22 percent on beef, 2 percent on pork, and 51 percent on lamb and mutton.l7 15 National Livestock and meat Board, Pricing_Retail Meat Cuts, p.15. 16 J.C.Purcell and V.John Brensike, Net Karketing and Slaughter of Livestock and Consumption‘by Regibns, 1950, Bur. of Agr.Econ.,U.S.Dept.ofAgr.,preliminary manuscript. 17 U.S.Dept.of Agr.,Fami1y Food Consumption in the United States, 1942, Miscellaneous Publication No.550, 1944. 84 The above discussion centered around the reconciliation of the differences in levels of meat consumption between the M.S.C. Consumer Panel and the United States averages. A ques- tion might also be raised about the pattern of consumption among the different kinds of meats. According to Table 12, the Lansing pattern is very similar to that found in the North Central Region during the 1948 Food Consumption Survey. In both cases, lamb and mutton consumption is extremely low, be- ing only a little more than one-third of the national average. The proportion of beef and veal is relatively larger in the 1948 study than in the Lansing panel data for 1951-52. This is probably due to shifts in relative supplies and prices of pork and beef between the two periods. In both periods, beef and veal consumption exceeded pork consumption. In general, the pattern of meat consumption in the panel seems to compare quite closely with the results of previous studies. Lansing Prices Compared to Detroit BLS Prices Since the price series derived from the panel data were weighted average prices obtained by dividing total expendi- tures by total quantities, some question may exist as to whether these prices accurately reflect price changes in the retail stores. Several comparisons of panel prices for the Lansing market with prices for similar commodities, reported CONSUHPTICN PATTERN FOR RED ME TO PATTERN AND 85 TABLE 12 "ATS IN LANSING COMPARED FOR THE NORTH ENERAL RTGION THE UEITED STATES Beef & Pork Lamb & Other Total Place Veal Mutton Meat b Percentages Based on Panel Data Lansing 42.0 40.5 1.4 16.5 100.0 United States North Central Region United States North Central Region Percentages Based on 1948 Survey0 45.7 56.7 4.5 15.5 100.0 45.5 58.1 1.5 16.9 100.0 Percentages Based on 0PA Records for 1944 49.5 , 47.9 2.8 -" 100.0 48.9 50.2 , 0.9 ' -- 100.0 a All data, other than for Lansing, taken from "Net Market- ings and Slaughter of Livestock and Consumption’by Regions, I950, preliminary report by 32C.Purcell and V.John Brensike, Bur.of Agr.Econ.,U.S.Dept.of Agr. b M.S.C.Consumer Panel, July 1951 to June 1952. c 1948 Food Consumption Surveys conducted by the Bureau of Human Nutrition and Home Economics, U.S. Department of Agriculture. 86 by the Bureau of Labor Statistics for Detroit, have been made. Direct comparisons are difficult because the BLS price series are quoted for Specific retail meat cuts.18 For beef items, U.S. Choice and U.S. Good, grades are the basis of reporting. The panel price series are for fairly broad groups of retail cuts with no grades specified. The BLS prices are taken by market reporters, who make the rounds of sample stores during the first three days of the week, during which the 15th of the month falls. The panel prices are based on purchases over the entire week and therefore, are more likely to reflect the influence of meat price specials featured during late week trading when more than one-half of the meat is purchased. Since there is considerable difference in the definition of the retail cuts involved in the two price series, a com- parison of the level of prices probably is of little signifi- cance for most cuts. More important is a comparison of price trends over time for similar items. Such a comparison has been made graphically and by correlating the pairs of price series. Figure V shows a graphical comparison of retail prices for beef item . The prices of ground beef moved together very closely, with a correlation of .98. The prices for roasts and steaks showed similar patterns, with the Detroit price declin— ing less rapidly than Lansing prices. The wide difference in 18’Bur.of Labor Stat.,U.S.Dept.of Labor, Food Pricing Specification Manual, January 1954. 87 [ Cents perpound — Detroit -- Lansing RD. STEAK IOO *- 90- \ /\ P/’\\ RIB ROASTS eo— ROASTS/\ \ /\/ \ \J \v * GROUND " BEEF eo— (‘2lllllllelllllllllllle$ JULY ocr. JAN. APR. JUL. oer. JAN. APR. l95l I952 l9 55 Figure V. Comparison of Lansing M.S.C. Consumer Panel prices and Detroit B.L.S.prices for selected cuts of beef, July 1951-June 1955. 88 level of prices is due largely to the differences in the grade of beef. The BLS series specified U.S.Choice and U.S. Good grade in rib roasts, while the Lansing price is an average for all kinds and grades of roasts. The BLS series also specifies U.S. choice and U.S.Good round steak, while the Lansing price represents all kinds of steak. The correlation between steak prices was .85, and between roasts and rib roasts it was .92. Figures VI and VII, show a graphical comparison of prices for selected pork cuts. The prices on pork chOps moved to- gether fairly well with differences in price level again due to the item specifications. For the BLS price only center cut pork chops from No.1 loins are represented. Rib ends or shoulder ends or should end chops and soft or oily pork was excluded. The Lansing price represented all kinds and quali- ties of chops. The correlation between the two price series was .80. The correlation between bacon prices was .89 with the BLS Splsice representing sliced and packaged, one pound units of Standard Grade A bacon. Although ham prices tended to follow the same general Ifilttern, there were much wider variations in the Lansing price SEiries than in the Detroit prices. The correlation between tile two series was .40. The BLS specification calls for skin- 89 Gents _ DetrOif per pound -- Lonsmg - J 90 - ’ - -- A q 4 °€ " lulnlllllllllljllllllll T JULY ocr. JAN. APR. JUL. OCT. JAN. APR. I95| l952 l953 Figure VI. Comparison of Lansing M.S.C. Consumer Panel prices and Detroit B. L.S. prices for selec- ted cuts of pork, July 1951-June 1955. 90 Cenrts -— Detroit per pound --- Lonsmg 80 '- 50 »~- ~ I " éIlllllillllflllillllllll} JULY 0011 JAN. ARR. JUL. ocr. JAN. APR. I95I l952 I953 Figure VII. Comparison of Lansing, M.S.C. Consumer Panel prices and Detroit B.L.S. prices for ham and frankfurters, July 1951- June 1953. 91 ned, smoked, tenderized ham weighing between 10 and 16 pounds and of the "best quality." The Lansing price probably reflects to a greater extent the impact of "specials“ on whole or half hams . . The prices of frankfurters (includes almost all types of weiners) were closely related with few exceptions. A correla- tion of .74 existed between the Detroit and Lansing prices. In conclusion, it appeared that Lansing prices derived from panel data displayed a close relationship to Detroit BLS prices when allowance is made for differences in commodi- ty specifications. The major price changes are reflected in both sets of prices; however, the Lansing prices showed more variability. Presumably this variation in Lansing prices re- flected more of the effects of "specials" and to some extent changes in commodity composition from one period to another. CHAPTER V SINGLE EQU TIUN DELAND 10 BLS FOR GROUPS OF KEATS Introduction This chapter deals primarily with the problems encountered in formulating the single equation-models used in measuring de- mand for meat groups. The food purchase diary for collecting data from panel members and the basic IBN tabulating procedures were accepted as relatively fixed elements in constructing the demand equa- tions. The measurement of demand for broad groups of meats, such as beef and pork, received the most attention because these data became available at an earlier date than the more detailed data for specific meat cuts, such as beef roasts and pork chops. Although the demand equations were relatively simple, £3everal problems arose with respect to the length of the time Ineriod for individual observations, the choice of variables tc) be included in the system, the handling of disturbances Sllch as holidays and changes in demand, and the specification 01? the mathematical functions to be fitted. 95' Pctfmds j, per Family 5.2 - _ BEEF 2.8 ' i _ I 2.4 - I A 2.0; - 3.: - j‘ 2.? - - 2.4 - 4 2.0 - - '.4' SAI'S’AG: 4 ‘ L0-”\Av/V\J~VA\fwi”VA/\f/~WJP«/~J~ I I I 'I as 'i RCULTRY I I 39 52 Weeks I?) 26 Figure VIII. Weekly average purchases of different kinds of meats by families, M.S.C. Consumer Panel July 1951-June 1952. 94 I Pounds per Family 3.: r - BFFF f?§3f' ”\ié/N/dr‘L/~_a - v2.4 - A’ ' 20- ; J <> a <> ask , . . PORK 2.8 "' - .L4- _ 20[ - < > IA- SAUSAGF 1 LG - WWW _ 0.6 r .J ‘ 4.3 4.6 f L8- POULTRY a IA- - LO- A KJJS /N\Elfi:ifaVf"'I\VNI§~IU[~AV’jh\v\v\¥\_~_,a\A\/\ ;;2 27 59 52 IS 26 Weeks Figure IX. Weekly average purchases of different kinds of meats by families, M.S.C. Consumer Panel, July 1952-June 1955. 95 F Pounds per Family R IOr 9: {l l : NW“ N NW 27 39 52 I5 26' Weeks Figure x. Weekly average purchases of all meats by families, M.S.C.Consumer Panel, July 1951- JUne 1952. 96 Pounds per Family i0 P ‘ - "l 9 _ .. 8 _ .. 7 _ a 6 t' " 27 39 52 I?) 26 Weeks Figure XI. Weekly average purchases of all meats by families, M.S.C.Consumer Panel, July 1952- June 1955. 97 Evaluation of Weekly Observations Panel members report their food purchases on a weekly basis. It is possible, however, to aggregate weekly observa- tions for pusposes of analysis. An examination of week-to- week variations in quantities of meats purchased revealed that sizable fluctuations had occurred. (See Figures VIII, IX, x, and XI.) If the week-to-week variations were largely due to in- stability because of the small size of the sample, the com- bining of several weekly observations would tend to reduce the period-to-period fluctuation. It should increase the multiple correlation coefficients for the demand equations but might have little effect on the regression coefficients as compared with equations based on weekly observations. A question might also be raised about the amount of time which consumers require to make adjustments to changes in re- tail food prices. If this requires several weeks or months there might be reasonable doubt that weekly obServations represent "true" equilibrium adjustments in the static sense. Still another argument that might be raised against the use of weekly observations is that statistical difficulties with autocorrelation may be greater than when using data based on monthly or annual observations.l Additional explanatory 1 Lawrence R. Klein, Econometrics. Row, Peterson and Company, Evanston, Illinois, 1955, p.517. 98 variables may be required to allow for ladged relationships. There are, however, soue strong arguments for the use of weekly observations. One argument is that retailers adjust meat prices on a weekly basis.2 Each week different meat items are featured as specials. Another argument for the use of weekly observations is that most families shop weekly, or more often, for meat. In any given week about 80 percent of the M.S.C. Consumer Panel families buy some kind of beef and about 75 percent purchase pork in some form. This is expected in view of the perishability of most meat items and the is- portance of the meat dish in meal planning.3 Another factor considered was that aggregating weekly data into longer time periods would have greatly reduced the number of degrees of freedom in applying regression analysis. This is not to say, however, that the amount of information lost in the aggrega- tion process is directly prOportional to the reduction in the number of degrees of freedom.4 Last, but still of considerable 2 For a recent example of short-term adjustments in re- tail meat prices, see Karketing Margins for Beef, U.S. Dept.of Agr., Agr.Hkt.Ser., December, 1955. See also: Kenneth D. Naden and George A. Jackson, Some Economic Aspects of_3etailing Chicken heat, California Agr. Exp. Sta. Bul. 754, 1953. pp.41-42. 3 An REA study now under way at Harvard University should provide some additional information on this point. 4 Klein, op.cit., pp.513-514. 99 importance, is the consideration that little empirical evidence has been available on consumer meat purchases on a weekly basis. An analysis of these shorter run adjustments would provide some additional information not available previously. For these reasons most of the analysis in this study will deal with week- ly observations. Principal Variables to be Included in the Models Subject matter considerations suggest that the quantity variable should be taken as dependent in the single equation models used to explain weekly meat purchases of families in the M.S.C. Consumer Panel. In any given week, these families step up to the retail meat counters to make their purchases, accepting as fixed the price tags appearing on the different cuts of meat. The quantities purchased by these families can be considered as their response to the price structure con- fronting them along with the many other complex forces which motivate these consumers to buy. This raises a question as to which variables should be included in the equations explaining the variations in week- ly purchases of different groups of meats. Some restrictions on choice of variables are imposed by the availability of data and the increasing complexity of computations as more. x... which theory indicates as most important are2‘(l) price of ’ 100 product being studied, (2) prices of closely related products, and (5) available income of the purchasers. Following this pattern, the demand for a particular meat group such as beef can be represented by the following functional relationship: =f(P P P P P I) beef beef, pork, sausage, poultry, fish, where Q is the average quantity purchased per family per week, P is the city-wide average price, and I is a measure of average family income. The relative importance of these different groups of meats was pointed out in Table 10, page 73. Because of the small quantities purchased, lamb and veal were excluded from the analysis. There is some question about the inclusion of fish; however, it was decided to leave it in initially so as to have a1nore complete coverage of meat items. Eggs might have been included, but it was reasoned that they were primarily a break- fast item and hence did not compete directly with most meat products. Other Variables Affecting Weekly Meat Purchases Holidays and religious customs. An inspection of fluc- tuations in weekly purchases of different kinds of meats, as shown in Figures VIIIand IX, indicates that the simple model shown above is probably inadequate to account for the wide lOl variations observed. The most noticeable disturbances occur in conjunction with the major holidays, Thanksgiving, Christ- mas, and Easter. Due to customs deve10ped over the years, Thanksgiving and Christmas are holidays when poultry meats are traditionally served. At Easter ham has become the pOpular item. These holiday customs cause tremendous shifts in the demand functions for all kinds of meats (Table 13). During Thanksgiving week 1952, purchases of poultry meats were over 500 percent above the average level of weekly pur- chases for the rest of the year (Table 15). Purchases of all other meat groups averaged about 15 to 20 percent below the yearly average. Even so, total meat purchases were 28 per- cent above average. Not only were there sizable shifts in quantities purchased, but changes also occurred in prices. The average price of poultry meats advanced about seven cents- per pound. This is probably due to a change in the average composition of this group of meats, with a larger than usual proportion being higher priced roasting fowl. The price of fish was about seven cents higher than usual; the increase being due to a larger than normal proportion of high priced oysters and other seafood delicacies. Poultry is also the most popular meat at Christmas, with ham coming in as a preferred item durin'r the Christmas-New Year's holiday period. Total meat purchases exceeded the yearly average by about the same amount as observed during 102 .pmppflfio mxmoa hapfiaoz spas mm0H pom mmmwflohfim haxooa ommho>< A. l U ‘I mHH 05.0 mnH 00.0 00H rm.0 pwofi Ha< mHH hm. m0 ow. on ma. swab em em. as we. we om. omemssm me mm.¢ .H0 00.0 mm 00.0 . Maom mo mh.a up No.0 00 0H.m moon we eo.H ewe sm.e sow mm.e aprsOm $mwwho>¢ hHHEmm *mmwho>< hHHEdm *ommao>« haafidm Mo pom Mo mom mo hog paws mo UGHM psooaom meSom unooaom meSOm psooaoa magnom some wmpmam some waspmnweo some mefi>fimmseeee qm<flM HEB mo mflQZH< EBHB.QHm< mmmfl .umzam mmeemzoo .o.m.e 0H mqm<8 103 Thanksgiving (Table 13). During 1951, the Christmas-New Yearls food shopping was divided between two different weeks, making it difficult to combine the data with those of 1952 to arrive at some sort of average (Figure VIII). Due to extensive promotional efforts of the meat trade, ham has become the favorite meat item at Easter. In 1952, ham promotions boosted total pork purchases to 88 percent above the average for the rest of the year. -In 1953, pork sales rose much less at Easter than in 1952 (Figure IX). This was prob- ably due largely to the sharply lower beef prices and somewhat higher pork prices in 1953 as compared with 1952. In 1952, the average price of beef was 75 cents per pound while pork sold at 54 cent at Easter. By Easter week 1953, beef prices had fallen to 58 cents and pork had risen to 62 cents per pound. Another point of interest about the effects of Easter is that pork sales tend to be depressed for a week or two after the holiday. Since ham.is storable for two to four weeks in an ordinary refrigerator, the large sales at Easter appear to be partially at the expense of pork sales and total meat sales for a short time after the holiday (Figures VIII and_IX). Another disturbance, similar in nature to the holiday situa- tions just discussed, is the effect of the Lenten period on meat purchases. In 1952, Lent began during week 9 and lasted until week 15. In 1953, it began during week 8 and ended with week 104 14. The effects are most noticeable in the pattern of fish and seafood purchases, with substantial increases being particularly marked in the first two or three weeks of the Lenten period (Figures VIII and IX). Lent lasts about six weeks in all and ends on Easter Sunday. In total, the cur- tailment of meat purchases by those families who abide by the Lenten customs does not appear to have a very significant in- fluence on total meat purchases of the M.S.C. Consumer Panel (Figures X and XI). Thanksgiving, Christmas-New Year's, and Easter appear to be the major holiday periods influencing demand for meats. Other widely observed holidays, such as Memorial Day, Inde- pendence Day, and Labor Day, do not appear to be associated with noticeable shifts in meat purchases. (See Table 13 and Figures VIII, IX, X, and XI.) A question may arise in regard to Labor Day because a sizable upward shift in meat consump- tion tends to occur about this time of year. In 1951, this upward adjustment occurred during the week following the Labor Day weekend, while in 1952 it coincided with the Labor Day weekend. Further consideration will be given below to this seasonal increase in meat purchases. At this point the tentative conclusion would be that Labor Day in itself prob- ably has only a minor effect on meat purchases. In constructing the single equation demand models, some provision must be made for shifts in demand due to the holi- 105 .xoms mcfipooopg one CH Umpoc ohm momwSOASQ poom Co mpooumo emu .popmsm mm £05m «hepndm So Afiooo whmvaaofl Con» .hprSpwm so muse paw hveSm no mcflwon moms hswap one J: o“ hwpmhsne mm hammode mm mwapmflpno hwpmhdsa ow hwpmpdfle bw mda>fiwmxcsne assess en messes me has honed bmpfihh hm hammocpog hm bwn monopcomopQH hprSpwm mm hspflmm mm hag Hwasoaoa asessm ma seessm ea seamen hapmhdfle H hammose H . ham m.hde%.Boz ham x003 ham mama ham goo; seefiaem mama mmma Hmma wanna maps . anma maps .qame mmaquom mes monm some mama ems mo man any has mmmmbeoo 333% 5:5 camps ~3me mamzmqso mam. no means wH mqm<8 106 day disturbances discussed above. One alternative would be to omit from the analysis weeks involving major holidays. Another possibility would be to add a separate explanatory variable to each equation with zero for all observations except the holi- day week, where a value of one would be inserted. The coef- ficient of the holiday variable would then provide an estimate of the influence of the holiday disturbance. A third alternative would be to construct a holiday vari- able with each holiday beinglgiven a value approx'mating its effect in shifting the demand function. The influence of the holiday might be approximated roughly by the deviation of holiday purchases from a regression line of quantity on price for a particular meat. This procedure would not take into account the effects of price and quantity changes among com- peting meats. An estimate of the relative influence of the holiday disturbances could also be taken from the results of the second procedure suggested above, where a separate ex- planatory variable is included for each of the major holidays. Seasonal shifts in demand. A similar seasonal pattern “-- of meat purchases appears during both of the years for which panel data are summarized. As might be expected, purchases of "all meats" are smaller during the summer and larger during the fall and winter months (Figure X and XI). Pork purchases exhibit greater seasonal variation than any of the other major 107 groups (Figure VIII and IX). The patterns of purchases for both beef and pork are similar to that just described for all meats. However, the patterns of purchases of beef and pork for the year, July 1952 to June 1955, are distorted by substantial cyclical changes in prices of these meats, which in turn are the result of shifts in supplies offered on the national market. Although seasonal variations are not large, purchases of sausage meats tend to follow a pattern different from the patterns of beef and pork purchases. Sausage items are pur- chased in slightly larger volume during the summer months as compared with other seasons of the year (Figures VIII and IX). Poultry purchases appear to have no well-defined seasonal pattern except the holiday variations already described. Fish purchases are highest during Lent and lowest during the rest of the spring and summer. A large portion of the seasonal variation in total supplies and prices of beef and pork is due to rather well-established seasonal fluctuations in livestock slaughter. Beef slaughter rises in volume during the fall as cattle are marketed off pastures. Hog slaughter also rises rather steadily from late August through November as the spring pig crop moves to mar- ket. Variations in fresh meat supplies are closely associated with corresponding seasonal price variations at the wholesale 108 level.5 These price changes are soon reflected in retail prices as supplies increase and decrease. An examination of the panel data revealed that a major portion of the season- al variation in purchases was associated with corresponding adjustments in meat prices. A question which requires consideration is the extent to which seasonal variations in meat purchases are due to shifts in the demand function. There is considerable evidence that demand for "all meats" and "all food" actually declines dur- ing the summer months.6 High temperatures tend to retard food intake and cause shifts in demand among different food items. Among the meat items, ready-to-eat cold meats and easily pre- pared steaks and chops gain in pOpularity, while heavy roasts and stewing items are less desired.7 This preference pattern is again altered as cooler weather arrives in the fall. Selected excerpts, from weekly wholesale meat trade re- ports, lend further support to the notion that demand for 5 F. L. Thomsen and R. J. Foote, Agricultural Prices, McGraw Hill, New York, 1952, Chapters‘lg and 20. 6 ”Bur. of Human Nutr. and Home Econ., U.S.Dept.Agr. Seasonal Patterns of Food Consumption, City Families, 1948. Special’fieport No.3, February 195I, pil 7 A. A. Dowell and Knute Bjorka, Livestock harketing McGraw Hill, New York, 1941, p.48. 109 . . . 8 fresn meats declines during the summer: July 4, 1955, Chicago: "With narrow consumer out- lets, due primarily to hot, humid weather, trading during the week was dull and most sales forced." July 4, 1953, New York: "The impending holiday and the beginning of family vacations were factors contributing to a relatively slow trend." July 14, 1955, Chicago: "Demand for most classes of fresh meat showed a considerable improvement over the previous week with more normal temperatures a stimulating factor." August 25, 1955, Chicago: "Sparked by cooler tem- peratures demand for fresh meat improved materially." September 8, 1955, Chicago: "Trading was marked by a series of slow, mostly forced sessions with the market in all classes and cuts very unsettled. Ex- ceedingly high temperatures curtailed consumer outlets and proved a depressing factor in the meat trade." September 15, 1955, Chicago: "Demand for fresh meat showed considerable improvement with cooler weather a stimulating factor." These comments suggest that extremely hot, humid weather Ireduces consumer demand for fresh meat, and that the advent Of cooler weather in late summer and early fall has a stimu- lating influence on appetites which increases demand for meat. In addition to the effects of extremely high summer tempera- tures, it seems that the irregular pattern of living activi- ties interspersed with vacations, picnics, and travel, con- 8 These excerpts are taken from the Weekly Livestock Market News report entitled "Market News and Statistics," issued by the Livestock Branch, Production and Market- ing Administration, U.S.Dept. of Agr., Vashington, D.C. 110 tributes to a reduced demand for fresh meats and many of the heavier foods. Eith the reopening of schools on, or about, September 1, and the end of the vacation period, more regular food habits are resumed and demand for fresh meat probably increases significantly. No satisfactory method of allowing for these seasonal shifts in demand was arrived at during the early stages of this investigation. It was decided that a close examination of the residuals from the basic equations, outlined earlier in this chapter, might be the best approach to the problem. The pattern of the residuals might yield an approximation of the magnitude and timing of the shifts in demand for different meat groups, realizing, of course, that other disturbances were likely to be compounded in the residual patterns. Another alternative would be to try to add temperature as an additional explanatory variable. Still another possibility would be to segregate the summer period and analyze it separately; but the short period over which data were available made this rather impractical. Kerchandising activities of retailers. It must be recOg- nized that there are several methods that managers of indivi- dual retail stores use to influence meat purchases in their particular stores.9 One method is advertising in conjunction 9Naden and Jackson, op.cit., pp.50-70. 111 with a price reduction on selected meat items to attract customers into their stores. Once in the store, the cus- tomers are likely to be influenced in their purchases by point-of-sale promotional material and by the location and amount of space allocated to different items in the display. In stores with service meat departments, the customer can also be influenced by the sales talk of the butcher. Then an item is out of stock, the butcher is always ready to push some other item that happens to be long in supply. Vhen attempting to explain weekly average purchases of meats by the M.S.C. Consumer Panel, there is a strong possi- bility that many of these merchandising practices cancel out, because all stores are not likely to be using the same pro- motional schemes in any given week. Nevertheless, it must be reCOgnized that the effect of advertising is not likely to be a random disturbance from week to week. In the first place, when.one of the major chain stores features a special on a large traffic item, such as ground beef or chuck roasts, a mark-down of ten cents a pound may affect nearly 10 percent of the total sales of these items in a given week for the entire city. In an effort to meet competition, other stores will feature specials, frequently on the same item or a close- ly related item. n some weeks when several of the large chains feature Specials on one class of meat, such as pork, a substantial increase may occur in purchases by panel members. 112 Frequently, some large meat wholesalers develop promotional programs in which several independent stores may feature the same meat items at the same time, with attractive in- store display materials tied in with newspaper advertisements. Weekly average purchases of meats by panel members will probably by influenced by these merchandising activities. Part of the adjustment can be accounted for by changes in prices. However, the city-wide average prices used in this study can reflect only imperfectly the prices confronting in- dividual consumers. Some individuals will purchase an item in one store at the advertised special price, while other families will have purchased in other stores where a more normal mark—up is being charged on the corresponding item. It appears possible that sizable week-to-week fluctua- tions in meat purchases may be associated with corresponding adjustments in prices. However, shifts in demand may occur, depending on the extent of advertising and other promotional activities. There is no variable in the basic demand equa- tion for meat groups to account for this type of demand shift. It would be very ‘ifficult to arrive at an empirical index representing the demand shifting effects of these merchandising practices, such as advertising. In this reSpect, the demand models are incomplete. Sizable residuals may occur for certain weeks and the multiple correlation coefficient may be reduced by the incompleteness of the model. 115 Lagged relationships. Little is known about the lagged relationships that may exist between meat purchases in one week compared with purchases during preceding weeks. Personal observation indicates that most consumers desire to provide variety in their meat diet. This variety can be obtained by purchasing different cuts of the same class of meat, differ- ent classes of meat, or by preparing the same retail cuts in a different manner. If a beef rib roast is served for Sun- day dinner one week, it is highly probable that the family will prefer some other meat item the following Sunday. When working with the combined purchases of the panel members, most of this week-to-week shift in demand by individual fami- lies is likely to average out.' However, in weeks following purchases an extensive promotion and large/of a particular kind of meat, it would be expected that purchases the next week would be reduced as families purchasing the previous week shifted to other meat items. Although separate analyses were made to obtain an indi- cation of these lag relationships, no attempt was made to include lagged variables in the basic demand equations. This should be attempted in future research. The results of a multiple regression analysis indicate that there is a significant relationship between the quantity-* of beef purchased by panel members in any given veek and the quantity of certain other meat items purchased the previous 114 week. The prediction equation computed was: X : .3804 + .2288 X2 + 1.2008 X + .1295 X + .4888 X 1 5 4 5 where X1 : the average quantity of beef purchased per family each week; X2 = the average quantity of pork purchased per family in preceding week; X:5 = the average quantity of sau- sage purchased per family the preceding week; X4 = the average quantity of poultry purchased per family the preceding week; and X5 : the average quantity of fish purchased per family the preceding week. Weekly observations for the period July 1951 to December 1952 were used, omitting the major holiday weeks. The regression coefficient for X2, pork, was significant at the 5 percent level while the coefficient for X3, sausage, was significant at the 1 percent level. The coefficients for X4 and X5 were non-significant at the 5 percent level. It is interesting to note that the coefficients are all positive, indicating that large purchases of a competing meat last week will be associated with increased purchases of beef this week. This type of analysis suggests the nature of the lagged relationships between some of the different groups of meat. It is difficult to include variables in a single equation de- mand model that will account for this type of disturbance. One possibility would be to add the quantity of competing 115 meats on a lagged basis as an additional explanatory variable. Since the price of the competing meat item is alrea y one of the explanatory variables, a high intercorrelation with the quantity series would probably cause statistical difficulties. The high degree of perishability of most meat items is largely reSponsible for the frequent purchase of these prod- ucts. Lagged relationships due to week-to-week variations in consumer stocks of meats are probably much less important than for more durable food items. Families with ordinary home refrigerators are not likely to store fresh meats for more than a few days or a week. Cured and smoked items, such as ham, may be stored two or three weeks; for this reason greater week-to-week variation might be expected in the pur- chase of these items as consumers take advantage of special prices. As more and more families become users of home freezers or large freezer compartments in regular kitchen refrigerators, the problem of lagged relationships will be- come more important. In this study, large meat purchases for home freezers or locker storage were adjusted out of the data. (See page 72, Chapter IV.) The Form of the Mathematical Function A linear relationship between the explanatory and the de- pendent variables was accepted as reasonable and practical. Scatter diagrams of the relationships between pairs of variables 116 on arithmetic scales indicated no marked deviation from linear- ity. It was also reasoned that the ranges of variation in the price and quantity observations represented such a small seg- ment of the total demand curve that the relationships obtained could reasonably be expressed by a linear function. No strong preference for an arithmetic versus a logarith— mic function could be arrived at by mere observation of scatter diagrams of price quantity observations. However, due to the large number of observations in each series of data and the number of variables in each equation, there were practical reasons for preferring the less laborious procedure of fitting a function that was linear in arithmetic terms. After experi- menting with arithmetic relationships to determine the basic factors affecting meat purchases, it probably would be desirable to try some of the equations in logarithms in order to compare results. The basic demand function used was of this type: Y a+bX1f°X2 where the coefficients b and c, represent the amount by which Y changes for each one unit change in the explanatory variables X1 and X It is important to keep in mind that the elastic- 2. ity, which is in percentage terms, varies for each point on the function represented by this type of equation. The demand F Y (Pounds per Family) l l 1 MV 50 60 7O 80 X. (Cents per Pound) 1 Y 3 3.998 - .OSSSXI 2 Y = 4.498 - .0555X 1 Figure XII. Variations in price elasticity along a demand function which is linear in arithmetic values and the effect of a parallel shift in the demand function. 117 118 function, I, shown in Figure XII, illustrates this situation. In this case the price elasticity at the mean values (Y1: 2 pounds, 2 = 60 cents) is -l.O,lO while at a price of 50 cents per pound the elasticity is —O.7 and at 70 cents it is -l.4. Assume a parallel shift in the demand function with the quantity increasing to 2; pounds at 50 cents per pound. Here it can be seen that the price elasticity at the mean values has been reduced to -O.8. Both demand functions, I and II, have the same slope; therefore, the regression co- efficients are identical. The constants, a, in the two equa- tions are different, however. The variation in elasticities, as shown in Figure XII, suggests that one should be careful in quoting elasticity measurements, particularly when an arithmetic function has been fitted. It is customary to give greater emphasis to the elasticity measured at the mean values of the variables in- volved. The Basic Demand Equations The dependent variables that were used in this series of single equation models were the average weekly purchases of five main groups of meat by all families in the M.S.C. Consumer Panel: They are as follows: 10 Computed as follows: b if or .0355 .52. . 1.0 2.5 t< to «+4»: 3 4 5 119 3 pounds of beef per family - pounds of pork per family 3 pounds of sausage per family 2 pounds of poultry per family 3 pounds of fish per family The explanatory variables are the average weekly prices of these meat groups, total expenditures, total quantity, and a measure of J5 wf1c3< mg kg ll >4 0: income. price of beef price of pork price of sausage price of poultry price of fish average weekly income per family (4 week or 13 week moving average) CHAPTER VI THE DEHAHD E*'0h BEA. Introduction Beef is the most important meat item in the food budget of Lansing families. During the period July 1951 to June 1955, weekly expenditures for beef averaged 51.95 per family. this was 59 percent of the total meat bill, compared with 29 percent for pork, 15 percent for sausage, 12 percent for poultry, and 5 percent for fish (Table l5). During an aver- age week approximately 81 percent of the panel families bought some kind of beef. TABLE 15 AVERAGE unhnLY PZRC IASES LEAT BY M.S.C. CcnoYVfiR PANEL FAIILIES, JULY 1951 to JUNE 1955* r ‘- r“ T 1 ‘ __;-t L r==lfi tn. Quantity Expendi— Expenditure Percent Kind of Heat per ture per as Percent Buying Family Family of Ieatffill pounds dollars Beef 2.929 1.952 59.1 81.2 Pork 2.557 1.445 28.9 74.8 Veal .119 ..065 1.5 8.5 Lamb .047 .050 .6 2.7 Sausage 1.115 .624 12.5 65.6 Poultry 1.245 .609 12.2 55.7 Fish .511 .268 5.4 59.0 All meat 8.501 4.905 100.0 -- Holiday weeks included. ....... QQQQQQ ....... In] t. brawl}!!! . Ill-III. v 121 The demand for beef at the retail level is actually a composite demand derived from the summation of individuals' demands for a wide variety of retail beef cuts. There are approximately 20 to 25 basic retail cuts that are obtained from a beef carcass. Tide variations exist in the cutting methods used in different retail outlets. Differences in the amount of bone and fat trimmed from the retail cuts also contribute to the non-homogeneity of cuts between stores. In addition to differences in cutting and trimming policies, there is a wide range in the quality of beef sold in different markets. In terms of U. S. Government grades, the "quality" of beef sold in retail stores in Lansing ranges from cow beef grading U. S. Utility to steer and heifer beef grading U. S. Prime. The bulk of the beef sold as fresh cuts will grade U. S. Commercial, U. S. Good, or U. S. Choice.1 A large por- tion of the lower grades is merchandised in the form of ground beef and processed sausage items. Price spreads be- tween grades of beef carcasses are usually large (Figure XIII). Among the different retail cuts, price spreads due to grade differences are greatest on the more desirable steaks, such as porterhouse, T-bone, and sirloin. Price Spreads due 1 North Central Livestock’larketing Research,Committee, "Retailing heats in the North Central States," preliminary draft of a bulletin manuscript summarizing data obtained in a survey of 1551 meat retailing outlets in the North Central States. 122 to grade differences are relatively small for the less de- sirable cuts, such as chuck roasts, stewing and boiling beef, and ground beef. These wide differences in retail beef cuts complicate the analysis of the demand for beef. The impracticality of obtaining information on grades of beef from panel members was mentioned earlier. It was also impractical to obtain a detailed breakdown of individual cuts of beef. The following sub-groupings of retail beef cuts were used: Canned beef Corned or chipped beef Ground beef, hamburger Liver Heart, tongue, other organ parts Prepared baby food, beef Roasts Steak Stewing, boiling, soup All other beef Although this classification permits a great deal of price variation within each sub-group, the M.S.C. Consumer Panel probably is one of the best sources of data that have be- come available to date to study demand for retail beef cuts. The analysis of panel data on beef purchases has yielded some interesting results. The procedures used and the results 123 obtained are described below. Attention was first centered on the demand for beef as a composite commodity. The basic demand equations formulated in the previous chapter were tested using least squares multiple regression techniques. This was followed by some preliminary analysis of the demand for retail cuts of beef. Variations in Prices and Quantities Substantial downward adjustments in beef prices occurred during the two-year period, July 1951-June 1955 (Figure XIII). This was the period for which data were available for this study. The average price of beef in Lansing ranged from 77 cents down to 55 cents per pound. Weekly average purchases by panel members varied from about two pounds per family to more than three and one-half pounds. Beef prices were subject to controls by the Office of Price Stabilization from Kay 1951, until February 6, 1955. During the last half of 1951 and the first half of 1952, beef prices and the quantities purchased by panel members remained very stable. No widespread "shortages" of beef were reported; however, prices held near record high levels and the average per capita consumption of beef for the United States as a whole declined to the lowest levels since the latter part of World War II. National average annual beef consumption per person, on a wholesale carcass weight basis, was only 55 ._ 6 i1! . Pi 124 r Cegts d BEEF PRICES per oun L- Retoil - Lonsinoa / ALL BEE? 70%- Wholesale - Chicago]D L44 ,cmonce" l [UTILITY (Cow) —| % I1 I I I I I I I I I I l I I I I I I I I I I I LI I f JULY JAN. 4-WeeK Periods JAN. JUNF l95l l952 1953 a M.S.C. Consumer Panel. Wholesale Dressed Meat Prices: Weekly Average of Daily Quotations, taken from weekly Market Reviews and Sta- tistics, Livestock Br.,P.M.A., U.S. Dept. of Agr. ° 500-600 pounds no quotations Figure XIII. Comparison of retail beef prices in Lansing with wholesale prices for selected grades of beef at Chicago, four week averages, July 1951-June 1955. 125 pounds in 1951 (Table 16). This compares with 61 pounds per person in 1952 and an all-time record high of 77 pounds for 1955. TABLE 16 MEAT CONSUMPTICN PER PERSON BY QUARTEE YEARS, U.S. AVER GE, 1950 T0 LID-1955 b Period Beef Pork Total Red heats pounds pounds pounds 1950 January-March 15.5 18.2 56.6 April-June 15.4 16.5 54.7 July-Septel‘nber’ 16.0 14.8 55.9 October-December 15.6 18.8 57.2 Year 62.5 68.1 142.4 1951 January-March 14.4 18.1 55.0 April-June 15.1 16.9 52.5 July-September 14.2 16.1 52.9 October-December 15.5 19.5 55.6 Year 55.2 70.6 155.8 1952 January—Karch 14.5 19.6 56.0 April-June 14.5 16.9 55.9 July-September 16.2 16.0 55.2 October-December 16.2 19.5 58.9 Year 61.2 71.6 144.0 195:571 January-Karch 17.5 17.9 58.4 April-June 18.9 14.9 37.1 a Agr.Mktg.Ser., U.S.Dept.of Agr. The Livestock and Heat Situation, Dec.-June, 1954, p.7. b Pork excluding lard 0 Includes beef, pork, veal and lamb, wholesale carcass weight equivalents. d. Preliminary estimates for 1955. 1“»ch \lfli III; 126 Beef supplies began to expand by the second half of 1952 (Table 16). Supplies of the lower grades of beef became more plentiful, forcing wholesale prices downward by late summer of 1952. (Figure XXX). Prices of Choice steer beef, which was being handled in most of the large self-service meat de- partments, did not decline rapidly until early 1955. The average price paid for beef by panel members began to decline during the last half of 1952, following the de- cline in wholesale prices for the lower grades of beef (Figure XIII). The quantities purchased responded promptly to the price reductions. The scatter diagram of price-quantity rela- tionships (Figure XIV) suggests that consumers may have over- responded, considering the size of the price decline. This might have been brought about by the low level of consumption in the preceding year and a half, causing consumersto become "beef-hungry." It is also possible that beef prices were low- er during the last half of 1951 and early 1952 than they would have been in the absence of control. If this were true, the response to price reductions in late 1952 may have been a return to more normal relationships between beef price and quantities purchased by panel members. The big increase in beef supplies that occurred during (the first half of 1955 was accompanied by sharp price reduc- tions, particularly on the higher grades of beef (Figure XIII). .eoppaso maaeaaoh .mmma ocssuamma ease .Hmcwm pmssmnoo .o.m.z .heon Mo mooanm was momwfloasg omwao>m haxoos Coospon QHSmGOdeHom .>Hx ondwflm 7 m Devon. 3a «2.00 m\ on my 05 w onl> w M 1 . . .71 a 1 _ _ a d _ a _ 4 q a _ p _ _ fl _ . a . _ 3 m n M .l LON r. a o o L I. o o x0 L vl . . .w XX X 0 l4 .1 xom m o J . O x .I . .xxuxxxoo as Low .vl . .XO m I4 I #X X 0 O I 7| X. .00 O Q L ..| O J 0) 0 U 00 d I I o \ d IO.M .r a 1 w 11. Q I 1 3 1 . :2 .2 £8: 2 1 . a m m m a 6 Inn to; ncm N090 1 :9. am. .39 x 258 5a . < mussel to: new Em.- r _ 128 During this period most of the retail outlets were promot- ing the sale of beef. During the summer and fall of 1955, an industry-wide program was developed to stimulate beef sales. It is difficult to evaluate the results of such a program, but it is possible that demand for beef may have been increased by the special promotions that were carried on at this time. In the scatter diagram shown in Figure XIV, there are two weeks during which beef purchases were extremely large. These occurred during the weeks ending larch 21 and Kay 2, 1955. During these weeks beef prices dropped below 55 cents per pound and average purchases per family climbed to approxi- mately four pounds. Extensive promotion of beef cuts at special prices apparently was responsible for these large purchases. he pattern of beef purchases may have been affected by the supplies and prices of competing meats. The influence of these factors was taken into consideration in a regression analysis. During most of the two—year period for which panel data are available, large supplies of pork were available (Table 16). During the spring and early summer of 1955, pork production began to contract, as a cyclical reduction in hog numbers began to be reflected in market supplies. This down- vard adjustnent in pork supplies happened to coincide with 129 the rise in beef supplies. Nevertheless, total supplies of red meats were increasing from mid-1952 until mid-1955 C.) (Table 16). Results of Regression Analyses July 1951 to December 1952. The first equation to be fitted mathematically was of the type described in Chapter V, page 116. The weekly average purchases of beef were ex- pressed as a linear function of the price of beef, the prices of competing meats, and family income. Observations for the weeks in which the major holidays (Thank giving, Christmas, and Easter) occurred were omitted. A four-week moving aver- age was used. as the measure of current income of panel familieS. The simple correlation between the quantity of beef and the price was 0.61 (Table 17). The simple correlations be- tween the quantity of beef and the price of competing meats were all relatively small, ranging from -.15 for the price of poultry to .55 for sausage. A relatively high correlation was observed between quantity of beef and income, and there was a high intercorrelation with the price of beef. With the exception of a correlation of .52 between the price of beef and the price of sausage, the correlations between pairs of prices were relatively small. 150 TABLE 17 SIKPLE CORRELATICNS SETUZLH PAIRS CF VARIABLES, BEEF EQUATIOH, JULY 1951 TO DECEIBER 1952 3 Variables Variable _ \r X1 X2 X3 X4 A5 X6 Q or beef, Y -0608 “.172 -0326 -0155' .317 .519 1 P or beef, X1 .042 .521 .108 “.312 “.599 P of pork, X2 .551 .079 .280 -.571 P of sausage, X5 .005 -.l29 -.402 P of poultry, X4 -.001 -.295 P of fish, X5 -.018 Inc one , X6 Fitting the equation by least squares procedures produced araultiple correlation coefficient of .71 with a standard error Of estimate of .16 pounds and a mean of 2.54 pounds. The pre- diction.equation was (1.1) Y1 = 5.9547 - .0588Xl - .0096X2 + JXT77X — .0024X4 + .0140X5 + .0094X6. The regression co- 5 efficient for X2, price of pork, was negative and non-signi- fiCEUQt. .5 priori reasoning suggests that pork is competitive Witfli beef; therefore, the Sign of the coefficient would be eKipected to be positive. The coefficient for X4, the price or Poultry, was also negative; however, no strong relation— ship was expected between the price of poultry meat and beef purchases. 151 The beta coefficients and their standard deviations were computed and t values were obtained on each of the co- efficients. The results are summarized in Table 18. The tests of significance indicate that only the price of beef had a strong influence on the quantity of beef purchased. It was rather suprising to find the price of fish having a significant effect on beef purchases, while non-significant influences were registered for the prices of pork, poultry, and sausage meats. The price elasticity of demand computed at the mean price and quantity was -l.1l. This indicated that a 1 per- cent change in the price of beef was associated with a change in the Opposite direction of 1.11 percent in quantity of beef purchased by families in the M.S.C. Consumer Panel during the period studied. Stated in absolute terms, the regression co- efficient indicated that a change of 5 cents per pound in the price of beef was associated with a change in the opposite direction of .19 pounds in weekly average purchases of beef by Lansing families. The residuals were computed for the demand equation de- scribed above (Figure XV). As might be expected from an equa- tion having a R of .71, there are quite large residuals for some weeks. This suggests several possible problems. One is I ..'¢(1l.llil'.l. ‘ il. 152 that additional explanatory variables are needed to account for the week-to-week changes in beef purchases. These vari- ables might include some measurements of advertising activity and certain lagged relationships in purchases. TABLE 18 SUMMARY OF RLGRESSICN RESULTS, BEEF EQUATION (1.1) JULY 1951-DECEMBER 1952 Statistical Variables lbasure . . .. . . Price Price Price Price Price Income of of of of of' Beef Pork Sausage Poultry Fish x1 x2 x5 x4 x5 x6 Beta's -.4864 -.1565 .0697 -.0261 .2456 .2555 €73 .1516 .1054 .1099 .0917 .0976 .1270 t value* 5.70 l.49 .65 .05 2.50 1.85 deans 72.8 59.2 60.8 50.9 61.3 82.45 w With 55d,f,, t.05 = 1.997, t.01 ; 2.654. Based on table of t values in George W. Snedecor's, Statistical Methods, Iowa State College Press, Ames, Iowa, 4th ed., 1946, p.65. ho ommho>m haxoms map on pwwmwmquAflmHmomwwmmwwwowoMMEfiMthMWMWUMMMW .>K ohdmflm Nam. 33m? 3m. 5 00 av 0? on On mN ON 9 O. 0 mm D» NV up. mm KN new —JJ—~qd._d__da—_Adfifi_qq__qqqq__dd_41~4Nm_:__a.q_~dhad_~dqq—1jfi\:.~__.nq_fifiufi— com _ _ a ‘4 I . n :— om . . .22. o . 1., _ r . . a \ l « :L Lbs ”58.363 N. A: rl. IN... I i T L. o I ‘ 1., l_. .l LN. A_.:co:o:Um >269... 8a mem I m4<39mmm season. all]- a 154 Perhaps one of the more striking characteristics of the XV is the seasonal pattern. This is not (D residuals in Figur surprising since it was expected that there were some seasonal shifts in demand for beef. The residual pattern shows that beef purchases were low relative to price during the summer months and high during the fall and winter. This suggests that a demand shifter, closely related to these seasons, would explain some of the variability in the data. Temperature has been used by others as such as a demand shifter.2 When dealing with the meats, it is doubtful that tempera- tures below a certain critical level have much effect on de- mand. Weekly averages of mean daily temperatures in Lansing were plotted with the residuals in Figure XV. It appeared that some relationship existed, but it was difficult to deter- mine the critical level of temperature below which demand was not affected. In moving from summer to fall, it appears that somewhere around 60 to 65 degrees was the level at which de- mand increased as temperatures declined. Above 65 degrees, weekly purchases of beef seen to be inversely correlated with sharp temperature increases. This same temperature level seems 2 G. G. Quackenbush and J. D. Shaffer, "Consumer Purchases of Ice Cream for Home Use," Unpublished manuscript, Dept., of Agr. Econ., Michigan State College. See also: George M. Kuznets and R. L. Klein, A Statistical Analysis of the Domestic Demand for Lemons, 1921-41, Giannini Found., Agr. Econ. Rept. 84, 1945. 155 to be closely related with the decline in demand which be- gan in hay 1952. Part of the negative residuals immediately following Easter are probably due to the lag effects of ham purchases for the holiday. A significant correlation was found to exist between temperature and the beef residuals (Equation 1.1) for the warm season of the year. The period considered as part of the warm season included all of the weeks occurring in 1951 between July 1 and October 20, and in 1952 between April 27 and October 18. The temperature variable was constructed by subtracting 60 degrees from the weekly averages of mean daily temperatures in Lansing. All weekly averages of 60 degrees or less were assigned a value of zero based on an assumption that below this level temperature becomes unimportant as a demand shifter. The correlation between temperature and the beef residuals was -.58 with 41 weekly observations. A least squares regression of beef residuals on temperature, using the same data, produced a regression coefficient of -.0122. This indicates that a 10 degree increase in weekly average temperatures would decrease the quantity of beef purchased by the average Lansing family by .12 pounds. This is equiva— .1ent to about a 5 percent decline from the average of weekly purchases for the July 1951-December‘l952 period. 156 Similar correlations and regressions were computed using all the weeks in the July 1951-December 1952 period.5 Here again all weekly average temperatures of 60 degrees or less were given a value of.zero. The correlation between the tempera- ture variable and beef residuals was —.51 and the regression coefficient was -.0101. This coefficient was significant with a t ratio of 2.75. These results were reasonably consistent with the statistical results obtained for the warm season period. July 1952 to June 1955. As data became available a new series of analyses were made using observations for the period during which sizable changes occurred in the prices and pur- chases of beef. Data for the period July 1951 to June 1952 were not included since there is some question about the dis-~ turbance of normal relationships caused by price controls. A lS-week moving average of family income was used in place of the 4-week moving average used in the previous analysis. The principal reason for making this substitution was to eliminate the variability in the income series due to the non- uniform pattern of pay periods among families in the panel. A simple correlation of-«SV (Table 19) between the quan- tity and price of beef was obtained from this analysis as com- pared to the -.61 for the earlier period when only small changes were observed in prices. The correlations between income, 3 Major holiday weeks were omitted. 157 beef prices, and quantities were also substantially in- creased. The high correlation of -.95 between beef prices and income indicates that the multiple regression results may be unduly biased by the intercorrelation among the ex- planatory variables.4 An intercorrelation of .546 is also noted between the price of sausage and the price of beef. This relationship is not surprising considering the fact that beef is one of the major components of sausage. _There was also an intercorrelation of 0.55 between the price of sausage and income. The high correlation between the income variable, the price of beef, and the price of sausage was believed to be partly a chance relationship. Although the rise in income probably had some influence on demand for beef and other meats, the steady increase in beef supplies and the constant downward pressure on beef prices are closely associated with the rise in incomes which began in the fall of 1952 and ex- tended through the first half of 955. When data are avail- able over a longer period of time, during which beef supplies turn downward, he correlation between income and beef prices dwould be expected to decline substantially.‘ 4 Karl A. Fox and James F. Cooney, Jr., affects of Intercorrelation upon Iultiple Correlation and Regression. Agr. hkt. Ser., U.S.Dept.of Agr., Processed Report, 1954. 28 pp. S ILIPLE TABLE 19 158 CGRRELATIONS sswwasw PAIRS cs VARIABLES, , JLY 1952-JUNE 1955. BEEF EQUATION Variables Variables X1 X2 X5 (X4 X5 X6 Q of beef, Y1 -.874 .098 -.509 .555 .225 .815 P of beef, X1 -.116 .546 -.486 -.216 -.952 P of pork, X2 .592 .050 .257 -.021 P of sausage, X5 -.~84 -.256 -.649 P of poultry, X4 .139 .480 P of fish, x5 .216 Income, X6 Three multiple regression equations were fitted to the data for the period July 1952 to June 1955. The first, (1.2), was the basic demand equation similar in structure to equation (1.1). The second equation, (1.5), was the same as (1.2), but the income variable was omitted in an attempt to avoid part of the intercorrelation problem mentioned above. Equation (1.4) involved a substitution of the quantity of fish purchased in place of the price of fish as the X5 variable. Equation (1.5) is the same as (1.5) except that temperature was added as X7.v The prediction equations multiple correlation coefficients, and the standard errors of estimates are shown on the follow- ing page. 159 - .0165X - .0181K 11.2818 - .0706Kl- .0015X .004 - .0272K + 5X5 5 '6 .1755 Sy.x (1.5)Y1 - 7.5755 - .0551Xl ¢ .0008X2 - .0084X5 - .0189X4 + .oouox5 sybx 2 .1763 (1.4)Y1 _ 8.8416 - .0562kl - .001042 - .0117A3 - .0244h4 'b 0422].){5 R .886 .1758 Sy.x + .0142X + .0018X - .0079X 1 + .0012X5 - .0255X 4 7 R a .909 Sy.x _ .1559 Although the variations in these equations had almost no effect on the multiple correlation coefficient, there is a striking differencetmflwmxnlthe regression coefficients for X1, the price of beef. When the income variable was included in the equation, a regression coefficient of .0706 was obtained SULJI .3133? 017 QW‘I’VDTW(_‘1 aIr‘\‘T .LxJJ‘JJ. L111.J.D U‘.. JULY 1952 - JUHfi 1955 RESULTS, 3332 EQUATION page 159. . . , . , D1. . Statistical Price Price Price Price .iice Family Veasure of of of of 01 Income ” Beef Pork Sausage Poultry Fish X X X X X l 2 X5 4 5 6 Equation (1.2) Beta's -l.1945 -.0161 -.0955 -.l205 .0447 -.5569 0'5 .2608 .0041 .1151 .0814 .0796 .2744- t-ValllCS 4.058 017 .83 1.48 056 1.25 Equation (1.5) Beta'S -0899]. .0084 “.0481 ”.1264 .0411 45. .1029 .0956 .1088 .0827 .0209 t-values 8.74 .09 .44 1.55 .51 Equation (1.4) . a Beta's -.9515 -.0197 -.0670 -.1651 -.1105 «5 .1112 .005 .1010 .0375 .032 t-Values 8.56 .21 .66 1086 1.24: Equation Beta's -.7516 .1659 .0105 -.0526 .0128 -.5207 ti .1010 .0055 .0032 .0770 .0729 .0057 t-values 7.59 1.72 .106 .695 .176 5.42 a Quantity of fish substituted as X5 in place of price of fish. ‘ b Temperature variable, noted as X7 in Equation (1.5), for X1, as compared to .0551 when income was omitted. The corresponding elasticities at the mean values of price 035.9 cents) and quantity (5.01 pounds) were -l.50 and -l.l5. IhHDpping the income variable had relatively little effect on 'mie other regression coefficients. The coeff'cient for pork, 3&3, changed from negative to positive, but neither coefficient was: significant (Table 20). ubstituting quantity for the price of fish, as X5 in the equation, yielded a slight change in the importance of fisli as a factor influencing beef purchases (Table 20). The .increase in fish purchases, during the Lenten period, was quit;e apparent in the quantity series but was hardly discern- 3b1£3 in the price series. Consequent y, he shift in demand froni red meats and towards fish was probably better repre- sentued.by the quantity series on fish than by the price series. Thiss, plus the fact that the average price of fish is relative- 1y Luastable, were the principal reasons that could be offered for the shift in relationships. Although the t value for the qu£uatity of fish was much larger than for price of fish, it “3&3 still not large enough to be significant at the 5 percent level, Adding temperature as a variable in equation (1.5) pro- C1Heedsome interesting results. The coefficient for X1, the Price of beef, declined to -.0444. The price elasticity based On this coefficient was -.94 at the mean values of price and 142 quantity. The coefficient for K2, the price of pork, became significant at the 10 percent level. The cross elasticity with the price of pork was .51 measured at the means. The coefficients for the prices of sausage, poultry, and fish again were non-significant; however, the temperature variable 'was highly significant. The temperature coefficient indicated that an increase of 10 degrees in mean temperature was associ- ated with a decrease of .255 pounds (about 8 percent) in beef purchases per family. The residuals for equation (1.5) were computed and plotted graphically. No significant seasonal variation was discern- ible. This indicated that the temperature variable, used as I a "demand shifter,‘ had accounted for most of the shift in demand between the warm and cool seasons. A question arose with regard to the possible autocorrela- tion of the residuals for the above equations. It was pointed out earlier that one of the underlying assumptions of ordinary least squares regression is that the residual errors are inde— pendent of one another. The residuals for equation (1.5) were tested using the ratio of the mean square successive differ- . a 5 . ‘52 ences to tne variance. The ratio,———Tr—-, was equal to 1.97, ' s which was non-significant aCcording to the distribution table 5 There is some disagreement over the validity of this test, however, some of the other tests are also question- able. See Lawrence R. Klein, Econometrics, Row, Peterson and 00., Evanston, Ill.,1955, p.89. See also: B.L.Anderson, ‘The Problems of Autocorrelation in Regression Analysis," Jour.Amer.Stat.Assoc.,, Vol.49,1954, p.117. 143 tabulated by Hart and von Neumann? This test was followed by a simple correlation of the residuals where X1 - residuals for time period t and X2 = he residuals for time period t - 1. The correlation coefficient was .056, which was very low and non-significant. A tentative conclusion was that autocorrelation of residuals was not a serious problem; how- ever, further analysis using other methods of testing may be desirable. In cases where a high degree of autocorrelation existed there have been attempts to minimize the difficulty by con- verting the raw data into first-differenceS.7 An attempt to use first-differences on beef purchase data yielded a rather low multiple correlation coefficient. An R of .55 was obtain- ed when the quantity of beef was taken as a linear function of the price of beef and the price of pork. Weekly observations for the July 1952-June 1953 period were used. Holiday weeks were omitted. The simple correlation between the quantity and price, in first-differences, was -.25. The regression coefficient for the price of beef was -.0408 with a T value of 1.84. The regression coefficient for the price of pork 6 B. I. Hart, "Significance Levels for the Ratio of the Mean Square Successive Difference to the Variance," Annals of Math. Stat., Vol.15, 1942, pp.445-47. 7 D. Cochrane and G.H.Orcutt, "Application of Least Squares Regression to Relationships Containing Auto- Correlated Error Terms," Jour.Amer.Stat.Assoc., 44:32-61, 1949. 144 was .0249 with a t value of 1.68. (With 49 observations a t value of 2.0 is significant at the 5 percent level.) Based on this first difference analysis, the price elasticity of demand for beef was -.75 at the mean quantity and price. The cross elasticity with the price of pork was .21. Summary Which of the regression equations produced the "best" estimates of the structural relationships among the studied variables? An answer to this question must be based partly on Subjective considerations. The results of the first equa- tion (1.1) were subject to criticism because the period July 1951 to December 1952, included the period during which beef prices were subject to government control. Equation (1.2) appeared to be unduly disturbed by the high intercorrelations existing among the variables incomes, the price of beef, and the quantity of beef. The results of equations (1.3) and (1.4) appear to be superior to either of the first two equa- tions. In some respects equation (1.5) may have yielded even better results than (1.3) or (1.4). In (1.5) an attempt was made to provide a variable to act as a "demand shifter" be- tween the warm and cool seasons of the year. The multiple correlation coefficient was increased to .909 and the rela- tionship between the price of pork and beef purchases seemed more reasonable than the results indicated in the previous 145 equations. The regression coefficient for the price of beef was lower in equation (1.5) than in any other equation except (1.1), but appeared to be reasonable. The lower esti- mate indicates that a decline of 5 cents per pound in the weekly average price of beef was associated with an increase of .22 pounds in average beef purchases per family. The higher estimate indicates an increase in purchases of .28 pounds per family with each 5 cent decline in price. The coefficient for the price of pork, in equation (1.5),indi- cates that a 5 cent rise in the price of pork was associated vith an increase of .07 pounds in beef purchases." The Demand for Selected Retail Cuts From the ten sub-groups of retail beef cuts listed in the panel diary, three were selected for special study. These are ground beef, roasts, and steaks. During the period July 1952 to June 1955, these three items made up 88 percent of the total quantity of beef purchased at retail. Ground beef was the largest item with weekly average purchases of 1.09 pounds per family. This compares with .88 pounds for roasts and .68 pounds for steak (Table 21). Ground beef is by far the largest traffic item with about 60 percent of the families buying each week. Steaks attracted about 55 percent of the families each week while 25 percent purchased roasts. These percenta5es and relative quantities vary with prices as will be pointed out below. TAT LE 21 n3LA1I7 I PCTTAKC; ‘F SLLECTH RETAIL CUTS CFflBZLF M.S. C. Cc”SJ1nR FAILL, JULY 1952-JUII" 1953“ Pounds Percent Expendi- Price Retail Cuts per Buying ture per per Family Family Pound dollars cents Ground beef '1.09 59 0.58 53.6 Roasts .88 26 0.57 64.9 Steaks . .68 37 0.56 82.1 Stewin5 and boiling .13 7 0.06 46.6 All others .23 -- 0.15 65.2 Total beef 3.01 -- 1.92 63.9 ‘.I '.\ Holiday weeks, Thanks givings, Christmas and Easter omitted in computin5 avera5es Consumer responses to declining beef prices were some- what different for the three sub-groups of retail beef cuts. The nature of these differences are revealed in Table 22. In order to avoid disturbances due to seasonal variations in demand, purchase data for the 2d quarter of 1952 were com- pared to correspondin5 data for the 2d quarter of 1953. During this period of time, beef prices declined more than 20 percent. The price of ground beef fell from 66 to 45 147 cents per pound. Consumers responded by increasin; average weekly purchases from .96 pounds per family to 1.12 pounds. This response was insufficient to prevent expenditures from declinin5 from 64 cents per family per week to 51 cents. This indicates that the "price elasticity of demand" for ground beef was inelastic within this ran5e of observations. It is also interesting to note that the percent of families buyin5 5round beef in any 5ivcn week was practically unchanged between the two periods. Between these same periods the avera5e price of beef roasts declined about 18 cents from 73 cents per pound to 55 cents. Consumers responded by increasing their purchases from .75 pounds to 1.02 pounds per family per week. Part of this increase in purchases was due to a 5reater proportion of the families buyin5 each week.‘ Expenditures per family for roasts increased from 54 to 56 cents-per week, indicatin5 that the price elasticity of demand was slightly elastic. The greatest response to price changes occurred on steaks, which was not particularly surprisin5. The avera5e price of steak declined about 17 cents, from 91 cents per pound to 74 cents. Purchases nearly doubled in response to the price de- cline. The percent of families buying steaks in any 5iven week rose from 29 to 42 percent. Weekly eXpenditures for steak increased from 42 cents per family to 62 cents indicat- ing a hi5h1y elastic demand. CCILPARISOTT OF PU‘RCI‘TA 0F fi"-P 7f", 1 T "' DJJJJJH 1'1 . TABLfi (7‘ ”‘0 min 53135.) L; S. C. 22 2d QUARTER OF 1952 and 148 Roman RETAIL eras CONSUJER PAEHL 2d QU‘RTER CF Retail Cut weekly Average and Period Price Pounds Percent EXpendi- per per 3uyin5 ture per Pound Family Family cents cents Ground beef: 2d quarter 1952 66.4 .96 58.3 63.7 2d quarter 1953 45.4 1.12 58.0 50.9 Roasts: 2d quarter 1952 72.5 .75 .22.9 54.4 2d quarter 1953 54.7 1.03 29.8 56.3 Steaks: 2d quarter 1952 90.9 .46 28.5 41.8 2d quarter 1953 74.0 .84 41.7 62.2 Total 3 outs 2d quarter 1952 75.7 2.17 -- 159.9 2d quarter 1953 56.6 2.99 -- 169.4 149 It was interesting to observe the shifts which occurred in relative quantities purchased of these three major sub- groups of beef cuts (Table 23). In percentage terms there was a sizable shift from ground beef to steaks, which poses an interesting question with regard to the relative prOpor- tions of the different retail cuts. Individual beef carcasses of the same grade and weight, broken into retail cuts by the same cutting procedure, will normally yield nearly fixed prOportions of the different cuts. Therefore, the variation in the relative proportions of hamburger and steak sold in Lansing in 1952, as compared to 1953, might be due to two factors: (1) a change in the composition of Wholesale beef cuts shipped into the Lansing market; (2) changes in cutting procedures in the retail stores so as to merchandise a higher proportion of the beef as steaks. It is entirely possible that the composition of whole- sale beef supplies received in Lansing retail outlets did change between the second quarter of 1952 and the second quarter of 1953. During the late spring and summer of 1953 an unusually large percentage of the total beef supply for the U.S. graded TABLE 23 AJT LXCD DITURES FOR GRCTI. D 078 T .233 PANEL, 2d HART R d C'TmLIDJ CF 1955 CIAACHS IN THE RELATIVE QUANTITIES 1341.1? ’ A-{k’ASTS .A-.D Sr ‘31, 7:3 , Milli. b QC . 'T‘ w, L 0? 1952 COLPARID T0 Quantities Expenditures Retail Gut 1952 1955 1952 1953 pct. pct. pct. pot} Ground beef 44 37 40 30 Roasts 35 34 34 33 Steaks 21 28 26 37 Total_ . 100 100 100 100 in the choice and prime grades.8 This was the result of record numbers of cattle being placed in feed lots the .1" previous fall. As these large supplies 01 the better grades of cattle moved into trade channels, the spread in prices be- came extremely narrow (Figure XIII). It is likely that a larger proportion of choice beef was sold in the retail stores of Lansing as a result of the overall adjustment in supplies 8 According to statistics from market reports issued by the Livestock Division of the Agricultural Iarketing Service, U.S.Dept.,of Agriculture, 71 percent of the beef steers sold out of the first hands for slaughter at Omaha, Chicago, and Sioux City, during the second quarter 1953, were choice and prime. This compares with 62 percent for the second quarter of 1952. 151 and prices. It is also reasonable to expect that a relatively larger proportion of a choice carcass can be merchandised as steaks as compared to lower grades of beef carcasses. Another explanation of the shift toward more steak and less ground beef is that a larger proportion of the wholesale cuts shipped into the Lansing area in 1953 may have been hind- quarters as compared to 1952. The hindquarter of beef is the source of most of he steaks in the carcass. Yith the sharp decline in beef prices and a general broadening of the market for beef cuts, an increasing prOportion of the front quarters of beef may have been channeled to lower income areas of the country. These lower priced cuts probably could be mer- chandised to a greater advantage there than in the Lansing market where the average level of incomes is relatively high. It is difficult to estimate the extent to which retailers were able to change their cutting procedures so as to increase the proportion of steaks in relation to ground beef. However, the use of steak machines to tenderize cuts makes it possible to move a larger proportion of a beef carcass as steaks rather than roasts or ground beef. Regression analyses of price-quantity relationships for the three groups of retail beef cuts further substantiated some of the general conclusions indicated in Table 22. This analysis was limited to simple two-variable correlations and regressions, treating each group of retail cuts separately. 152 Weekly observations for the period July 1, 1052 to June 1053 were used with major holiday weeks omitted. The results of these computations are briefly summarized .t! .1. in Figure XVI. Here it can be seen that family purchases 0 ground beef increased an average of only .0031 pounds for *‘3 each one cent decline in the p ice. The low correlation of —.32 between prices and quantities indicated that the rela- ‘ tionship was rather weak. Purchases of roasts tended to in- crease ab ut .0174 pounds for each one cent decline in average price. The correlation between prices and quantities was -.7G ['1‘ which indicated a fairly strong relationship. consumers to price reductions on steak was the greatest among the three groups of cuts studied. Family purchases tended to increase by .0207 pounds for each one cent decline in steak prices. A high correlation of -.90 was observed between prices and quantities. The preliuiaary analysis of data for retail beef cuts J-‘._ offers at least tentative evidence that there are signiiicant differences in the dekand characteristics among the various Ground beef apparently is a staple neat iteu in the food budgets of uost fanilies. The small increase in purchases associated with rather large price declines, can be rational- ’3 n terms Oi both an incoge and substitution effect. For I“ zed H- ine response of .-—.—-.——-——_-._’——-» a-i 153 l PCUWJ? per Fa 1Hy I.3- ‘ -‘ \ ' _. Ll— GROUND BEEF \ =—.003| r = —.32 .4r- “ {\fi 1 I 1 1 1 l m haxoos nooauon manageapaaom .HHH>x ohsmflm 250.”. 3a macoo 2 on no om on On Jw___:injjh__fi_;:fi:::_daéw i. 1 Hi 6 Q d nun“. d 7 no a H II 0 W? 98 - O LO.N T1 .0 < 4.x A. 1 rl . O O . OXX d l I. ‘ O Q .fl 60 “X C X l T o no. x 0 ca 1 l . a lm.~ I. . . e. XXXX X XX I. I o o .- one a 1 I. O O .a x L I. X XX I. o x x o x lo.n to; t. .32 a . h 0 CN. _0 L t c U .Nmm a X 1 Cl :0: am. Nnm. x 3E5... 3a to; new Jam. 0 moaned _ Kore detailed plotting of the pork purchases as related to prices revealed that shifts in the demand function may have occurred between summer and fall periods. This would be logical if there were a significant decline in de1and for pork during the summer months. The sharp rise in prices in the spring and early summer of 1955 was the beginning of a period of high pork prices which extended into 1954 Data available for this study cover only the beginning of this period; however, it can be observed in Fig ure XVIII that consumers responded rather quickly to the rise in prices. Quantities purchased declined far below those L, bi of other weeks in the two year period covered by 1is study. These adjustments in pork purchases and prices occurred during a period in which beef supplies had soared to record levels causing the retail price of beef to fall sharply. The sudden rise in pork price, v.hile beef prices were making th r down- ward adjustnent, suggests that many consumers are reluctant to substitute beef for pork. Purchase data for both beef and pork give indications that substitutabili ity between these two meats is far from perfect. Analysis of data over a longer period wiiich includes the last half of 1955 and 1954 will provide a more adequate basis for testing some of these re- lationships. 164 Results of Regressien AnalySis July 1951-December 1952. Following similar procedures used in analyzins the demand for beef, a single equation model s was fitted to the pork purchase data for the 13 monfiiperiod beginning July 1, 1951. The week-y average purchases of pork were taken as a function of the prices of beef, pork, sausage, 1 ion and fanily income. Data V Pb poultry, i or the major holiday weeks were omitted. I The simple correlation between the quantity of pork and he price of pork was 0.59. The simple correlations between the quantity of pork and the prices of "competing" meats and family income were all non-significant at the 5 percent level. The correlation with the price of beef was .21 which approach— es significance at the 5 percent level. ( Tith 70 d.f. a cor- relation coefficient of .25 is significant at the 5 percent level.) form. Cl' 0 ('1 :_J (D p. (‘1' p P- H ._J :3 PS H0 (‘1‘ r i 1 5 C1. Ho 0 The equation was fitted #3 The mathenatical fit by least squares yielded an R of .68 and a standard error of estimate of .225 pounds; the mean ‘was 2.d7 pounds. The multiple regression coefficient was not a great deal larger than the simple correlation between the quantity and price of pork taken alone. The prediction equation was as follows: (2.1)‘r2 - .5730 + .omaxl- .osesx2 + .0515}:5 + .oz-jzssx4 + .OOQlXS+~ .OOSQXG The signs of all the coefficients were consistent with expec- tations based upon a priori reasoning. Casual inspection indicates that the price of pork, X2, is the most important variable in the equation explaining weekly pork purchases. To further compare the relative importance of these variables, beta coefficients were computed and tested for significance. (See Table 25.) Only the price of pork was significant at or above the 5 percent level. Income was the least important variable in the equation. The price elasticity of demand for pork measured at the mean was -l.55, indicating an elastic demand. The largest cross elasticity of demand was with sausage meats, with beef and poultry following in that order of importance. These above computations provide tentative estimates of the structural relationships between the average weekly pork purchases by Lansing families and the explanatory variables in the regression equation. The "low" multiple regression coefficient suggests, however, that all the explanatory vari- ables have not been taken into account. It is also possible that some function other than a linear function in absolute .arithmetic values could better fit the data. An examination of the pattern of residuals provides a basis for judging the adequacy of the single equation model used in this problem (Figure XIX). 166 .mQqueq Ga mogpwmogep Megan @336 secs .Ho 3.3998 5032» on penmmfioo 3.3 «83.380 View E09.“ mawsvammm .xHx magmas mom. mxeog .mm_ Om me ov me OK on ON m_ o. m um we NV hm mm hm :_::_:_=:_____:_::_::.J__:_::_::_._:_:_:_:JJ‘jJZZ—«qdj‘onum . : as S —\I a a .9 — ‘f _: _ I ~ . 10w ’ ~ 1"; o ._ ) : a , z , :r 5: a __ , \ "x On . a no oSaSanoh I—x S. vied - .093an me. N .N.» >.::ou Lea mugged L 16? The residuals for the pork equation (2.1) show a season- al pattern similar to that observed for beef. Temperature 1 For the warm season was correlated with the pork residuals. weeks a correlation of -.55 existed. For the entire July 1951-December 1952 period, the correlation was -.51. The re- gression coefficients for these two groupings of weekly data were almost the same where pork residuals were expressed as a linear function of the temperature variable. For the warm season weeks, the regression coefficient was -.0149 and for the entire period (July 1951-December 1952), the co-efficient was 0.157. The t ratio for this latter coefficient was 2.75 which was highly significant. TABLE 25 SWILARY OF REGLBSSION RESULTS, PORK EQUATION JULY 1951-DECLIDER 1952 Variables statistical i§ice i§ice :gice :giee :gice Measure Beef Pork Sausage Poultry Fish Income X1 X2 X5 - X4 Beta's .1368 -.6921 .2145 .1370 .1192 .0727 ‘3 .1562 .1091 .1158 .0950 .1010 .5500 t value 1.37 6.54 1.88 1.97 1.18 .55 l The temperature variable was coded as actual mean temperature in Lansing-60 with all temperatures of less than 60 given a value of zero. 168 July 1%) Jun e 1955. After data became available for the first half of 1955, a separate re5ression analysis was made for the period July 1952 to June 1955. This was a period of substantial price chan5 es for both pork and beef. The same basic demand equation was used as described in the preceding section, with one exception. The four-week movin5 avera5' e of fa1r1ily income was re} laced by a thirteen week moving avera5e. hajor holiday wee ks a5 ain were omitted from the series of observations. The prediction equation fitted to the data was as follows: (2.2) Y0 = 5.0542,! + oOOSSXl " .0560}:2 - .0016}: + oOO45X 5 4* 5 6. An R of .82 and a standard error of es tinate of .514 were as- sociated with this equation. The mean quantity was 2.17 pounds. The si5ns of two variables, sausa5e as X5 and income as XG’ were chan5ed from those obtained in equation (2.1). However, neither of these coefficients was si5nificant (Table 26). The coefficient for X2, the price of pork, was highly si5nificant and was approximately the same as the p1evious estL1ate in equation (2.1). T1e coeii icient for the price of beef was considerably smaller an‘ the level of sig- nificance was reduced from the previous estimate. ClDCO the inccne variable proved to be non-815nificant and was intercorrelated with beef prices it was decided to ('3! drop it and compute a new prea diction equatior1. (2.5) v = 5.2096 . .0120x - .054 4x2 . .0041x . .0057x *2 1 5 4 + .0110x5‘ The multiple R for this equation was .82, essen ially un- chan5ed from the previous equation. The si5n chan5 ed for sausa5e price, K5, but t1:e coefficient was still non—515ni- f5 ficant. The coefficient for the price of be21 ch~15aisub- [En stantially and became si5n icant at the 1 percent level. ‘ he coeff H. P- C 153 ent for the price of pork, X5, was practically un han5ed from equation (2.2) and (2.1). Usin5 the coefficients for the last equation the price elasticity of deaand for pork would be estimated at approxi- mately-1.E 5 at the mean values of price and quantity. The 0 cross elast1city of demand with the price of beef was .55, indicatin5 that a 10 percent chan5e in the price of beef was a85001atcd with a 5. 5 percent cl1an:5e in pork p1rchases in the same direction. This a5ain applies to chan5es at the ;uean values,assumin5 prices of other meats are held constant. The apparent relationship between temperature and pork ;purchases was mentioned earlier. When added to a multiple :re5ression, temperature proved to be si5nificant in explain— 1 lenand for pork Equation (2.4) 1 i15 the seasonal snifts in siLanarized the results of this analys1s: 170 1.5553 + .0241X1 - .0442X2 + .0114K4 + .0115X5 1 .0091X6 - .0166X8 (2.4) Y2 = he multiple re5ression coefficient was .91, the hi5hest ob- tained for any of the pork equations. The standard error of estimate was .164 pounds and the mean quantity was 2.17. The temperature coefficient indicated that a 10 degree increase in the weekly average of mean daily temperatures, above the 60 degree level, was associated with a decrease of .186 pounds (about 7% percent) in pork purchases per family. There were also some noticeable changes in some of the other re5res- sion coefficients. The coefficient for X2, the price of pork, was smaller han any of the previous estimates from other equations. The coefficient for X1, the price of beef, was the largest obtained from any of the equations. Althou5h _there were some changes in the coefficients for prices of other COnpeting meats, none approached si5nificance (Table 26). heasured at the mean values, the price elasticity for pork was -l.25 and the cross elasticity with the price of beef was .71. In absolute terms, an increase of 5 cents per pound in the price of pork was associated with a decrease of .22 pounds in average weekly pork purchases per family. A decrease of 5 Germs per pound in the price of beef was associated with a decrease of .l2 pounds in average weekly pork purchases per family. 171 The results from the pork regression analyses Summary . were more consistent than those for beef. The coefficients for X2, the price of pork, ranged from —.0444 to -.O565 with the lower estimate coming from equation (2.4) where tempera- ture was used as a demand shifter. Intuitively, the -.0444 coefficient seemed to be superior to the higher estimates which may be biased by seasonal shifts in demand. TABLE 26 SUELARY OF REGRESSION RESTLTS, PORK EQU TICNS, JULY 1952-JUNE 1955 Variables Statistical Price Price Price Price Price measure of of of of of Family Beef Pork Sausa5e Poultry Fish kl X2 X3 X4 X5 X6 Equation (2.2) Beta's .1160 -.7512 -.0015 .056l .1450 -.5006 .5225 .1164 .1599 .1006 .0984 .5595 t values .56 6.45 .08 .56 1.47 .89 Equation (2.5) Beta's .5794 -.7293 .0292 .0509 .1420 a? .1262 .1149 .1555 .1014 .0992 t values 5.01 6.55 .22 .51 1.45 Equation (2.4) Beta's .5080 -.5951 .0911 .0956 .1174 -.2814* d? .1529 .1246 .1295 .1007 .955 .1226 t values 5.82 4.76 .65 .95 1.25 2.50 Temperature variable, noted as X7 in Equation (2.4) 172 The coefficient for X1, the price of beef, ranged from .0055 to .0241, but the lower estimate was rejected because of the disturbing influence of the intercorrelation between beef prices and the income variable in equation (2.2). There- fore, the reasonable range of estimates for X1 lies between .0180 and .0241 with some preference for the higher coeffi- cient where seasonal shifts in demand were partially accounted for in the regression equation. Although none of the coefficients for sausage or poultry prices was significant at the 5 percent level, it appears reasonable that mild substitutability does-net exist between pork and these items.2 In most instances this was indicated in the regression results. The use of a temperature variable provided a reasonably satisfactory procedure for measuring the influence of chang- ing demand between the warm and cool seasons of the year. It was surprising to find the temperature coefficient for pork was less than the corresponding coefficient in the beef equa- tion. This relationship should be tested by analyzing data over a longer time period where several seasons can be ob- served. In a recent study R. C. Smith found some evidence that fryers and pork loins were competitive. See Factors q 0 ~‘ _ 1 1’) .. o a o .""“t—"""‘—r' Afiecting Consumer Purcnases 01 Fry1n; Cnickens, bniverS1ty of Delaware, Agr. Exp.8ta. Tech. Bul.298, Newark, Delaware 1955, p.7. 175 Demand for Selected Retail Cuts Preliminary analysis of panel data on retail pork cuts indicates that the demand for pork chops and ham is elastic with reSpect to price-quantity relationships, while the de- mand for bacon is slightly inelastic. These generalizations apply to the period July 1952 to June 1955 without fully ac- counting for the effects of changing prices for beef or possible seasonal shifts in demand. Further refinements in methods of measurement will be required to obtain reliable estimates for the true structural price-quantity relationships. Weekly average price, percent of fauilies buying, uanti- ties, and expenditures per family for several pork cuts were plotted on semi-logarithmic paper to study the interrelation- ships anong these variables. Prices and expenditures for pork chOps were inversely related indicating an elastic de- mand. A siuple correlation of 0.76 existed between the weekly average prices of chops and quantities purchased per family. This was based on 49 weekly observations, with the major holi- days being omitted. A two variable regression with quanti- ties purchased expressed as a function of weekly average prices yielded a regression coefficient of —.0059. The price elasticity was -1.55 at the nean values of 74.8 cents per pound for price and .552 pounds per fahily for quantity. From the graphs, mentioned above, there appeared to be a 174 significant inverse relationship between the prices of pork chops and the percent of fanilies buying. During the second quarter of 1955, when prices rose from 70 cents per pound to 89 cents, the percent of fanilies buyiné declined from about 27 percent to 18 percent. These observations provide the basis for a tentative conclusion that the denand for pork chops was elastic for the period studied. Graphical exanination of purchase data for fresh pork sausage indicated that prices for this commodity did not in— crease in preportion to prices for other fresh pork items during the first half of 1955. hevertheless, expenditures for sausage and quantities purchased varied considerably and inversely with prices. Does this mean, then, that the de- mand for pork sausage is hiéhlj elastic? Probably not in this case, since there is reason to believe that the demand for fresh fat pork cuts declines significantly fron winter to summer. The period of large supplies and low prices for these items occurs during the winter when demand is seasonal— ly strong. As summer approaches, demand declines seasonally, while at the same time, total pork supplies decline and ’3 prices tend to rise. nder these conditions iittin~ a line of average relationship to prices and quantities observed in C the market over tile, is lidely to yield a denand function having nuch 5 eater elasticftv than the true denand function. 175 This sane veneral relationship, described Eor pork sausage, appears to be true ior pork roasts. For both pork sausage and roasts, the percent of wilies buying varies widely by seasons of the year. DHP1NQ the win- ter, about 17 to 20 percent of the fanilies bouéht sausage in any given week. The percentage declined to around 12 during ausaoe purchased is closely re- (.9 t1e Sizu1er. The quantity of lated to the pe rce:1t buyiig with ‘he &VGT&QG purchase beinb slightly over one pound. The percentage of families buying pork roasts was around 10 percent durin51e Vi- inter, 1ihile only 5 to 6 percent purchased during the summer. Tiese vari- est that sizable ations in percent of fanilies buying su”* Qt.) shifts in demand may have occurred. Further research will be 1 . reouired to verify t -is. Ham and bacon are the two most important cured pork items with each having schewhat different demand characteristics. Bacon is prinarily a breakfast item; as such, it is doubtful that other heats are readily substituuxifor it in the averare household. Pork sausabe and han are probably substitutable for bacon to a limited extent.‘ Under these conditions it would be expected that the dehand for bacon might be inelastic. The I" purchase data fro1 the 3.3.o. 0011su;i er Panel seexs to verify this hypot1et ical relationship. During the period Jul; 135 2- June 1955, expenditures for bacon were positiv 013 related ‘0 weekly average prices. A simple correlation of -.50 existed between weekly average quantity purchased per fanily and price. A linear regression of quantity as a function of l D “n coefficient 0i \ i C) 0) weekly average prices yielde‘ a TGQTG C. -.OOGl. At tne mean value of price (55.» cents) and quan— tity (.452 pounds) the price elasticity of demand was -.75. The price elasticity for ham for the 1952—55 period appeared to be elastic. Due to wide weekly fluctuations in purchases it is difficult to deternine the true structural relationships frou a sinple function involving only price and quantity in the current week. Tron the graphs of the data, it appears that weekly expenditures per faiily are negatively related to price. This SH;QGStS that the price elasticity of denand is greater than one. A linea' regres- sion with weekly ban purchases as a function of price indi- cated an elasticity of-l.36 at the mean values of these variables. The mean price was 70.6 cents and the mean quantity was .45 pounds per family per week for the July 1952-June 1955 period. The simple correlation between weekly average prices and quantities of ham :urchased was -.46. It is interesting to note that the percent of families buying nan-remained relatively constant throughout the period of rising prices in the first half of 1955. This would sug- éest that fanilies were continuing to buy ham as frequently as previously, but the size of purchase was declining as prices increased. 177 In summary, it appears that the denand for some of the fat pork cuts may shift considerably from winter to summer. .1. .1 b3 Lay provide estinates of pr ce H° If this is true, ;.'.2.ar1;et da elasticity of dcnand that are biased upward because the seasonal Shifts 'n demand tend to accen uate the normal re— sponse to seasonal price changes. This would also bias some of the denand elasticities for "all pork" discussed earlier in this Chapter. Further research with data over a longer period will be required to deternine the "true" structural relationships between prices and quantities of retail pork cuts. CHATTER VIII TLE DEhAUD 3C3 SAUSAGE, PCUZTRY AhD FISH Introduction 1 Although sausage heats, poultry, and fisn can be clas- sified as minor meat groups, they collectively account for about 50 percent of the total quantity of meat purchased by M.S.C. Consumer Panel members (Table 15, page 120). Average weekly purchases of sausage were 1.12 pounds per family dur- ing the period July 1951 to June 1955. This is slightly less than the 1.24 pounds of poultry purchased by panel mem- bers and is more than twice as large as the average weekly purchases of fish. In the preceding chapters an attempt was made to measure the extent to which these "minor meats" are competitive with beef and pork. Results indicated that prices of sausage, ‘poultry, and fish had relatively little effect in shifting the demand for beef and pork. This does not mean, however, that the prices and purchases of beef and pork have no ap- preciable effect on the demand for each of the minor meat groups. In this chapter demand for sausage, poultry, and fish will each be discussed individual y. 179 The Demand for Sausage Keats ined earlier, sausage includes weiners, frank- H.) As de furters, bologna, salami and cold cuts of all descriptions. It also includes meat mixtures of various kinds, but these make up only about 20 percent of this moat grouping (Table 27). There appeared to be little relationm lip between weekly average prices and the quantity of sausab e purciziased (F iture XX). A simple correlation of -.089 existed between prices and quantities for the period July 1951-December 1952. A multiple regression analySis was made. ‘\ TABLE 27 SU.....LC\’Y OP UflCKASE DATA FOR DITITTJ EIIT SATTSACS TEXTS M.S. C.ocYSJIL PAIJEL, LIL... 19,152 "" IJMIS 19550 ‘I DC I\ Weekly Averages Product Average Quantity Expendi- Percent Price per ture per Buying Family Family cents per pounds cents pound ifeiners and franks 59 .51 18 29 IBologna and salami 59 .26 16 27 Cold cuts 75 .2g 17 26 other" .2 Tntal SC) 1 , 09 61; I.‘ i ' Holiday weeks omitted. ‘”‘ Other includes Prem, Spam. Tree t, chop suey neat, chili con carne, hash, soup and mincemeat. .nmaa ocshlamma has: .Hmcwm pofidmmoo .o.m.m .mmmmdwm p0 mooamm Una mmmafloasm owmhm>m haxoos GeoSpon mwzmcoapwaom .KM madwflm 930a 8a «:60 m we mm mm em 1 an n I I I I I I _ I l I 2M l lo. 1|: x Tm. .l x T . I . h fl .-0 xx xxfi‘fl x d a a my I. .I. I; 10.. O x %v ”k . d x add 1 x o my 0.00096 0 % ca 0 c 0 46m 1 II 0 x Ox 0 % x d .l: ,I x x Q 6 OX I. 00 .u x a . fll x 0 d 4. O IN _ m0w haxeoa on poameEoo AH.nV coapdsvo owdmswm Seam mawsefimom Nam: 00 av. O¢ mm Om 0N ON 0. O. m .Hxa owsmfia exam; .mm_ N0 NV N¢ NM NM NN fi—‘dqdd—q_—:—l—d—1fi_.fiqflq_—_fl4d-.’——q:~d.4~—_—.d«ddd—dd—q_dfiJ:—~_q———u——-—3d— hm I. :2 : I. I, III... 1 . \ [om . \ v.4». . I a :7 ":7 .\ __J’ x ,I _< I“ ~ & lfi‘s ..muh . ‘ 329.363 ’<>4‘ r. A! “N. T modmaqm .. mLIqao 5mm Pu. 325... 8a moaned _ a 1- I‘V '. —-‘.~ NA -, . -,- A, . -' I: [:3 N 1" —. ‘f‘ s. ' - -,‘ tween tne residuals and tangerature during the warn season -‘- " '7- T ~.- '1 3 ‘x‘ x .1. fi'fi —. 1w 3 .‘a 1" rs . . .T D of the year. a regiesSIoa analysis for this sane perioc 0i time inlicated tllat a 10 de gree increase 1L the weekly aver- age of mean daily te \eracuies would ~e aces- Inicd by an increase of .C31 paunds In s usage lurcnases. In co delusion, this analysis indicated that consumer purchases of sausage itens To not respond significantly to changes in the sausage prices. The i11terrela tionsdip of (‘0 sausage purchases to prices OI pork and poultry are not clear based on the asove regression results. A regress oiCfl aaaly Mi 9, using data for a later period, also yielded "negative" results. Teeklv sausage purchases were expressed as a function of sausage prices, temperature and the prices of other neat groups. The prediction equation was as follows, based on weekly observations for the July 1952-June 1353 period: (5.2) Y5 a .4148 - .OOSKl + .OOGOKB + .0045X5 - .0019X4 + .OO2CK5 + .0014K7 ij was the tciverat re variaole. The multiple regression co- efficient was .24. None of tlIe beta coefficients approached vnificance, even at the 5 percent level. The 1219 ale cor- L) P. s Jselation be twe en temperature and sausage purchases was only .15. A tentative conclusion was that purchases of sausaQe, as appregated for this study, have practically no relation- ship to the variables used in the regression analysis. Fur- ther analyses should give greater attention to a study of smaller sub-proups of sausape iteus. The Denaud for Poultry Keats The analvsis of panel data on purchases of poultry heats was linited to the overall neat group witn little or no at- tempt to study the denand characteristics of particular kinds of poultry. In any given week, approximately one-third of the panel fanilies buy some kind of poultry meat. Poultry expenditures made up about 12 percent of the total meat bill over a two-year period. Table 28 shows that chicken meat 2 made up about 72 percent of total poultry purchases for the period July 1952-June 1955, omitting holidays. Fryers are by far the most important chicken product with stewing chickens and roasters following in order of importance. Tur- key purchases are quite snail mostly because Thanksgiving and Christmas weeks are omitted from the tabulation. Actually about 75 percent of the 1952 turkey purchases were made in these two holiday weeks. Inspection of a scatter diagran of weekly average prices and.family purchases for all poultry meats (Figure XXII) TABLE 28 RELATIVE IIPORTAHCB “F DIFFSHLET KINDS OF POULTRY KEATS, M.S.C. 00333133 ANEL, JULY 1952-JUNE 1955 h!“ 1 A a weekly Averages Product Quantity nxnendi- Percent Price per ture per Buying Family Family cents per pounds cents pound Chicken Broilers or fryers 57 .40 25 12.0 Stewing 51 .25 12 6.2 Roastgrs 50 .12 6 2.4 Other 57 .04 2 5.5 Turkey 79 ..06 4 1.5 c Other .25 Totals 52 1.12 58 a Holiday weeks omitted. b Includes chicken parts, canned chicken and chicken pie. c Includes duck, and mixtures, chiefly chicken. 411:} .emppaso maaeuaon .mmoa maze-amaa aHsn .HocaI essence . .o.: hthSoa Mo moowaa Use memmsoasg ewwpebs masses seewpon QHSmWOmemom .HHxx madman I]... ncaoa 5a 350 mm vm Om we. mm MN -._ _ _ a I fl :_ a .fi. fi.2 4 . fl _ I a o x o 0 TI! x d C d C a o d d 0 d O O x o 00 x T O O O x xix d d m d d x d x d C 0 d .i O Oflq O x I: o x x o.w a 0 «ea 0 o .I d 0 I o o M Ar 0 o x 0 VI x xx m I I . H o o o x m . / o o x ”baa: em. MnX&_q E2 2m .39 o n o o I :2 t. Immm: >KISDOQ :2 new :3. . '; lint. >_::om Lea mncaod a showed wide variability in quantities purchased for any given retail price. Part of this variation is due to the changes in the composition of the overall commodity from.week to week. For example, in some weeks fryers make up an unusually high proportion of total purchases. This is likely to occur when fryers are featured as specials by some of the large chains. Variations in average relationships between prices and quan- tities may also be caused by lack of refinement in process- ing the data. In this study, no attempt was made to convert each poultry purchase to some standard basis such as "ready- 1 to-cook" weight. The large shifts in demand for poultry associated with the Thanksgiving and Christmas holidays were discussed in Chapter V. The effects of these holidays have not been com- pletely removed from the data by omitting the weeks during which the holidays occurred. Close examination of the data indicates that the demand for poultry meats tends to be de- pressed during the weeks before and after Thanksgiving. For the period July 1951-December 1952, there was a correlation of -.40 between the weekly average quantities of poultry purchased per family and the city-wide average price. A seven variable multiple regression with the quan- tity of poultry purchased as the dependent variable yielded an R of .58. The standard error of estimate was .148 pounds per family with a mean of 1.022. The prediction equation 1In the food purchase diary poultry purchases are classified as alive, dressed, ready-to-cook, boned or selected parts. See page 10 of the diary in Appendix. 189 was as follows: ( 1) Y4. 1 ’ 2 ’ 3 4 The price of poultry and the price of pork were both signi- ficant explanatory variables with the price of poultry being the most important according to a comparison of the beta co- efficients (Table 29). The regression coefficients indicate that a change of 5 cents a pound in the price of poultry meat is associated with an opposite change in quantity pur- chased of .144 pounds. The price elasticity measured at the mean was -1.45. The cross elasticity of demand with respect to perk prices was 1.02. This was about the same as the cross elas- ticity of 1.08 with sausage meats. The corresponding beta coefficients were tested for significance, with the result being that the pork coefficient was significant at the one percent level (t u 2.98) and the sausage Coefficient was significant at 10 percent level (t = 1.65). These cross elasticities may be biased upward by the interrelations in seasonal shifts in demand for these three commodities. This same problem was raised in the preceding section dealing with demand for sausage meats. The residuals were computed for the regression equation described above (Figure XXIII). he definite seasonal pattern was apparent in these residuals. There were substantial de- viations in the actual poultry purchases compared with the predicted quantities. Further research would be required to develop a more precise prediction equation. Attention prob- ably should be centered more on different poultry items such as fryers, roasters and stewing chickens. The Demand for Fish Fish expenditures made up only 5.4 percent of the total meat purchases of panel families during the two year period beginning in July 1951. Kevertheless, an average of 59 per- cent of the families were buying some kind of fish each week. The group of products labeled as "fish" includes all fresh and processed fish and seafood. See the purchase diary in. Appendix.) It was pointed out in the preceding sections of this manuscript that the price of fish has little influence on the purchases of other meat groups. Further analysis indi- cated that fish purchases are not greatly affected by chang- ing prices of hese other kinds of meats. There was a sig- nificant relationship between the price of fish and the quan- tity purchased each week (Figure XXIV). A correlation of 3.3 soapsswo hhpasom Sofia madduwmom .HHHVQ ohswam Nmm_ mxomg .mm_ mm mm. mm m_ _ mm km 191 N. m. EEO“. 5a >m.:30n_ . mh<39mmm wagon. .UopuHEo mhwpaaon .nmmH esdhuamma mafia .Heswm headmsoo .o.m.5 .flmHM mo meowha use memafloadm ewwwe>w hammos seespen QHQmQOdeHom .>Hxx ehdmfih ocaoa .ma macoo % ms Ox. .3 cm mm cm 1. %\ . w , Ad __a::_“:am.__x:w:___Ta__>\v Tl To .% x o .llm m A» Au .0-0 0 Aux x o ”1 0 Cd <68 0 lg a oo o o o o& x x O Ox 4 .- . TI 0 Q of q x0 x '1? O Ox Cd 0 O. O [I Q Q Q. C l G x II o a a :Ihm . a I. O < d x0 .1 a I40. x x x O x x In :2 E .32 q a s. :2 2N.mnm.o 1m; :2 t... .mmm; x :28... .2 :2 2m .62 a a mncaoa ) a 195 -.57 existed between weekly average purchases and prices of fish for the period July 1952-December 1952. A multiple re- gression for this same period, with the quantity of 1181 as the dependent variable, produced an R of .65 and a standard error of estimate of .079. The mean quantity was .424 pounds. The prediction equation was a follows: (5.1) Y5 = 1.6205 + .OOlQXl - .OOG4X2 - .0027X5 — .0051X4 - .012 X + .OOlQXG 5 I The beta coefficient for the price of fish, X5, proved to be icant at the 5 percent level (Table 29). Since Pb highly signi a of the pork cocflicient was negative, there is cl- ; C.) m Ho :2) reason to doubt that this is a true structural estimate. The residuals for the aboveequation show that purchases were greater than predicted during the Lenten period of 1952 and during the fall months of both 1951 and 1952 (Figure XXV). Purchases were less han the predicted amount during the sum- :ner months. This pattern supports the belief that the demand for fish increases durin; he Lenten period. The increase in demand during. he fall months is probably part of an overall seasonal increase in demand for meats. The decline in fish purchases during the summer is probably related partially to -Uie increased use of fresh fish caught in local lakes and streams and the general decline in demand for all meats. 194 AH.mv moapwsvo swam Soak madsvamom .>xx onswam «l:—__:41...:ZZJZZJZ Nmm__ mxm03 mm mu m. rm; .. m4<39mwm _u_f:_4____:_______Lg.____:___._f:d____:___ .mmw_ _ mm LN I.¢s| ILnJI ILN.I IL. n >:Eou emu moaned FD «.0 be. ObmH. meH. 0. maoo. m «. oso. 02 we. mesa. emma. . Heoo. m LO m0.H mwa. ¢O.H wfiwa. oe.e nmofi. oeme.u mmao.- we. ,eoo. neso.u Hmoo.u 0 [0 CL) :3 o C) .r'? O (O OPDrHD o C3C7O o ( W V_; Om.m mwOH. 00.H osma. meow. mmao. 03. (P\ owma. em. omvfl. ammo. Omoo. wo.m bmHH. m 0.: $000.: mm.N 0H. mam-. Hmma. @009. N®m00l ObHO. MH00.: 00.¢ ww.a mafia. omva. mafia. mm00. 000m.: 0H00.: have. 000a.: Q 00.H ws0. 00. $000. H000. 5000.: 0000.: 0HHO. 0000.: Q o efldmfiwm .z% m 13‘- m 6 1&1 n J.‘ NU.” H ”\m :mah mapflsom o semen MAOm Meow M. M.hm m esoodH no me we we we I eoahm eoapg eoflam moaam mowed mohdmdo; :: - 4 ii. Hecapmfipmpm meapdflse> hQOQmCdeKm was poipoam mega man Home 09 Head wane .eeam qnmam m - yd ; J: .1 1, ‘ .mee are smegboe mean leIP : q 4.11 Q 0.5 PD. C r 1.4.4434. firfteE 114,9 24.) «J «1 r.:rr Hr. H .U .: fill.-.) .55.). O O .4. Lita? 0 0 m L .3. «1512... NO W341i: n1 _.l,- r.\....,‘) Pb. . Arm ._, L . Dru I/‘( O - "\ “Why'd“! ‘2' \J -H‘LA LHLL .‘4AL m"-" "‘ 1' I“. ‘T“‘ ‘ T ‘l ‘ r71 Jun“ 11...- 41.41.) -‘ Vii Ales.) l‘na—Ji—L-L .. A... .‘~. -.. '1’. . Studies of the agéseéate demand for all meats nave been nade by Iorkinq, Fox, Shepherd, and others. These stud cs -§ .‘ 1 o o s o I 1 ~—, a were described earlier in this manuscript. 0338a on annual data for the 1322-41 period, price elasticities of deaand of approximately -.7 were reported. In these studies all teat was defined to include the red neats--beef, pork, veal, and lanb. Poultry and fish were excluded. In this study all meat has been defined to include beef, pork, sausage meats, poultry, and fish purchased for home consumption. Veal and lamb were not included because of the 1 small amounts purchased by panel families. the city-wide average price of all meat was computed as a weighted average. The sums of the expenditures for the various meat groups were divided by the sums of the quantities of different meats pur- chased. As the composition of the meat supply shifted toward more and lower priced beef in 1952 and 955, the average price of all meats declined more rapidly than if a price in- dex with base period weights were used in computing the aver- age price. 1 See Chapter III. A scatter diagran of weekly average prices and pounds purchased per fahily showed rather wide variations (Figure "XVI). In spite of Opposing trends in beef and pork prices, 11 - -f’ ‘ i f h a d- i i i ' n. d e;-t — t e averaue r ce +or all 18°tS eel red nIo1 6 c «‘3 er ’6 pound to 57 cents during the two year period, July 1951 to June 1955. The decline in prices appears to have been ac- - companied by increasinp quantities of neat purcnased, but the relationship is far from perfect. The correlation be- tween weekly average prices and quantities for the two year period was -.52. A regression of quantity on price pro- duced this prediction equation: ..’7 o o n o o Q ~ 0 I o unere Y 18 quantity and X is price.9 The price elastiCity l at the mean values of price and quantity was —.75. The standard error of estimate was .55 on a mean of 7.59 pounds per family per week. When a thirteen week moving average of weekly family in- come was added as a third variable the multiple regression coefficient became .59. The prediction equation was as follows: 6 s The coefficients for price and income were both unreasonable. 2 Lajor holiday weeks omitted. .weppago £320: .33 05733 33. .388 $5800 .05.: 33s Ham mo mooamm new memdnomsm omwpebw Magoo; seespen Qaflmsoauwaom .H>XN emdwam % ucaoa Ea word 3823 1 ON mm 00 wok/sir A _ _ _ ‘ _ _ d _ _ _ J _ _ _ . A _ .lJm.® W x p l .. _ oo o o _ T o O O im..fi O O o o x x o O O O O 0 d C C X. Q T. o 00 x 58 x a a 0 an [on o x aux a a d 0 00 xx x G do 0 x 8. a a dd x a 1| *0 a 4 IO m 0 x. x d d . o 00 d a a to; .m. mom: 0 x :2. new .39 o 0 Lee :2 ,5. Jam? mh \r,-. r \{3 J- .‘ -‘g to glue .1o‘il si s <3i :_,;;o .,e_-a;ii *LALJ. 3.1)t..i.is.-ot. De, Anes, 13 43, 55. p. ::i.§J. 4--Lk/|~J_L.Q, _LO¥.8L JLCIUe UC)llU A Friednan, Iilton. "The Iarshallian Demand Curve," Jonr. Girshick, X.A., and Trydve Iaavelmo. "Stati stical Analvsis of the Demand for Food: Exanyles oi Simultaneous Esti— mation of Structural Equations." Econometrica, 15:73- 113, 1:47. Gold, Norman L., and Iaz; ine Lnlow. "The Demand for Food b: Low Income FaAilie s." Quar.Jonr.2con., 57:526- 629 ,134 . Haas, 6.3., and Yordecai Lzekiel. Fact rs Affecting the Price 0 ‘3 T— I‘ ‘. T T‘ A”. - ~,". {71" 0;. .2; S. L.S.-)Opt. OJ. ADP“, JL‘LI. Ira-u, lapu. ht”.— Iaavelmo, Trygve. "The Statistical In2lications of a SJQtOn of Simultaneous anations." Leonometrica, llzl-l2, 1945. Hicks, J.R. ‘Jalve nd Canital, 2d ed., Oxford Univ.Press, LondOlAl, 13‘; j , ~'\ 91“) Hildretn, Clifford, and Frank Jarrett. "' “tatistic al Study f Livestock Productizwn and artetin- " WCOxles Com is- sion Discussion Paper (Scononics I .2055 and Statistics £0.575), November 15, 1352. Judge, Georme Garrett. An Leononetric Analysis of the De 13nd for n s. Unpublisned‘Ph.?. thesis, lowa State 0011060, mics, 3:.)52, 250. pp. Katona, George. Psycholo;ical Analysis of Economic eliav ior, KcGraw-Hill, Rew York, 1351, 547'pp. Klein, Lawrence R. Econometrics. Row Peterson and Co., Evanston, Ill., 1355, 355 pp. Kni Oht, Frank H. "Realism and Relevance in tne Theory of Demand." Joar.Pol.Lcon., 52:23‘-518, 1944. Kuznets, G.1. "I.easore :ent of "arket Demand with Particular Reference to ConSLLHer Demand for Food," Jour. Farm Leon., 55:273-295,1055. Kaznecs, G.I., a:1d Lawrence R. Klein. A Statistical Anal"sis of e 130798461 Demand for Lemons, Siamlini I-omld” nor. .LJ'COLLQ :Le Sgt. C‘I, lz‘fa’ 112131). 214 Larshall, Alfred. rrncwol 3 3? Scenes cs, Iacmil1an, London, 8th ed., 132 , 371 pp. geat Lerchandisinc Inc. Laster Zeat Pricer, 105 8.9tL St., St.Louis 2, A0., 1949, 87 pp. Killer, Earl E. tnanres in Demand for For Products." The Livestock nd eat Situation. Bur.of Morgenstern, Econ., Morrissett, Loss, Thomas N. Adr.3conomics, a De t. CL A r., Lay-July, 1955, pc.14 JO ‘ Oskar. "Deaand Theory Reconsidered 8:154-201, 1948. Irving. btitution--A Survej. Some ." Quar.Jour. "Bone recent U es of Blast icity of Sue- Lconometrica, 21:41-62, 1955. Relationships of Selected Socio-Eoonomic Factors to OOdivOHbthCL\H and Axpezidftt res, Lansin~ SprihQ SJ, oApuolisAcJ’lA.D. tiles EB, cellebe, 1955, 565 pm. Losteller, Leasurenent of 19 51. National Live 5 +rederick and Phillip Utility," took and Aichiban SCate Logee. "An AAneriuental Joal.Pol.Lcon., 59:371- 404, heat Board, Pri cin n37 Retail [eat Cuts, 407 S. Dearborn St.,Chica o, Ill.,_23 p1. Norris, Ruby Turner. The Theory of C nsumer's emand, rev.ed. Yale Univ.Press, em Laven,VI§52, 257 pp. North Central Livesto Larketing csearch Committee. Leat Aeta111n* in t‘w *ortn Central Su letin maAuscri fpt. Prest, A.R. "Some tates, preliminary eul- xperlients in Demand AnalySis. Review of Leon. Statistics, 51:53-49, 1949. Purcell, J. C., and V. John Brensike. Net “arlietin; and Slatx éhter of Livestock and Consurntion DrCliAlnarJJTeport Bar. of A r. Loon., Quackenbash, Gerald G samer Panel." Sta. and D.C. December 50, "Deaand Analysis fr01 tA Journal Article 30.1594, “' A paper delivered at the JOint Aeetin Stat. Assoc. ’ Bar me C)ions, 50. :ToSoUepto UT A DIR, 1.5.0. Con- Alch.A5r.pr. g of the Amer. 1e Amer.Farm Econ.Assoc., VaS1id ton, 1953. Fa- Gerald G., of Ice Creai an I”! Q‘Llac 1:.enbu811 , Chas C" o manuscript, Dept. Rousseas, Ctephen 3., cation of a Composite Econ.,S :283-3 8, 19 Sartorius, fiarketers of Lester C., and Food Pro and A.G. d J.D. Shaffer. Consumer Pur— or Zone Use. Unpublishefl Qcon., uiciigan State College. n :1. “‘X erinental Verifi- Lap." Jour.?ol. Hart. Indifference 51. of Aur., in coofic rati larguerite Burk. Eating Places as ducts, Bur.of Agr. Loon., 3.3.be3 Ion with the Lil LV of .Zirm., ---arket- . . ‘ -_ . H firm 115 esearcn hep ort $0.0, 1334, 11:3 pp. ‘ 1 r - ~ _‘ , 3 7.- 4 -_ , ( 1 'i .- o Scnraoer, 3.“. iue be ana for Ledt in Candoa. JCOHOfilCS “\_’ ,3- . . " __ \n r .[TL_ ,. . fl. ., (“ff LiVisioa, C333 a Lost. oi “91., Cubdn&, 341], 1335, 10 pp. 1. --_ - 3,. I! '1'. : - H» , . t‘ .. n 32, 1‘1le, ”01112) . Lie 3tatist ical . edsare ‘Out OJ. L asticity '3 .43. 45" “ " l! . '-" . .r‘r , (\{\, oi Leland ior geei. J01r.33r1 3con., 6.434-278, 33 . . 'N/ (:1 1 ‘ \ x. ‘n ‘3‘. q ".r- . ._.___ ‘Tie T oory cud lea * eleat oi D0l3fld Lliv. of --° \; I r! ‘V T‘ '7‘ (AT- W s.- Vii—LLUL O AI}CQU, VLL~CL1UO’ lVC3, vll £J'i). (“1" ,. f1 3') "-~ r'w ‘ “ ,_ , __I 1‘ ‘T ‘1 ," «3,, 3uuiier, J.u., 3.3.Quachenbas3 and m. .w ss. Tie Cons p— .O , ., .4 ‘. _ ( ‘ r? ~.-‘ 1 1. 3 v ‘ V‘ . -‘ I tion Oi ,eat 333 Jeiated Procacts L3 Iansimv, icki-ua n, i-.. .- . if? r* . J g ”173‘ _ 3.: l ' «A, .;_._,uu. -.-..'_C'_ Lox-10.1.”). 4.13.: ’OubLJ-o, ,LCC. quljo:uu, Lin-3:, bi) pp. 1 :0 "' *' " -‘ ‘“ « -’-‘ '-‘. — . ' -. Siaifer, J.D. :e3103olow-cal ”es 3 lap ,ie (berat:ioa of a n,...' ... , ~_~’._,~,,“r . ,. - 3 __ ‘1 TT ‘r n ~ l.‘ mar CULJJ‘.3.;O ‘ 1.....L 13313-38 _ kiwi; . 31-333il34pu i 1.1.1). b4lCDlS, - fiT. ,. i.4_, " q art) r‘ ...l --_Loc~;l .3 3.3.36 3311ch , - 331;. , 7'37 pp . _ , I‘V‘ ' '3 v r _V' \ I“: v .___—— A Plan for 3a. q>linu 33anbinu Bogulation LVGP .- - " u, - '7’“ a fi {7'7 - r: '31- L3 0 LT'jLilq. 17‘an . .JCO--. ’ OK.) .15 0.3-le), lEJU4‘. j ‘4 1 "I 'F‘ 'n - '— 1 - n V‘ —.. -.. - J-I . ‘Aw n n ‘ . VF Snep3ero, 3ooiiroJ. Cialges in 310 ‘e and for e3t and Lal3l "‘ J- 9 . . ~_ - .L. V1 - ._.3 fl - \ JPPQ 31.13123 L-S J. W. t J3 1 {J (:1 '3. L.) U'VL :16...) n.) L- CV I- fl .L.\) o :vaa 31‘) A 0.944;; 3 0 TI.“ ,..V-.- T1’“~.-.. Ff“. .. 01.155... l-JLLI. Ukli‘, .Lv‘izu , 1.)./ 13);). \14 m“ N ‘7 "> . 3616". " m- "' ‘ - "" H T ‘ 7" -———_— ‘Lile I -1C_LC .08 C'.— 461-8 1 COL/u _._-L *(11\ 0",]. out) 0 do -1 "1 1,. V" ‘ r‘v. ) .-_-..-.,~ 'IPCVC‘ 4.! 9.1 L ~_\.IC"".. , l/ .Ufal ”U , .L 20.)!) :.: ‘ “I-.(-q " —. a - r. \ 1" '3 ". ,\. J- .'\ ‘ ( .' ’ .'* ‘ ‘ \/“‘fV "' 1’ DJ- “Lulu ’ “31.1.4. .. 0 '31 — .‘Q .. \JJI. _ A“ 91' 100 'v L L/ J .L ~ :V): K/ s... C L L-— g h .- , .. n (‘1‘. ~.-.—‘.A' . 1---7 I- 1 ‘- ;. - - -'-._ - 3. ".- l v. -»- ‘J 4 . .’ LL; ._LJ '..'. .. ..‘_ '-‘ J. _ , _~J , g . __, . ( IL).J 5.1013.) ‘iuij . .LJ‘L ). ”j— - ’\ A _ F‘ f..."\\ '\'\ @u3l.’ “VJ... ;\J'—’, l L/J.’ U‘v i'lj. C —. ‘4 ‘7 Q - -/ - . up“ .3- . 1 - "'\ ~-v‘ " fl‘ , ~ fi- ~ ~- ‘fi ‘ I ;‘ L H.1— Li.) ’ _, .. O 10;) (J . , p1. A.»L -__ ..C LlJuc {Q ' $.'.L.)\‘.LJ.. ‘A"3\,)v1qc‘.‘-)fi_lu¢' (3.1L 3.1L) ~.-' ”-.‘1 ‘r- »‘.- .4—-° - . n. ‘- . J— r . . H 11. °.. .- A ‘— Ami.“i rol¢c3i_- 333 “cat 3033-: tier. "&3 Liiestcc- ‘a r v\ "--. z—‘vr‘ ’- ,- ~‘j‘1 \‘0 ll - ‘ ' .‘ ,\~- I.— '5 ‘1 'C 63..) .3 .. 22.3-1 'u_'-C\' in -J.Lro(-'i J“; ’0‘;- 'COHO, v.9. 'I-IUIV. (J‘L _ ~, 4 .1. r‘, ""3 . 0'7 $101 0 , 4. VLUV‘LU U, j. u' .22.), P jolv-LIL). .I ll F 1‘ .o u c».-.- u, :r .2. rx —. - -. :I n C - t 1.- -. f" 4. A J— o.“.c1, .JlflOlL..u., .Co 3 er In3.aJiL.Lo: -;..mmi CL3,gClileLl to v,‘ ,. in n .._1 ”3.4- ,‘-J- A"- , . .-f T,!_. .419 -8. KC- ... \‘I’ LUDCA... , \-.-'..'-.LJ 0 .../"x/‘J u. U“ 1103. ...:CUiL. a...“ ,,,x.. I“. , -- J o __ _ A. .- _ .1 ~ .r _ ,_fi ‘_ \ '1 r 3 _, [‘1‘ -?_ 1' _o .L.’ '\. ..- ogivcrsitJ oi deJLiJQ, “iscellaucols -JQlLCQQLvA .o. O ."~L'.' '20 - — llu, luul, (JO 5):). :‘- -9 J.‘ “, fl ,;: .L —.‘-, “ PM!" 4.: M .4...” -,... - ‘. In ...—3- ..-- gain 1 -L .1; o J. 210 CUES .1-.. CC 123.1-“ Ungls 9-4.01” - JI‘C..3.SGS- 0.. ' I“, '519 {1‘ .' , "I .11 “no, ' . Y ... \ f .s..‘ "'1 ._“r"‘ -" _‘ *7”: f'/\ v-1'CZL’01u, ”blur-«3.173 4.50.1.1. “4.1...uutlolo amLQJJVL .L’auv’ Ivuu, r-vrj- _ . UU pg. or A u. 7' -° 1- -- "1* n - 4- --~- - m ~.—\ '1 n . switn, Jictor 1e olaSSicists’ ,so i. uc-&ac. Uoar F01. .3COl“.., 5‘):2/“2- -25'7, 10:31. Snedecor, 300153 1. Statistical Lethods. 4th ed., Iowa State r1 110 “ T) (‘0 ' 1 1"r ‘ F- ,1" :1" 1‘ U0 «DO .Lre-JU, it. $053 , iUb-Cl, Lu‘iu’, ..A-LJK) $-19. Staehle, Hans. "Llelative Pri es and Postwar “ar”ets for Anigul food :roducts." i-er. Jou_r. Econ., 59:257-13, 245 Stiglcr, George J. T“ Tleor" of Price, rev.ed., fiacuillan, fie L New York, 1952, 310 p“. Szatrowski, Zenon. "Tiue Series Correlated \itl the Jeef-Pork Consumption datio." ‘cooo‘etrwca, 15: CO— ~78, 1945. TL0usen, F.L., and R.J.Foote. A tural Price, 2d.ed., LcGran- -3ill, low York, 19 o2, SJD p' , Gerhard. Econ ometrics. John Wiley, Sew York, 1952, 070 pp. U.S.:ur. of Census. "Cliaracteris i s of the Population." 1950 Cer sus of Population. Vol.II U.S. Dept.S of Agr. Family Food Consumption in the Tnitcd State L1942, Llscellaneous Publication $0.530, 194$. .,.- l k, and Poultry Products. Unpublished Pn.J. thesis, Rahbv Omar. Econometric Analysis of the Demands for Beef, 1 r Iowa state Ubllch, lQSl} Raite, 9.0., and R.?. 00):. A Stidv of the Consom tion of Ieats in Linnea1301is, C31. ufnxm sota Abr.;: {p.Sta . 3.1.11.2,1 19 5-, 26 pp. _~-fl - I Jaite, U.C. and H.C.Trelogan. A)TlCJlL“P&l {arket Prices. 2d ed., John Wiley, dew Yelii, 1351. 49 pp. ‘ ‘ V'v!1{.gl'.l In. -' in. 1 Hull" .. 217 .‘e £u1alENsis 2n1d .1- "Statistical Techniques for Pricinpa and Trade, a :52: pp. Ralsh, R.n. Interpretati n of Price Data. report of the Katienal Lari-:1; ting ‘.5Tcrl;;sicp, 107-18 aU5h, 3.7., "Applicability of Recent Developaeqts in Jethodo- lorf to A ricultural Leonoaics.” Jour. Farm Econ., 692- -706, 1955. Iovza State Col- 55: ...... Readings on Agricultural :arketi in. . lege Press, Ancs, Iowa, 1954, 456 pp. lectures on Mel tical Lconomy, Reutledge den , IJSe, 220 pp. 'V‘ Jf“ ..u "icksell, Knut. and Kegar , LEJ., Lon S, Jenn Wiley, e“and Analys Told, Herman, and Lars Jureen. New York, 1955, 553 pp. Noodlan, .3. The Infl11ence of Prices 01 tne Relative Con- sumption of Lee? a1d Perk, Dco1o¢ics Division, Canada Dept. of Apr., 1955, 9 pp. , v \Wor ng , Elmer J. "Agricultural De :nand Du “l ing Rearmament." Jour. Farm Econ., 54:200-224 952. ——_——. Studies in t1e Teasurene nt of Demand with Special Reference to t11 e Demand fer_Zeat. Ln Mli we Ph.D. 1952. thesis, Karvard, 218 APPENDIX MICHIGAN STATE COLLEGE WEEKLY CONSUMER FOOD PURCHASE DIARY This diary is for recording all food purchases for the week of Sunday .............................. through Saturday .......................... I. May we emphasize that each of your diaries is important to us, whether your food purchases are many or few. Your diaries will be of most value if made out accurately and returned promptly — every week. 2. We suggest that you enter food items in the diary each day as you make the purchase. 3. If a food item that you use is home- grown or a gift, show this by writing “home- grown" or ”gift" In the price column. ’ 4. If you don't know under which heading to enter a food item, you can list it in one of the blank spaces on page I5 5. At the end of the week check through the diary to make sure you haven't forgotten any purchase or made any incomplete entries. 6. As you are checking the diary also V the squares (D None) if appropriate. 7. If you want any information, call us at the college—number 8-I 51 I, extension 7364. INDEX FAG! PAGE PAGE BAKED GOODS ........ l2 FATS and OILS ........ 3, SUGAR, SWEETS ...... 13 BEVERAGES .......... I4 FIRSH and SEA FOOD. .IO VEGETABLES ..... -. .6 8: 7 BABY FOODS ......... I I ............ VITAMINS ........... I4 CANDY ............. 1‘3 GRRAIN PRODUCTS ..... II MINERALS ........... I4 COOKING AIDS ....... T ............. VITAL DATA DAIRY PRODUCTS. .2 8.13 POULTRY ............ IO Questions .......... I5 EGGS ............... I0 5 ............... I3 WHAT YOU CAN EARN IB_Y KEEPING THE DIARY If you return the diary for 52 weeks or more without missing a Week, you earn 40 points for each diary returned in the sequence. OI’ If you return the diary for I2 to 5I weeks without missing a week, you earn 35 points for each diary returned In the sequence. or If you return the dairy for 5 to II weeks without missing a week, you earn 25 points for each diary returned in the sequence. or If you return the diary less than five weeks in a raw, you earn 10 points for each diary. PLUS I. A bonus of 5 points for each diary returned on time (postmarked before Tuesday noon of the following week). 2. A bonus of 70 points if you return every diary on time for a year. . 3. A bonus of 10 points for each diary returned during July and August. 4. A bonus of 5 points for each diary returned after returning 52 diaries. You can earn 2500 points the first year and 2760 points for each additional year. (2) DAIRY! PRODUCTS MILK NONE [J n a FRESH I I00 321:. perhgirart A;:':'uld Partner“ Brand Homogenized—Vit. D. I I I0 3 o Homogenized—Plain I I20 2’ 5 Regular Pasteurized I I30 ‘2 Jersey or Guernsey I I40 Buttermilk I l50 Chocolate I I60 Skim ‘ I I70 Sour Milk, Yoghurt, etc. I l80 Egg nog, etc. I IBI Other Milk I I90 CANNED 1200 5:33: £32.25. 3‘12. .Ei'ét... Aim... 3...... Eva porated—Unsweetened I 2i 0 Condensed—Sweetened I 220 DRIED I300 .i'i'fi'ubii. ”:2:sz Autixlnd Brand Powdered—Skim Milk I310 Powdered—Whole Milk I320 Powdered—Baby Formulas I330 Ice Cream Mix I340 Melted Milk Powder 1321 CREAM NONE [:l I400 .iimlk p.333... 43%... 9.323.. Brand Coffee Cream I410 3 ;; Whipping Cream—bottle I42] ‘3 E Whipping Cream—can I422 "" Sour Cream I430 i *For Fresh Milk and Cream—Please indicate from whom it was bought in the fourth column as follows: hum—- . it delivered by milkman . If bought from grocery store . If bought from cash and carry specialized dairy store . If bought from other source 3 DAIRY PRODUCTS (cont) ( ’ N mb Pri T tal Wher .53 ICE CREAM NONE D I500 oful’intesr per Pint Am: Paid Purchaseed Brand Hand Packed I510 Pre-Packaged 1520 Other 1530 Check one: u , N mbe f Pric Total Pr - CP‘EESE NONE D Liz’s" 0'1: per Potemd Amount Paid Bulk Jar Pkgd. Natural American Cheese # 1610 Processed American Cheese 1710 Cheese Spread 1720 Other Cured Cheese 1620 Cream Cheese 1810 Cottage Cheese 1820 N mb r Pri e T tal FATS NONE D 2100 of Poursds per Pcciund Am: kid Butter 21 IO Oleomargarine 2120 Lard 2130 Vegetable Shortening 2140 Other Fats (name kind) 2150 N be of Pric Total OILS NONE E] 2200 PinlIlflor 231:. per Uchit Amt. Paid #Cooking Oils 2210 Mayonnaise 2220 Salad Dressing 2221 ' fSalad Oils, French Dressing, etc. 2230 Q fOther Oils 2240 ’rTartar Sauce 2241 r—Sandwich Spreads 2242 ' Whips and Toppings 2250 ’— The extra spaces are for additional purchases of listed items and for items not listed. If there aren't enough extra spaces on the classified pages, turn to the last page. I" FRUITS Number Sin Price To I BERRIES ownm ofUnlt pcrUnIt AmiJsaId Blueberries Cranberries Currants Dewberries Strawberries Other Berries CITRUS Lemons Lemon Juice Limes Other Citrus Other Citrus Juice Mixed Citrus Fruit Mixed Citrus Juices OTHER FRUITS NONE Ci — and I - Please don't forget to enter home grown, home canned, and gift items. m ‘vnu.‘K FRUITS (cont.) ‘5’ . . e i E 2 5‘; OTHER FRUITS Cont. czar: .f'5:.. .352... 233'... £- E 8 E 2 E Avocados 3330 Bananas 3340 Cherries—Sour 3 351 Cherries—Sweet and Maraschino ‘ 3352 Dates 3360 Figs 3370 Fig Juice 3379 Grapes 3380 Grape Juice 3389 MELONS 3410 Cantalope 34I I Watermelon 3412 Other Melon (name kind) 3410 Nectarines 3420 ‘ Olives 3435 Persimmons 3430 Peaches 3440 Pears 3450 Pineapple 3460 Pineapple Juice 3469 Plums 3470 Prunes 3480 Prune Juice 3489 Raisins 3510 Rhubarb 3520 All Other Fruit (name kind) 3530 All Other Fruit Juice (kind) 35-9 Mixed Fruits (except citrus) 3590 Fruit Cocktail 3590 In reporting Fruits and Vegetables please indicate, where possible, the actual quantity purchased in weight or liquid measure. ' iii ‘6’ VEGETABLES 5‘ I 1 a I s GREEN LEAFY ‘ 3 f 3 E E “- VEGETABLES NONEE] 41.43 2:337: ““57." .352» A331,“: 3 E 5 5 3 Brussel Sprouts 41 IO Cabbage 4120 Sauerkraut, Cabbage Salad, etc. 4121 Celery 4130 Celery Cabbage 4140 Endive, Chicory, Escarole 4160 Lettuce—Head 42 I 0 Lettuce—Leaf 4220 Mustard 4240 Parsley, Swiss Chard, Water Cress 4250 Spinach 4260 Mixed Leafy Vegetables 4290 Other Leafy Vegetables 4300 I r: E '5 3 E I’. g I 355329;? "“3332 a 44-45 2:312: .f‘az. .::'a:.. .1231... E a .2 Artichokes 4410 : Asparagus 4420 Beans—Lima 4430 Beans—Snap 4440 I Beans—Sprout 4450 Broccoli 4460 Carrots 4470 Corn—Sweet 4480 Peas 4530 Peppers 4540 Pumpkin 4550 Squash 4560 Sweet Potatoes and Yams 4570 Mixed Green and Yellow Vegetables 4590 Other Be sure to fiill in the "size of unit column," and at least two of the other three columns, as well as check the method of preservation. Mr VEGETABLES (cont.) (7) .c E 3 Gtéflgfgs NONE g 47.49 2:35: .3113. Ami,” s s Beans—Navy, Baked, White 4701 Beans—Kidney 4703 Beets 4710 Cauliflower 4720 Cucumbers 4731 Pickles and Relish 4732 __~_. Egg Plant 4740 Garlic 4750 ll Mushrooms 4780 . Onions—Mature 481 1 Onions—Green 481 2 Oyster Plant (Salsify) 4820 Parsnips 4830 POTATOES 4840 Michigan Potatoes 4841 Maine Potatoes 4842 Idaho Potatoes 4843 California Potatoes 4844 Other state grown Potatoes 4845 ' Potatoes (Don't know state) 4846 — Potato Chips and Sticks 4847 Potato Salad 4848 Radishes 4850 Tomatoes . 4871 Tomato Catsup and Sauces 4872 Tomato Juice 4873 Turnips and Rutabagas 4880 Other Vagetables (name kind) 4900 Mixed Vegetables 4990 Chop Suey, Chow Mein, etc. 4990 Mixed Vegetable Juice 4999 Vegetable Soup 4991 Vegetable and Meat Soup 4992 . Please don't forget to enter home grown, home canned, and gift items. 18) BEEF Beef Corned or Beefi All Other Beef PORK Bacon Canadian Bacon Canned Pork Steaks Ham Picnic Ham—Shoulder—Butts Uver other Roast Pork ' Side or Salt Pork- — »— 5100 5110 5120 5130 5141 5142 5160 5170 5300 5311 5312 5320 5330 5335 5341 5342 5351 5352 5370 v5381 5382 5390 A‘5391- MEAT Number at lbs., On. Number of lbs" On. Price per Pound Price per Pound Total Amt. Paid Total Amt. Paid Boned Boned . l I! Check ll: Check If: MEAT (cont.) (9) _ Check It: , Number of Prlce Total [ LAMB-MUTTON NONE E] 5200 lbs., On. per Pound Amt. Pold Boned Frozen PT; - Canned 5210 - Chops-Steaks 5220 _l 1 Ground or Grinding 5230 1 Heart, Liver, Other Organ Parts 5240 ' Roast (Leg, etc.) 5260 - Stewing, Soup, etc. 5270 . Other Lamb-Mutton (kind) 5280 VEAL NONE D 5400 11:31.02: per? 2:.“ AnltTPlald Boned Frozen Pratt. Canned Veal 5410 Cutlets, Chops 5420 Ground Veal 5430 ' ' Liver 5441 Heart, Tongue, Other Organ Parts 5442 Roast 5460 Stewing, Soup Veal 5470 Veal loaf 5480 Other Veal (name kind) 5490 1 OTHER MEAT NONE E] 5500 Number of lbs., 01:. Prlce per Pound Total Amt. Pald Wieners and Franks, etc. 5510 Bologna, Salami, etc. 5520 Cold Cuts 5530 Rabbit and Other Game 5540 Other Meat (name kind) 5550 Prem, Spam, Treet, etc. 5551 Chop Suey Meat 5552 ‘i’ Be sure to if meat is frozen, boned or pre-packaged. Other meats, poultry, eggs,rflsh and other sea food are on page 10. 00 Anotgincludesales (9’91“ reporting price or totalamountpaid. Report prepared baby food on page 1 1. I i as ., ... ... B Sawfly was m «mm “W‘s-J!" , “01 MEATS, POULTRY, FISH, EGGS ‘ MIXTURES — N b F Lb . CHIEFLY MEAT NONE D 5590 :r'I‘d/trOOzs.‘ Price per Pound Total Amt. Paid Chili Can Come 5591 Hash 5592 Soup 5593 Mincemeat 5594 Check One Check One 1 I 3 1 S I 1 I POULTRY NONE E) 5600 Number Te? «:23... g "g :3, s 15 u. I E E Lbs., Ozs. Pound Paid :: 2 8° g 4; E E 2 5 CHICKEN 5610 < O a: 2 a: me. u. u. U Broilers or Fryers 561 1 Roasters 5612 Stewing 5613 ; TURKEY z 5520 l DUCK 5530 1 OTHER POULTRY (kind) 5640 j MIXTURES—CHIEFLY CHICKEN 5690 1 Chicken Noodle Dinner 5691 1 Chicken 0 la King_ 5692 1 Soup 5693 1 Chicken Chop Suey, etc. 5694 j N mb Prl T tal n :1 Ch ck 'f EGGS NONE D 5700 orb“: per 0:... AmoSnt Paid Size Grade Ungarad;d = 1 It 51 51 Check One Check One I I I O I X 3 Number of Price Total 1: 2:; ‘a '3 = 3 Am “ 2 E ; a FISH AND SEA man “57:22:... .23. p.13" 5:5 2 s E ‘3 5 NONEE] 5800 as v: u o a u. 0 Tuna 581 1 Salmon 5812 Other Fish 5813 .‘t Oysters 5820 Scallops 5830 in‘ Shrimp 5840 Other (name kind) 5850 MIXTURES — CHIEFLY FISH 5890 PREPARED BABY rooo 111) FRUITS first: .3132» .357.“ 4.1331.“ cm Applesauce 331 1 __9_ Apricots 3320 ___9__ Bananas 3340 ___9___ Orange Juice 3249 __9___ Peaches 3440 _‘3_ Pears 3450 _9___ Plums 3470 __9__ Prunes 3480 __9__ Other Fruits 3530 9 I Mixed Fruits 3590 9 ’5— Puddings 7420 T :2: VEGETABLES 9 Beets 4710 9 Carrots 4470 9 Green Beans 4440 9 Peas 4530 ‘9— Spinach 4260 9 Squash 4560 9 Sweet Potatoes 4570 9 Other Vegetables 4900 9 T Mixed Vegetables 4990 T 9 MIXED VEGETABLES AND MEAT 4992 9 9 9 9 MEATS, CHICKEN AND FISH 9 Beef 5150 9 Chicken 5610 9 Lamb 5250 9 “—Ijver 5141 9 Park 5360 9 Tuna 581 1 9 Veal 5450 9 Other 5500 9 _. .3 ll. "2’ BAKED GOODS AND OTHER GRAIN PRODUCTS NONE D Kind of Bread BREAD Number of Prlco par Total ! 1 I G Loaves Loot Amt. Pald What. 61 00 Whlto Wheat Rye Other Total Amount Amount Paid QUICK BREADS AND ROLLS 6200 Where convenient TOTAL CAKES 6310 Amount Paid may be re- COOKIES 6320 ported as price X quantity. DOUGHNUTS 6330 That is, if ou bu 3 doz. I' PIES (name kind) 6400 y Y i cookies at 35¢ per doz. you may report it in the PREPARED BAKED total amount paid column GOODS MIXES 6500 Cake Mix 6510 as 3 doz. X 35¢. Cookie Mix ' 6520 Quick Bread Mix 6530 Pancake Mix 6540 Pie Crust Mix 6550 W l P' M' 6 PLEASE hoe le Ix 560 Check Type of Other (name kind) 6570 Grain Below V ‘. Check ONE 1 N b Lb . Pric T t I s a a l s a 7 NONE D oily/3’01; per Potmd Am: Paid >~ ,- - OTHER GRAIN £~§§ g .3 g g g PRODUCTS 6600 a: a: 3 U 0 a: a: 3 Breakfast Cereals r r— All Other (name kind) _ _ ”-— (lncludes crackers, meal, popcorn, __ spaghetti, pretzels, noodles, etc.) Be sure to record your purchases on the same day they are made so that you don't forget any at them. Check the package for weight, etc. Be sure to record the proper price. Use the extra spaces for additional purchases. I Please don't forget to fill in the Vital Data Questions on page 15. ' - «f» we SUGAR, SWEETS, CANDY 3,: 6. ME, 7. . (13) SUGAR NONE D 7100 Number of Pounds Price per Pound Total Amt. Paid White or Powdered 7100 Brown 71 20 Maple 7130 SYRUP AND N h I P d HONEY NONE E] 7200 :rrI'Ii 31:. 6::3.‘ Price per Unit Total Amt. Paid Corn Syrup 7210 Cane Syrup 7220 Maple Syrup 7230 Molasses 7240 Sorghum 7250 Other Syrup 7260 Honey 7270 N b I P d CANDY NONE D 7300 “ZnJrgunczzn S Price per Pound Total Amt. Paid PR EPAR ED DESSERT MIXES NONE E] 7400 Gelatin, Jello, etc. 7410 Pudding 7420 Other Mix (name kind) 7440 ALL OTHER SWEETS NONE 1:] 7450 NUTS AND NUT PRODUCTS Check One '1 NONE D 7500 T335322! ”fitting “1:;le In Shetl Shelled Canned Coconuts 7530 Peanuts 7541 Peanut Butter 7542 Other Nuts (name kind) 7570 Have you included all of the food purchases by other members of the household? Do not include sales tax in reporting price or total amount paid. l " ”a 4! «en-1m r. .a we» as means wags-35w“ will? . ... «tam wwwE-EJEV .5 i 4 5 “ ’ BEVERAGES g Size of Unit : N mb 5 cif Pric T tal , NONE D 8100 oqunitsr Ozsielbz, etc. per Uiit Amo:nt Paid i Beer 81 10 I Liquors 81 20 E I t Wine 8130 ( Cocoa 8210 l‘ Cofl’ee 8220 Tea 8230 - ’ Soft Drinks—bottled 8310 ‘ ’ Soft Drinks—powdered 8320 NONEE] VITAMINS AND MINERALS VITAMINS (name kirid) 8400 Quantity Purchased Total Amount Paid MINERALS (name kind) 8500 COOKING AIDS Number . Size Price Total 8900 of Units of Unit per Unit Amount Paid 8911 8912 8969 8921 Extracts 8930 Meat Sauces Salt - EXTRA SPAC&,,(|f_or items not listed in diary) “5’ . 44- Number ‘ Size Price Total Ducrlpllofl "t of Units of Unit per Unit Amount Paid j VITAL DATA QUESTIONS 1. Has there been any change in your household membership since your last reporting week? YES NO (circle one) If yes, what was their: Relationship to homemaker Age How many are there in your household now? 2. How many regular meals were eaten away from home by members of your household last week?____ (One meal consists of either breakfast, dinner or supper for ONE person). Total amount spent 3. How many guest meals were served duringithe past week? (A guest is anyone not a regular household member). 4. What was the total income payment actually received during the diary. week by: The male and female head of the household? Other members of the household? Check if none 5. Was this before or after Federal income Tax deductions? Before ( ) After ( ) (In reporting income payments, please keep in mind that they might come from many sources. These include wages, salaries, commissions, pensions, interest and dividends, annuities, profit from business and professional services, profit from rent, government payments, gifts, and any other sources. This information will be held strictly confidential, and your name will not be associated with it. it is necessary to ask these questions in order to get the greatest yalue from your diary. W~~L h"w._,.——i w uobgqogw 'BUIEUD'I Isa; afieuo) mags uobgqegw leuod Jewnsuog '3 '3 °w qanueqaonp ‘9 ‘9 JOSSO}OJd ~ (010“ do” pewung euod) NO POSTAGE STAMP NECESSARY POSTAGE HAS BEEN PREPAID BY Professor Gerald Quackenbush 1 M. S. C. Consumer Panel ,. -‘ . Michigan State College '- East Lansing, Michigan Date Due mom “3E 0““ luLA‘w-L 1') "‘illilllllllll‘lllllES