CONSUMPTION PATTERNS UNDER CONDITIONS OF-RAPID ECONOMIC GROWTH:' ’ THE CASE OF :PUERTO RICO (1941’- 1965) Thesis for the Degree Of Ph. D. " MICHIGAN STATE UNIVERSITY ORLANDO T. ESPADAS 1970 ~~m LIBRAR Y Michigan State -» -. Diversity THE‘I‘ . 4,,- This is to certify that the thesis entitled CONSUMPTION PATTERNS UNDER CONDITIONS OF RAPID ECONOMIC GRWTH: THE CASE OF PUERTO RICO (1941-1965) presented by Orlando Tomas Espadas has been accepted towards fulfillment of the requirements for Ph.D. degree in Economics Datfi‘TYK/C l/ EN 0-169 Igflaqws IIIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII MSU LIBRARIES -_-. RETURNING MATERIALS: PTace in book drop to remove this checkout from your record. FINES wiTT be charged if book is returned after the date stamped below. I m: 8/1 '"fiT‘TfiT‘IS W7 1E 55 K215 Oifigééq X I D£C28"87§5§~ ', EZLI . {I II ABSTRACT CONSUMPTION PATTERNS UNDER CONDITIONS OF RAPID ECONOMIC GROWTH: THE CASE OF PUERTO RICO (19ul-l965) By Orlando T. Espadas The rapid rate of economic growth that characterized the Puerto Rican economy during the period from 1941 to 1965 provided the general background for stating several hypotheses regarding the behavior of consumption patterns under such conditions. Per capita income increased, in real terms, from $261 in l9UO to $900 in 1965. The main objectives of this investigation were to measure, using Time Series and Budget Studies data, the patterns of consumption for Puerto Rico during the period studied and to test several hypotheses regarding the sta- tistical significance of changes in the measured patterns of consumption at different periods of time (period 19N7—55 vs. 1956-65 and between budget studies: 19Nl, 1953 and 1963) and for different subgroups of the total population at the same point in time (urban vs. rural, wage earners vs. non-wage earners, San Juan vs. rest of Island). Orlando T. Espadas The introductory chapter presents a brief review of Pcunomlc growth in Puerto Rico, during the period covered by this study, as well as the specific hypotheses to be tested and related studies. Chapter II presents a historical review of progress in consumption analysis and discusses the theoretical foundations for the use of several variables used in con— sumption studies. Chapter III presents the results of the statistical analysis performed on the time series data (l9N7—l965) for approximately one hundred commodities or commodity groups. The statistical model employed was the multiple regression analysis with the data transformed into logarithms. The independent variables were: 1) Per capita income at constant prices 2) Relative Prices 3) Ratio of Non-Agricultural to Agricultural Employment (employed as a proxy for the structural changes that occurred in the economy during the period studied). In this chapter the hypothesis regarding rapid changes in the patterns of consumption was tested by means of the dummy variables technique. The coefficients were found, almost universally, to be not significantly different between the first period (1947-1955) and the second (1956- 1965). Orlando T. Espadas Chapter IV presents the statistical results and tests of hypotheses (t-test) performed on budget studies data. The budget studies were conducted in 19Ul, 1953, and 1963. The data for the 19u1 and 1953 budget studies were avail— able in summary tables by levels of income and this study analyzed eleven main commodity groups. The data for the 1063 budget study were available in four magnetic tapes containing all the information collected for each individual family. In this case thirty seven non-food and twenty five food items were included in the analysis. The statistical model was the same as in the time series data, multiple regression with observations trans- formed into logarithms but with different explanatory variables. The independent variables were: family total expenditures and family size. Regressions were performed for the following popula— tion groups: all families in Puerto Rico, wage-earners' families, non-wage earners' families, residents of San Juan, residents of the rest of Island, residents of urban areas, residents of rural areas, families with an annual income smaller than $3,000 and families with an annual income greater than $3,000. Several hypotheses were tested regarding the statis- tical significance of the difference in the coefficients derived for each set of regressions. In most cases coefficients were found to be significantly different Orlando T. Espadas between sub-groups at one point in time and between same sub—groups of total population at different points in time (19u1, 1953 and 1963). CONSUMPTION PATTERNS UNDER CONDITIONS OF RAPID ECONOMIC GROWTH: THE CASE OF PUERTO RICO (l9h1-l965) By Orlando TL Espadas A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Economics 1970 ACKNOWLEDGEMENTS This study and its author have received COOperation from many individuals and organizations for its successful completion. I would like to thank Professor Robert D. Stevens, member of the thesis committee, for his continual guidance and encouragement throughout the entire investi- gation as well as for his help in obtaining financial assistance for this study. The results of this study were greatly improved by his painstaking interest in the clarity of style and content. I would also like to acknowledge the kind cooperation of Professor Thomas R. Saving, who was the original Chairman of the thesis guidance committee and regretfully left Michigan State University before the completion of this study. Professor Saving helped in the original design and early statistical work. Dr. John P. Henderson kindly took over the Chairmanship and his sug- gestions have improved the clarity of presentation. The measurements and calculations performed in this investigation were a very tedious and difficult undertaking. Professor Roy Gilbert, member of the thesis committee, helped a great deal in this long process by performing most of the computer programming and by providing guidance ii in some of the problems of statistical measurement encountered. I would also like to express my gratitude to Professor Garland Wood, former Director of the Latin American Studies Center of Michigan State University, for the encouragement and financial assistance provided during the early stages of this study as well as throughout all my studies at Michigan State University. In the area of data collection I received invaluable assistance and cooperation from Rubén A. Vilches and Carmelo Fortufio of the Office of Labor Statistics of the Puerto Rico Department of Labor. I would also like to express my appreciation for the cooperation of Professor Salvador M. Padilla, Director of the Graduate Program in Planning of the University of Puerto Rico, in financing the last stages of this inves- tigation. The Department of Agricultural Economics of Michigan State University.made a contribution by administering a grant made to this study by the Center for International Programs of Michigan State University. Many other persons helped this investigation in various ways during the long process of writing this dissertation. My thanks to all. iii TABLE OF CONTENTS Page ACKNOWLEDGMENTS . . . . . . . . . . . . ii LIST OF TABLES . . . . . . . . . . . . . vi LIST OF APPENDICES . . . . . . . . . . . viii Chapter I. INTRODUCTION . . . . . . . . . . . 1 Brief Review of Economic Growth in Puerto Rico . . . . . . . . . 1 Objectives of the Study . . . . . . . 7 Related Studies . . . . . . . . . 10 II. THEORY AND VARIABLES USED IN EMPIRICAL STUDY OF CONSUMPTION . . . . . . . . 13 Review of Progress in Comsumption Analysis . . . . . . . . . . . l3 Engel' 3 Law . . . . 13 Other Empirical Laws of Consumption . . . 18 Theoretical Considerations Regarding Variables Used in Consumption Analysis . 20 Summary . . . . . . . . 3“ III. TIME SERIES ANALYSIS OF PERSONAL CONSUMPTION EXPENDITURES o o o c o o o o o o o 36 Data Used in Statistical Analysis . . . 37 Statistical Model . . . . “0 Relative Importance of Main Commodity Groups . . . . A3 Major Results of Statistical Analysis . . US Difference in Consumption Patterns Between l9u7- 1955 and 1956-1965 . . . . . . 59 Summary . . . . . . . . 61 iv Chapter Page IV. BUDGET STUDIES ANALYSIS OF HOUSEHOLD CONSUMPTION EXPENDITURES . . . . . . . 63 Review of Data Sources . . . . . . . 6A Statistical Model . . . . . 71 Relative Importance of Main Commodity Groups . . . . . ' 7U Changes in EXpenditure Elasticities (19u1 to 1963) . . . . . . . . 76 1963 Budget Study Results . . . . . . 79 Effects of Grouping Data . . . . . . 93 V. SUMMARY AND SUGGESTIONS FOR FURTHER RESEARCH. 97 Summary .‘ . . . . . . . . . 97 Further Research . . . . . . . . . 102 BIBLIOGRAPHY . . . . . . . . . . . . . 107 APPENDICES . . . . . . . . . . . . . . 115 Table 10. 11. LIST OF TABLES Growth in Per Capita Income, Population and Average Annual Migration, Five-Year Intervals, 1940-65 . . . . Net Income by Major Economic Sectors, Puerto RICO, 19140-1965 0 o o o o o o 0 Employed Persons by Major Industries, Puerto Rico, 19N0—l965 . . . . . . . . . Per Capita Income at Current and Constant (195“) Prices, Puerto Rico: 1947-1965 Urban-Rural and Non-Agricultural to Agri- cultural Employments Ratios, Puerto Rico: 19u7-1965 o o o o o o o o o o Relative Importance of Consumer Durables, Non-Durables, Services and Main Commodity_ Grgups Selected Years: 19A7, 1956 and ' 19 5 o o .o o o o o o o o o 0 Regression Results for Durables, Non-Durables, Services, & Main Commodity Groups: 191‘7'1965 o o o o o o o o Income Elasticity Derived from Time Series Data and Arranged According to Their Values . . . . . . . . . . Relative Price Elasticity Arranged According to Their Values . . . . . . . Statistical Significance of Difference in Coefficients between Period 19A7— 1955 and 1956- 1965 . . . . . . . . . . Relative Importance of Main Commodity Groups: 19Ul, 1953 and 1963 Budget Studies . . . . . . . . . . vi Page 38 41 NA A6 “7 “9 51 75 Table Page 1?. Expenditure Elasticities of Main Commodity Groups from 19Nl, 1953 and 1963 Budget Studies . . . . . . . . . . . . 77 13. Significance of Difference in Coefficients , (t-tests) for Various Sub-groups . . . . 80 1”. Expenditure Elasticities for Puerto Rico Arranged According to Their Values-~1963 Budget Study (Individual Observations). . . 82 15. EXpenditure Elasticities Derived from 1963 Budget Study . . . . . . . . . . . 8H 16. Family Size Elasticities for Puerto Rico Arranged According to Their Values—~1963 Budget Study (Individual Observations) . . 86 17. Family Size Elasticities Derived from 1963 Budget Study . . . . . . . . . . . 89 18. EXpenditure Elasticities, From Weekly Data, for Puerto Rico Arranged According to Their Values--1963 Budget Study . . . . . 91 19. Comparison of Regression Coefficients Based _ on Grouped Data and Individual Observations for Total Expenditures and Family Size (1963 Budget Study). . . . . . . . . 9N vii LIST OF APPENDICES Appendix Page A. Time Series Results (19A7-1965) . . . . 116 B. Time Series Results (1947-1955) . . . . 121 C. Time Series Results (1956-1965) . . . . 126 D. 1963 Budget Study Results (Ungrouped Data) 131 E. EXpenditure Elasticities for Selected Individual Food Items (Based on a Week's Expenditures) . . . . . . . . . 1N9 F. 1963 Budget Study Results (Ungrouped Data) (Based on Weekly Consumption Estimates). 15A viii CHAPTER I INTRODUCTION Brief Review of Economic Growth in Puerto Rico Puerto Rico experienced a very high average rate of economic growth during the period from 1940 through 1965. Per Capita Income, in 1965, prices more than tripled from $261 in 19uo to $900 in 1965.1 This increase in per capita income implies an annual average compound rate of growth of approximately five percent for the twenty-six year period.2 Despite the very rapid average rate of economic growth, considerable fluctuation occurred. This can be appreciated when the period is sub-divided into five-year intervals (Table l). The contraction in the fiVe-year 1Puerto Rico, Junta de Planificacién, Negociado de Analisis Econémico y Social. Informe Econ6mico a1 Gober- nador: 1965. San Juan, Puerto Rico, 1966, p. 5. 2For further details on the economic development of Puerto Rico see: The Long Uphill Path: A Historical Study of Puerto Rico's Program of Economic Development by David F. Ross, Talleres Grgficos Interamericanos, San Juan, P.R., 1966; Fomento: The Economic Development of Puerto Rico by William H. Stead, National Planning Association, 1958, and covering the period up to 196“, Wages, Productivity and Industrialization in Puerto Rico by Lloyd G. Reynolds and Peter gregory, Richard D. Irwin, Inc., 1965, Chapter I, pp. 3- 0. TABLE I.——Growth in Per Capita Income, Population and Average Annual Migration, Five-Year Intervals, lQUO-65. Annual Compound Rate of Growth of Per Annual Rate of Average Net Capita Income at Population Annual Period Constant Prices Growth Migration 19A0-A5 8.8 1.8 A,600 l9H5-50 0.0 1.6 31,000 1950-55 5.8 .3 50.700 1955-60 u.3 .9 39,700 196C665 7.2 2.1 9,000 , Source: Puerto Rico, Junta de Planificacién, Informe Economico al Gobernador: 1965, San Juan, Puerto Rico, p. 6. Note: The apparent overlap between the end of a period and the beginning of another is due to the use of fiscal year statistics. period following World War II was due mainly to a large scale reduction in federal expenditures following the end of hostilities. The significant effect of large scale migration to the continental United States can also be seen. The impact of this migratory flow in the economic development of Puerto Rico could, perhaps, be analyzed in three broad fields. First, the impact on the rate of growth of per capita income by reducing the rate of population growth (Table l). Secondly, the effect of remittances to Puerto Rican relatives from those residing in the United States. Lastly, the contribution, in terms of new skills and higher productivity, of those returning to live in Puerto Rico. Another significant point, indicated by Table l, is the acceleration of economic growth during the sixties. The rate of per capita income growth increased to 7.2 per cent. This increment is more significant when account is taken of the increase in the rate of population growth due to a reduction in the migration flow. The economic transformation that occurred in Puerto Rico is indicated by the large increases in per capita income. It was largely the result of a whole set of developmental policies emphasizing the attraction of United States manufacturing enterprises to Puerto Rico. These programs became generally known as "Operation Bootstrap." Large structural changes, in terms of output and employment, occurred in the Puerto Rican economy during the period 19N0-1965 as a result of this very rapid rate of development. Puerto Rico had a predominantly agricultural economy in 19N0 (Table 2). At that time, 31 percent the net national income was produced in the agricultural sector while the manufacturing sector produced only 12 per cent. In 1965, on the other hand, manufacturing composed 2A percent of net national product while agriculture con- tributed only 8 percent. The turning point when income created by manufacturing became greater than income produced in the agricultural sector occurred in 1955. TABLE 2.-—Net Income by Major Economic Sectors, Puerto Rico, (Millions of Dollars) Figures in Parentheses indicate Percentage of Total 19A0 1950 1960 1965 Agriculture 70 (31) 1A9 (2N) 180 (13) 190 ( 8) Manufacturing 27 (12) 89 (1N) 289 (21) 5A8 (23) Trade 26 (11) 102 (16) 237 (17) 805 (17) Services 21 ( 9) .NA ( 7) 132 ( 9) 276 (12) Government 19 ( 8) 70 (11) 175 (12) 307 (13) Other 62 (27) 160 (26) 3A2 (25) 563 (2A) Total 225 (100) 61A (100) 1355 (100) 2289 (100) Source: Puerto Rico, Junta de Planificacién, Informe Econémico al Gobernador: 1966, San Juan, Puerto Rico, p. A-U. Agricultural employment predominated over manufactur- ing throughout most of the period (Table 3). Only after 1965 did employment in manufacturing become greater than employment in agriculture. Thus, the period which began with predominance of the agricultural sector ends with a reversal of the original situation. Puerto Rico had changed in the brief time span of twenty-five years, from a poor- underdeveloped agricultural society to a relatively highly industrialized one whose leading sectors are found in the manufacturing and service industries. TABLE %.——Hmponed Persons by Major Industries, Puerto Rico, 19N0—1965. (Thousands) Figures in Parentheses Indicate Percentage of Total 1990 1950 1960 1965 Agriculture 229 (AU) 21A (35) 12A (23) 119 (17) Manufacturing 56 (10) 55 ( 9) 81 (15) 119 (17) Trade 59 (10) 9O (15) 97 (18) 123 (18) Services 73 (1”) 77 (12) 75 (19) 102 (15) Government 13 ( 2) 45 ( 7) 62 (ll) 87 (13) Other 87 (16) 115 (19) 10“ (19) 130 (19) Total 512 (100) 596 (100) 593 (100) 680 (100) Source: Puerto Rico, Junta de Planificacién, Informe Economico a1 Gobernador: 1966, San Juan, Puerto Rico, p. A-22. Mention should also be made of the demographic changes during the period studied. Puerto Rico has a high popula- tion density. There are approximately 700 persons per square mile. In 19A0, Puerto Rico had a population of 3 A approximately 1,869,255 which increased to 2,626,000 in 1965. This represents an average rate of growth of 1.A percent for the period from 19A0 to 1965. As indicated 3Elviro Borrero and Jose L. Vazquez, The Population 9! Puerto Rico (1500—196A). School of Medicine, University of Puerto Rico, no date, p. I ‘Based on estimates made by Mr. Severo Rivera of the Puerto Rico Planning Board. earlier, this somewhat low rate of population growth is due to the large scale migration to the United States. A total of 670,000 Puerto Ricans left the island during the period lQHO-65 (Table 1), almost as many as the absolute increase in population during the same period. These figures indi- cate that the rate of population grOwth, from 19AO to 1965, was approximately half of what it would have been without migration. Perhaps less than half if one considers the fact that a large proportion of those migrating were adults with reproductive potential. Approximately 30 percent of the population was defined as urban in 19A0 compared to 50 percent in 1965.5 Of, perhaps, more significance than the statistical increase in urban population is the tendency towards greater homo— geneity in the population of the entire island. The small— ness of the island (A0 miles wide by 100 miles long) coupled with the development of a good road system of approximately 3,AOO miles and a very large number of motor vehicles (87 vehicles per mile of road) are factors lessen- ing regional differences. Another factor that could be important in the process of diminishing differentiation between the urban and rural sectors of the population is the effort by Fomento (gov— ernmental agency in charge of promoting establishment of new industries in Puerto Rico) to locate new industrial k 5 Op. cit., p. 5. plants outside the San Juan Metropolitan area. Greater tax incentives and other benefits are available to firms willing to locate in other regions. Other communication channels, such as television, are available to the rural as well as the urban population. A very successful rural electrification program has made electricity available to approximately 95 percent of the families in Puerto Rico and with it the potential for purchasing and using all the modern home equipments and appliances such as refrigerators, washing machines, elec- tric heaters, stoves, televisions, high fidelity consoles, etc. Under these conditions, one may assume, that the patterns of consumption are going to be influenced to a diminishing degree by place of residence. It is hypothe— sized that a family with a given level of income living in a rural area of Puerto Rico would have a pattern of consumption quite similar to a family living in an urban area with the same level of income. Of course, there would always be some differences because of relative prices or availability of certain commodities, i.e., housing cheaper in rural areas, sport events available in urban areas. Obiectives of the Study Given these major economic changes in Puerto Rico during the last twenty—five years, it is expected that significant changes have also occurred in the patterns of consumption. The statistical measurement of such changes in the patterns of consumption, from 19Al to 1965, is the first major objective of this study. Three budget studies (lQAl, 1953 and 1963) and time series data (19A7 to 1965) are used to achieve this objective. The direction and rate of change in the consumption pattern of a detailed number of commodities and commodity groups will be deter- mined for the whole population and for more homogeneous sub—groups such as San Juan and Rest of Island, Urban and Rural Wage Earners and Non-Wage Earners, etc. The other major objective of this investigation is to test several hypotheses about consumption in Puerto Rico. The hypotheses contemplated in this study can be sub-divided into two categories. The first category includes testing whether the measured patterns of consump- tion are significantly different at different points in time. The specific hypotheses tested under this category are: 1. The pattern of consumption derived from the 19A1 budget study is significantly different from that derived from the 1953 budget study. The pattern of consumption derived from the 1953 budget study is significantly different from that derived from the 1963 budget study. R} 3. The pattern of consumption derived from the 19A1 budget study is significantly different from that derived from the 1963 budget study. A. The pattern of consumption derived from the analysis of the time series data corresponding to the period 19A7-1955 is significantly different from that derived from the analysis of the data corresponding to the period 1956- 1965. The second category of hypotheses include those referring to the testing of consumption patterns of certain sub—groups of the sample to determine if they are signifi- cantly different at the same point in time, for example, comparing coefficients derived from the 1963 budget study for Urban and Rural residents. The specific hypotheses are: l. The consumption pattern of urban dwellers is significantly different from that of rural dwellers. 2. The consumption pattern of residents of San Juan MetrOpolitan Area (capital of Puerto Rico) is significantly different from that of residents in the Rest of Puerto Rico. 3. The consumption pattern of Wage Earners is significantly different from that of Non—Wage Earners. A. The consumption pattern of low income families ($3,000 or less yearly income) is significantly different from that of higher income families (more than $3,000 annual income). In the process of doing the statistical analysis required for this study, two particularly interesting statistical problems appeared and were explored. One concerns the nature of the difference between coefficients obtained from "grouped" and "ungrouped" data, when a multi- ple regression is used and the grouping is performed on the basis of one variable with the others allowed to vary without any restraint. The other statistical problem 10 relates to the interpretation of results when a large num- ber of observations with zero values are present. This problem appeared particularly in the analysis of the results derived from the data for weekly expenditures on individual.1xxxi items. Related Studies Several studies were conducted, during the period covered by this investigation, related to consumption in Puerto Rico. A brief description of the most significant studies follows: Patterns of Living in Puerto Rican Families6 was published in 19u9 based on research conducted in 19A6. The authors are Home Economists and therefore the emphasis of the study and type of data collected in their question— naires was in terms of adequacy of diets, number and types of utensils used in homes, housing conditions, etc. There was no emphasis on how consumers allocated their budgets and the prices paid for the goods. For these reasons, it was not used in this study. Budget Studies (19Ul, 1952, 1953, and 1963) The dates in parenthesis indicate the years in which the Department of Labor of Puerto Rico has conducted island wide studies of the income and expenditures of families in 6Lydia J. Roberts‘and Rosa L. Stefani, Patterns Of Living in Puerto Rican Families, University of Puerto Rico, Rio Piedras, P. R., 1939. 11 Puerto Rico. The 19A1 and 1952 Budget Studies included only wage earners' families because they were designed and used mainly to provide weights for the Cost of Living Index of Wage Harners' Families which is published by the Depart- ment of Labor. The Budget Studies of 1953 and 1963 include "All Families," that is wage and non-wage earners. These studies provide a wealth of material for economic analysis of a quantitative nature. There is one important drawback since the original data from the 19U1, 1952 and 1953 studies were damaged or discarded, all that remains are the summary tables. These budget studies will be discussed in more detail in the data section of the chapter presenting the results from budget studies. The Agricultural Experiment Station Study During 1962, the University of Puerto Rico Agricul- tural EXperiment Station collected 1029 questionnaires on family food purchases. Part of the data has been analyzed using multiple regressions and different functional rela- tionships that might fit the data. The dependent variable has been quantities purchased rather than expenditures on the items, for their interest is in making quantity pro- JOctions of estimated foodstuffs to be required in 1970. ROCcntly, results for fresh beef, pork and poultry were l2 7 published. Results for other food items are expected to become available in the near future. Time Series Data Prepared by Puerto Rico's Planning Board Data for approximately 100 different commodities were collected and published by the Planning Board. These time series served in 1966 as the basis for projections of durables, non-durables and services.8 These projections used the recently published work by Taylor and Houthakker9 as the model. Projections are planned for more commodity groups in the future. We may conclude this chapter by emphasizing that there are considerable data available on household consump- tion expenditure in Puerto Rico. It is assumed that these data would permit a rather comprehensive statistical ana- lysis of the nature.and significance of changes in the patterns of household consumption in Puerto Rico from 19Al to 196r. {Rodriguez de Zapata, Lillian, "El Consumo de Carne de Res, Cerdo y Pollo en los Hogares de Puerto Rico," M. A. Thesis in Economics, University of Puerto Rico, May 1969. 8P. R. Planning Board, Bureau of Economic and Social Analysis, Consumption in Puerto Rico, Technical Report No. 2 San Juan, P. R., August 5, 1966. ( p )H. S. Houthakker and Lester D. Taylor, Consumer ngand in the United States, 1929-1970: Analyses and Projections (Cambridge, Harvard University Press), 1966. CHAPTER II THEORY AND VARIABLES USED IN EMPIRICAL STUDY OF CONSUMPTION Review of Prggress in Consumption Analysis Attempts to determine, empirically, the manner in which consumers allocate their income among available 0 0 l commodities go back to more than one hundred years. Engel's Law The first and probably best known of all empirical consumption studies was made by Ernest Engel in 1857. Engel used data derived from a survey of the income and expenditures of 153 Belgian families. On the basis of his analysis of the data, Engel proposed the following law of consumption: "The poorer a family, the greater the proportion of its total expenditure that must be devoted 1For an excellent survey of the empirical studies of Consumer behavior up to World War I, see George J. Stigler, "The Early History of Empirical Studies of Con- sumer Behavior," Journal of Political Economy, LIII (August, 1935), “33-A8l. For another excellent survey article with many references on the history and development of studies to determine consumption patterns, see Dorothy W. Douglas, "Family Budgets," in the Encyclopedia of the Social Sci— ences, Vol. V-VI, pp. 73-78 (1931). McMillan Co., New York. 13 1" to the provision of food."2 This statement has become generally known as "Engel's Law" and the functional rela- tionship between purchased quantities of a commodity and the level of income are identified as "Engel Curves." "Engel Curves" are theoretically derived from the income—consumption curve used in economic analysis. The income-consumption curve shows (Figure l) the different equilibrium combinations of commodities (X and Y below) that would be purchased at various levels of income, assuming constant prices, tastes, etc. Y L3 L2 Y3 Y") L1 \ III Y1 U 1 1 X 0 X1 X2M1X3 M2 M3 Figure l (IbiCL, p. 98. For example, given the indifference map and the income restraint represented by the line L1 Ml (Figure l), the consumer would maximize his satisfaction at point A. The amounts purchased would be OX of X and OY of Y. When 1’ the maximizing l: the income restraint is changed to L2 M2, position will be at point B and the amounts purchased will be OX2 of X and OY2 of Y. Once we have plotted the equilibrium points for various levels of income, we can proceed to draw an Engel Curve from the above income—consumption curve. The pro- cedure now involves putting income on one axis and quanti- ties purchased of the particular commodity on the other, as in Figure 2. Income 1X0H3 N P OH x X X X Quantity of X Figure 2 16 Then by plotting a point, corresponding to the level of income and the quantities purchased, for each of the equilibrium situations on the income—consumption curve we would have an Engel curve for this particular commodity and consumer. Figure 3 indicates three general forms that the Hngel curve might take. Income Quantity of X Figure 3 17 Curve 1 shows that as income increases quantities purchased of Commodity X do increase but less than proportionally, that is, the income elasticity would be positive but less than one. Curve II indicates that consumption of the good increases proportionally with income, that is, the income elasticity is equal to one. Curve III shows that quantities purchased increased more than proportionally with incre- ments in income which would imply an income elasticity greater than one. income elasticity is defined as: AQ A that is, percentage changes in Ye = —— e —% quantity divided by percentage change in income. where: Q = Original quantity of commodity purchased. A0 = Change in quantity purchased after change in income. Y = Original level of income. AY = Change in level of income. The concept of income elasticity can also be used to classify commodities into "necessities" and "luxuries." Commodities are considered necessities when their consump- tion remains about the same irrespective of changes in income, that is, their income elasticity is positive but less than one. On the other hand, an income elasticity greater than one indicates that the commodity is a "luxurY-" 18 Other Empirical Laws of Consumption Several other empirical "laws of consumption" were promulgated during the second half of the nineteenth cen- tury with regards to the changing proportions of expendi- tures in a commodity or commodity group out of total income. Hngel's law, with regards to food expenditures, was followed, in 1868, by "Schuabbe's law." It states: "The poorer anyone is, the greater the amount relative to "3 his income that he must spend for housing. Later, in 1875, the Commissioner of Labor Statistics of Massachusetts, Carrol Wright,Ll translated the original Engel's law into the following form: The distinct propositions are: First, that the greater the income, the smaller the relative per- centage of outlay for subsistance. Second, that the percentage of outlay for clothing is approxi- mately the same, whatever the income. Third, that the percentage of outlay for lodging, or rent, and for fuel and light, is invariably the same, what- ever the income. Fourth, that as the income increases in amount, the %ercentage outlay for sundries becomes greater. in terms of income elasticities the above would say that it is less than one for food, approximately one for clothing, exactly one for housing and greater than one for I 'Haundries' or luxury goods. 3Stigler, op. cit., p. 100. 1 ‘For a discussion of the evolution of Engel's ori— EiJial law and the one generally attributed to him including ccnnmodities other than food, see Stigler, op. cit. 19 Many studies were subsequently undertaken to test the validity of the above-mentioned empirical laws of consumption. Of special interest was the study made by Houthakker6 in which MO surveys of incomes and expenditures of families (budget studies) from 30 different countries were analyzed and Engel's original law referring to food expenditures was confirmed by all. Progress in the empirical derivation of demand, as 7 pointed out by Lee was slow until the publication, in 1935, of the work of Allen and Bowley8 using British family budgets data. Some of the most significant contri— 9 using U. S. agricultural butions have been made by Schultz, time series data; Wold and Jureen,lO using Swedish budget studies and time series data; Stone,ll using British time 6Hendrik S. Houthakker, "An Internation Comparison of Household Expenditure Patterns Commemorating the Cen- tenary of Engel's Law," Econometrics, XXI (October, 1957), 532-551. 7Feng-Yao Lee, An Econometrict Analysis of the Demand for Basic Living Materials in Japan, unpublished doctoral dissertation, Michigan State University, 1965, p. 10. 8Ray G. D. Allen and Arthur L. Bowley, Family Expen— diture (London: Staples, 1935). 9Henry Schultz, The Theory and Measurement of Demand (Chicago: The University of Chicago Press, 1938). loHermand Wold and Lars Jureen, Demand Analysis (New Ykark: John Wiley and Sons, Inc., 1953). 11Richard Stone, The Measurement of Consumers' Ex enditures and Behavior in the United Kin dom 1920—1938, V131. I (Cambridge University Press, 1955). 20 series data; Prais and Houthakker,l2 using British budget studies data; and more recently Houthakker and Taylor13 using U. S. time series data. A review of most of the above contributions is found in two excellent survey 15 articles by HoodlLl and Ferber. Theoretical Considerations Regarding Variables Used in_Consumption Analysis The number of variables that determine the pattern of consumption for a given commodity is quite large. They range from the two most emphasized in the theory of consumer demand, that is, income and prices, to an extremely large number of variables referred to as "tastes and pre- ferences" in economic theory. This larger set of variables includes family size, age of the head of the family, occu- pation of the head of the family, level of education, wealth of the family, eXpectations, place of residence (urban-rural or other geographical breakdown), availability ¥ 128. J. Prais and Hendrik S. Houthakker, The Analysis gngamily Budgets (England: Cambridge University Press, 1955). 13Hendrik S. Houthakker and Lester D. Taylor, Sflggsumer Demand in the United States, 1929-1970: Analysis égyi Projections(Cambridge: Harvard University Press, 1966). I llWilliam C. Hood, "Empirical Studies of Demand," Ckinadian Journal of Economic and Political Science, XXI (August , 1955) , 309-327 . 15Robert Ferber, "Research on Household Behavior," flflfiirican Economic Review, LII (March, 1962), 19—63. 21 of new products in the market, attitudes of consumers, previous consumption patterns, consumption patterns of other consumers, advertising and probably innumerable less obvious variables that psychologists try to discover by depth interviews and other consumer research methods. Some of the above-mentioned variables, also referred to as "nuisance variables," present very difficult statis- tical problems in their quantification and analysis. Even in the cases when the variable can be reasonably quantified, there still remains the statistical problem of the high correlation of most variables with the level of income of the consumer unit. These reasons make it impossible to include all the above-mentioned variables in a demand study. Another factor that limits the number of variables to be used is the nature of the data available. Some variables, such as prices, are better analyzed from time series data. Others, such as family size can be better studied from cross-section data.l6 Below we will discuss, first, the most important variables from the point of view of traditional demand theory; then, other variables considered relevant and used 131 this study. ¥ 16For a discussion on this subject see Vernon G. IJippitt, Determinants of Consumer Demand for House Furn— ishings and Equipment (Harvard University Press, 1959). pp. - o 22 Traditional Demand Theory The traditional presentation of Demand Theory assumes that consumers maximize a utility function of the following form: U = U (Xi) i = l,...,n where the Xi are rates of consumption of the various goods and services in the community, subject to an income con- straint such as where Y is income and Pi are the prices of the various goods and services (Xi). From this maximizing process, we can derive demand functions for each of the goods and services. These demand functions would be functions of Income and Prices. This is the reason why the most important variables in empirical consumption studies are income and prices. i Income.-—This is considered the most important of all variables explaining consumer behavior. However, from an empirical point of view, there is considerable discussion as to the appropriateness of its use, as it is extremely (iifficult to obtain reliable estimates. A major reason is El tendency for people to under-report their income to the imiterviewers for fear that the information might be used 23 somehow against them, by tax collectors or others. One would always expect a rather substantial downward bias in the income estimates. In order to avoid this difficulty, some prominent authors have substituted Total Expenditures for Income as the principal independent variable in their demand functions.17 Another argument in favor of using total expenditures instead of income is derived from Friedman's permanent income hypothesis.18 It is argued that total expenditures s a better explanatory variable of consumer behavior than measured income because it reflects future expectations of the consumers with regard to their permanent economic con- dition rather than the short run, transitory, aspects that would be measured if present income is used. Friedman considers that measured income (y) and con— sumption (c) can be divided into its "transitory" (t) and "permanent" (p) components so that: l7R. Stone, op. cit., R. G. D. Allen and A. L. Bowley, op. cit., and S. J. Prais and H. S. Houthakker, op. cit. 18Milton Friedman, A Theor of Consu_ption Function (Princeton University Press, 1957 . This theory has been widely discussed by many economists, in particular Franco Modigliani and Richard Brumberg, "Utility Analysis and the Consumption Function," in Post-Keynesian Economics, Kenneth K. Kunihars (Editor), (Rutgers University Press, 195“), pp. 388-u36; Margaret G. Reid and Marilyn Dunsing, "The Effect of Variability of Income on Level of Income Expenditure Curves of Farm Families," Review of Economics and Statistics, XXXVII (February, 1956)} pp. 90— -95; Irwin Friend and Irving B. Kravis, "Consumption Patterns and Permanent Income," Proceedings of the American Economic Review, XLVII (May, 1957), pp. 5H8-555; H. S. Houthakker, 2H 0 I! O + 0 It is further assumed that permanent consumption constitutes a permanent prOportion of permanent income: And that the permanent and transitory parts of income and consumption are uncorrelated. As a result of the above mentioned propositions, a consumer is assumed to make his decisions not on the basis of his current income but rather on his expected lifetime or permanent income. The discussion initiated by Friedman's theory has not concluded yet. In any case, this theory is a major reason for using Total Expenditures as the key explanatory variable in this study. This theory appears to be substantiated by the Puerto Rican data. The 19Ul, 1953 and 1963 Budget Studies show that low and middle income families were spending more than their total income for the year. It could be assumed that they had expecta- tions of higher incomes in the future. "The Permanent Income Hypothesis," American Economic Review, XLVIII (June, 1958), pp. 396-N0u; and Robert Eisner, "Permanent Income Hypothesis: Comment," American Economic Review, XLVIII (December, 1958), pp. 972-990. For a 800d brief discussion of Friedman's Permanent Income Hypothesis, see Robert Ferber, op. cit., pp. 25-32. 25 Prices.-—As indicated earlier, this variable is of fundamental importance in explaining consumer behavior according to the traditional demand theory. It is used only in the time series analysis of demand relationships. The reason for not considering it in the cross section analysis stems from the fact that the cross section data is collected at a given point in time and it is assumed that all consumers are facing the same set of prices. How- ever, this does not imply that all consumers paid exactly the same prices. Variations in price do exist at a given point in time for a given good due to market imperfections and product differentiation. I Other Variables Used in Empirical Demand Analysis The traditional theory of demand leaves something to be desired in explaining consumer demand because: (1) it assumes that every individual is a decision maker, and that (2) the income constraint, as shown above is the only constraint faced by the relevant decision makers. These problems suggested a few additional variables for consideration. They are: Family Size.--In most communities, the Family is the decision-making unit rather than the individual. Essen— tffally, this means that the utility function of the head 0? the family will not be independent of whether or not he 26 has a wife or the number of children. That is, a given bundle of goods will yield a different level of well being for an individual if he is a bachelor than if he has a wife and five children. The larger the number of people in a household, the greater the expenditures on food, clothing, housing, etc. However, the relationship between family size and expendi- tures need not be proportional. Houthakkerlg calculated such a relationship and obtained a family size coefficient of approximately .2 in a double logarithmic equation, which suggests significant economies of scale. According to Houthakker, if a family with ten members has a new child which represents a 10 percent increase in its size, one should expect an increase of approximately 2 percent in the total expenditures of this family. The economies of scale result from the indivisibility of goods, greater use of clothes as outgrown pieces are used by other members of the family, and families of different sizes usually have kitchen, bathroom, patios, living rooms, etc. of approxi- mately the same size. Another factor that could explain the low family size coefficient is that large families usually have more Children and living expenditures of children are less than for adults. For these reasons Family Size is also ‘ 19Houthakker, 0p. cit. 27 considered an important determinant of consumer behavior and is used in this study. Place of Residence.--Another significant determinant of consumer behavior is the constraints placed on the individual decision maker by the physical environment and the nature and extent of goods available to him. The difference between the patterns of consumption of two consumer units exactly the same in every respect except that one lives in a city and the other in a farm could be explained by the following: (1) Information about goods and services is cheaper in urban areas, i.e., because of economies of scale in dissemination of information. (2) Relative prices are different in cities. Picnics in the meadow becomes costlier in the cities than in rural areas. Museums are cheaper in the cities. }(3) There is a substantially bigger, non-market sector in rural areas than in cities. (U) The city environment would affect the utility function. One reason could be the availability of new goods and services. For the above reasons and others, it is generally accepted that place of residence is an important factor in determining patterns of consumption. The level of total expenditures and its composition seems to be different by 28 region. in particular, total expenditures of urban dwellers are found to be higher at any given income level than the average consumer living in a rural area. As development proceeds, the differentiation between the urban and rural areas share in the benefits of economic development. This proposition was analyzed in this study. Urban-Rural Ratio.-—This variable was introduced as a proxy for all the structural changes that have taken place in the Puerto Rican economy during the period studied. It was decided, a priori, that this variable could explain some of the consumption changes that have taken place. However, as we looked into the data, we found out that this was not probably as significant a variable as was originally expected. One reason was that since the Censuses in Puerto Rico are taken by the United States government, they use the same definitions of urban and rural as in the United States. A place is considered rural if it has less than 2,500 inhabitants. In Puerto Rico, according to Dr. Vazquez,2O many areas classified as rural by the Census could be considered urban if another set of criteria were used rather than mere number of inhabitants. So, on the basis of this argument, a considerable upward bias in the rural estimates could be expected. 2OIn conversation with the author, Dr. José L. Vazquez, Director of the Demographic Studies Center of the School of Medicine of the University of Puerto Rico. 29 Another point to mention is that the values for individual years were obtained by linearly extrapolating for the between Census years. Since the last Census was in 1960, we had to make some assumptions as to the behavior of population in order to extrapolate up to 1965. We have estimates of total population up to 1965 but not a break- down by urban and rural. We could have decided that the ratio existing in 1960 was going to keep constant for the period 1960-65. Or we could have decided that the ratio increased at a certain rate during the 50's and use this same rate of growth in the ratio for extrapolating up to 1965. However, we assumed that all increments in population since 1960 took place in the urban areas for the following reasons: (Table 2 in Chapter III shows the actual urban- rural ratios used.) During the 1990's the rural popula- tion increased by 1 percent while the urban increased by 58 percent. During the 1950's rural population decreased by 0.“ percent, so it was reasonable to assume that for the period 1960-65 all increases in population will go as in the last two decades, to the urban areas. Another fact that may affect the lack of usefulness of the location variable in explaining consumption changes in Puerto Rico is the geography of the island. Puerto Rico is approximately 90 miles wide by 100 miles long with a fairly good road and transportation system. Almost 3O everyone is within a two-hour driving time of an urban center. This phenomenom makes for a much more homogeneous population in terms of the urban-rural differences that are expected in some other countries where it is much harder for rural people to come to urban centers because of the larger distances or inadequate transportation systems. For these reasons, another variable was included to attempt to better reflect the structural changes that have taken place in Puerto Rico during the period studied. Ratio of Non-Agricultural to Agricultural Employ- mgpp.--The ratio of non-agricultural to agricultural employ- ment was considered because the source of income and employ- ment, rather than the place of residence, could better represent the structural changes occurring in an economy if for different reasons people remain living in rural areas as they change their employment from agricultural to non—agricultural, and if the physical constraints of living in rural areas are reduced or eliminated by roads, electri- city, schools, and other social services. It is not surprising that this situation has occurred in Puerto Rico. The smallness of the island and good trans- portation system mentioned earlier appear among the reasons for this situation in Puerto Rico. In another country there might not be a significant difference between these two variables and either one would yield approximately equal results. 31 Wage-Earners'vs. Non—Wage Earners'Familie§.--The data from the 1953 and 1963 budget study had this breakdown which we decided to use by running separate regressions for the wage earners and All Families to see whether the coefficients were significantly different. The assumption behind this breakdown is that the "basket" of consumer goods purchased by wage earners is different from other consumers who have other sources of income. It would have been more interesting to make the comparison between wage— earners and non-wage earners since "All Families" include, as a significant group, the wage-earners' group. In the 1963 budget study there was the option to run a separate set of regressions for "non—wage earners" and it was run to see whether the patterns of consumption varied signifi- cantly. Other Factors Influencing Consumption Patterns It is realized that the variables listed above does not include all the factors that did influence the decisions of households in Puerto Rico with regards to the allocation of their income among the available alternative consumer goods. Additional factors were excluded for either of two reasons: the variable was not quantifiable, or it extended the scope of the investigation beyond the resources avail- able to the researcher. A few of those that are considered most important but excluded follow. 32 Demonstration Effect.—-Because of the special poli— tical relationship between the United States and Puerto Rico, which makes Puerto Rico a part of the American economy, Puerto Ricans have been exposed to a large number of new consumption items as well as to the United States way of life. ‘The reasons being the great amount of travel between the two areas as well as large amount of communication and good transportation facilities at relatively low prices. Marketing Changes.--The retailing of foodstuffs and most consumer items has been revolutionized with the intro- duction, during the fifties, of modern merchandising and gadvertising techniques by large United States retail com- panies such as Grand Union, A & P and Pueblo Super Markets in food; and by Sears and other department stores in the retailing of most other consumer items. Today, it is dif— ficult to name a United States national retail company that does not have one or more outlets in Puerto Rico. It goes all the way from Zales Jewelers to McDonalds Hamburgers. The new stores were mainly located in the Metropolitan areas of San Juan, Ponce and Mayaguez. These Metropolitan areas comprised, according to the 1960 census, 39.8 percent of the total population. This figure, however, does not show the real impact of such marketing improvements because of the small size of the island, which makes it relatively easy for people outside the Metropolitan areas to go to them for their shepping. Another factor adding to the 33 mobility of people for their shopping, as mentioned earlier, is a relatively good road and transportation system on the island. This internal mobility develops an internal demon— stration effect which is the counterpart of the external demonstration effect received from the United States. In recent travels around the island the author saw girls wear- ing mini-skirts, go-go boots, and other new fashions in rural areas. So, in this particular instance, it could be noticed how rapidly these fashions moved from the United States to San Juan and from there to the rest of the island. Social Mobilipyand Status.--As in almost any other society, Puerto Ricans seek social status and recognition by their patterns of consumption. However, in the case of Puerto Rico, this factor may play even a more powerful influence due to the rapid transition from a poor to a rather well—to—do society. This rapid affluence may make the desire to show off the new richness much more powerful than it might be expected had the same affluence been acquired over a longer time period. Consumers' Wealth.-—This could be another important factor since one might argue that in 1991 the stock of consumer durables was very low due to lack of income and that as income began to grow it could have been used for the purchase of consumer durables; specially in view of the 3“ previously mentioned "demonstration effect" which is assumed to operate very strongly in Puerto Rico.21 Credit Availability.--Another factor that has contri- buted to increased purchases of consumer durables has been the development of credit institutions and the availability of relatively easy financing (even though at high interest rates) for such purchases. Summary This chapter presents: l. A brief review of early empirical studies of demand._ 2. The derivation of income curves and definition of income elasticity which would be useful in interpreting the results presented in Chapters III and Chapter IV. 3. The traditional demand theory was summarized and shown that it leaves much to be desired due to its emphasis on income and prices. U. Other variables used in the present study were discussed. These included family size, place of residence (urban, rural or geographical breakdown), variables used as proxies for the 21For a detailed discussion of how wealth affects consumer decisions, see Boris P. Pesek and Thomas R. Savings, Money, Wealth and Economic Theory (New York: The MacMillan Company, 1957), pp. 297-365? 35 structural changes occurring in the economy of Puerto Rico (urban-rural ratio and agricultural to non-agricultural employment ratio), and source of income (wage earners and non—wage earners). 5. Finally, other variables which were not included in this study but which would have been interest- ing to explore are briefly mentioned. They include the impact of the demonstration effect in the consumption patterns of Puerto Rico, significant marketing changes which occurred during the period studies, social mobility and the development of a large middle class in Puerto Rico, the extent of changes in consumers' wealth during the period and lastly the very significant increments in the availability of consumers' credit. The discussion so far presents the necessary back- ground information for the presentation and analysis of the findings of this study in its attempt to measure direction and extent of household expenditures for a substantial number of commodities based on time series and budget studies data. The next chapter presents the findings derived from the statistical analysis of time series data. CHAPTER III TIME SERIES ANALYSIS OF PERSONAL CONSUMPTION EXPENDITURES This chapter presents the results of statistical analysis performed on time series data available for approximately one hundred individual consumption items or commodity groups. Regression results for the complete period (1997—1965) are presented and discussed in terms of the postulates of economic theory. The complete period was divided into two sub-periods (1997-1955 and 1956-1965) and a test made to determine if the coefficients for the two periods are significantly different as postulated in one of the hypothesis of this study. The basic data used in this chapter was developed by the Planning Board of Puerto Rico. Data for an earlier period, from 19Ul to 1946, on consumption of certain aggre- gated commodity groups was prepared by Daniel and Henrietta Creamer.l However, it was decided not to use the data for 1Daniel and Henrietta L. Creamer, Gross Product of Puerto Rico, 19U0-l9u4, Social Science Research Center, Rio Piedras, University of Puerto Rico, 19U8. Daniel and Henrietta L. Creamer, Gross Product of Puerto Rico, l9uu-l9u6, Social Science Research Center, University of Puerto Rico, Rio Piedras, 19U9. 36 37 this period for two reasons. First, there were the war years, and therefore consumer preferences were greatly influenced by war scarcities and rationing which existed at the time. The scarcities and rationing in Puerto Rico were more severe than in the mainland due to a shortage of maritime transportation. The second reason for not using the limited data gathered for the period 19u1 to l9u6 is because it is not completely comparable with the data developed by the Planning Board. For these reasons data prior to l9U7 was not used in this time series analysis. . The year 1965 was selected as the terminating year because it was the last year for which there were statis- tics on consumption when the time series analysis was carried out. Nineteen years appeared to be a long enough period to make certain inferences about the patterns of consumption in Puerto Rico. Data Used in Statistical Analysis Details about the data used to quantify the variables employed in this analysis follow: Per Capita Consumption of Each Commodity Values for the years studied were obtained dividing total value of consumption of each commodity in a year by the population of Puerto Rico in that year and then adjust- ing this value for price changes by dividing it by its 38 price index (195“ base). The data thus obtained for the ninety-eight commodities or commodity groups is not pre- sented because it would be extremely lengthy and it is available from the Bureau of Economics and Social Planning of the Puerto Rico Planning Board. Per Capita Income To calculate the value of this variable, Total Dis- posable Personal Income was divided by the Population. The resulting value was divided by the general Price Index (195“ base), Table “. TABLE “.-—Per Capita Income at Current and Constant (195“) Prices, Puerto Rico: l9“7-l965. Per Capita Income General Price at Constant Per Capita Income Index Year (195“) Prices at Current Prices (195“ = 100) 19“? 289 299 89.5 19“8 290 267 92.2 l9“9 322 286 89.0 1950 338 290 85.6 1951 352 313 88.8 1952 385 366 95.2 1953 “O9 397 97.1 195“ “22 “22 100.0 1955 “38 “39 100.1 1956 “50 “55 101.2 1957 “6“ “81 103.5 1958 “75 508 106.9 1959 “93 537 108-9 1960 525 586 111.6 1961 562 6“l 11“.0 1962 607 702 115.7 1963 651 762 117.1 196“ 68“ 815 119.1 1965 720 872 121.1 39 Relative Prices The regular formula for this variable is: Pi/Po where P Price Index of the ith commodity. i P 0 General price index This way of defining relative prices could lead to biased results because the value of the numerator is also included in the denominator. This could be significant in the case of commodities constituting a large part of total expenditures such as food, housing, clothing, etc. In order to avoid this possible source of bias, the following form was used: Vni/Vbi Vno - Vni/Vbo - Vbi where: Vni = value of expenditures in year n for commodity i at current prices. Vbi = value of expenditures in year n for commodity i at base year prices. Vno = value of total expenditures in year n at current prices. VbO = value of total expenditures in year n at base year prices. By using this formula, the value of the numerator is com- pletely eliminated from the denominator and the bias elimi— nated. The result is then interpreted as the relative price Of commodity i with respect to all other commodities Eigluding i; instead of the regular relative price formula which refers to the price of commodity i relative to all “0 other prices including i. It is realized that for many, especially individual items, this adjustment is insignificant. However, since it was important for the aggregated commodity groups, the adjustment was performed for all items. Urban-Rural and Non-Agricultural to Agricultural Employment Ratios As indicated in Chapter II, these variables were con- sidered as proxies for the structural changes that occurred during the period studied. Statistical results were cal- culated for both; however, only the results using the Non- Agricultural to Agricultural Employment ratio are presented. This decision was based on the fact that the urban-rural ratio was of no statistical significance for a considerably larger number of consumption items than was the case for the ratio of Non-Agricultural to Agricultural Employment. Table 2 shows the values of these variables for the 19“7— 1965 period. Statistical Model The model used in this study to determine changes in the patterns of consumption in Puerto Rico is the single equation least squares with values transferred into logarithms so that elasticities are obtained directly from the regression coefficients.2 Another reason for using such a model was the conclusion reached by Wold and Jureen 2For a discussion of single equation least squares vs. simultaneous equations, see Lee's thesis, pp. 29-32. “1 TABLE 5.—-Urban—Rural and Non-Agricultural to Agricultural Employments Ratios, Puerto Rico: l9“7-1965. Urban-Rural Non-Agricultural to Agricul— Year Ratio tural Employment Ratio 19“7 .61 1.58 19“8 .63 1.66 19“9 .66 1.70 1950 .68 1.76 1951 .69 1.98 1952 .70 1.97 1953 .71 2.20 195“ .73 2.10 1955 .7“ 2.29 1956 .75 2.“7 1957 .76 2.63 1958 .77 2.68 1959 .78 2.99 1960 .70 3.3“ 1961 .8“ 3.21 1962 .87 3.21 1963 .92 3.27 196A .96 3.80 1965 1.00 “.55 Source: Commonwealth of Puerto Rico. Puerto Rico Planning Board. Bureau of Economic and Social Analysis. Selected Indices of Social and Economic Progress, San Juan, n. d., pp. 3-“. after a careful analysis of the advantages and drawbacks of the least square method. They summarized their discussion saying: "The final conclusion must be, no doubt, that the regression analysis as traditionally applied is essentially sound. In demand analysis, at least, it can still be safely recommended."3 The equation selected for the statistical analysis of the time series data represents a power function of the form. 3H. Wold and L. Jureen, op. cit., p. 59. “’3 l . _ bi C D Ci — ai . Yi . (Pi/Po)i i . (AgN/NonAgN)i i which when transformed into logarithmics take a ‘ linear form: log(AgN/NonAgN)i+U logC =A +b logY +c log(Pi/Po)i+di i i i i 1 Other models were tried such as a simple linear function, a semi—logarithmic model, and a regression using the data transformed into first differences which is recommended to deal with the problem of multicolinearity between the regressors as is the case in the Puerto Rico data. Because of limitations in time and resources to make a complete analysis of the alternative results plus the fact that the double logarithmic equation is normally used, it was decided to present the results only for the latter. Further research could consider other functional relationships-such as the semi—logarithmic, the logarithmic-inverse and the double— logarithmic inverse as well as the linear and first differ- ences to decide on an item by item basis which is the most appropriate in each case. We have provided a general and rather simple model for explaining consumer expenditures in Puerto Rico. The next effort, in our opinion, should be to refine the estimates obtained here in terms of the type of function and the variables selected for analysis. “3 Multicolinearity This term refers to the "degenerate situation in regression analysis where the regressors are highly inter- correlated so as to be themselves subject to a linear relationship."u This is a very serious problem in the time series data used in this study. Puerto Rico during the period studied has had, as stated earlier, continuous high rates of growth with the result that all statistical series are very highly positively correlated with each other. In the simple correlations matrix it was rarely that a correlation between the variables was less than ninety with the exception of relative prices which had a random character taking values from very highly positively corre- lated to very highly negatively correlated and all sorts of values in between depending on the items considered. This limitation should be kept in mind in interpreting the results presented in this study. Relative Importance of Main Commodity Groups During the period there were some significant changes in the relative importance of the main commodity groups (Table 6. As experience in other countries and economic theory would predict, there was a significant drop in the relative expenditures for food accompanied by increases of ’Wold and Jureen, op. cit., p. “6. ““ TABLE 6.—-Relative Importance of Consumer Durables, Non- Durables, Services, and Main Commodity Groups Selected Years: l9“7, 1956 and 1965. Items % of Total Consumption 1947 1956 1965 l. Durables (96) 6.7 10.3 15.5 2. Non-Durables (97) 65.8 60.6 53.2 3. Services (98) 27.5 29.1 31.3 1. Food & Tobacco (001) “7.9 “2.9 33.9 2. Clothing & Accessories (010) 9.6 9.8 11.3 3. Personal Care (018) 2.0 1.6 2.5 “. Housing (021) 8.“ 10.2 10.0 5. Household Operations (02“) 10.1 12.1 13.“ 6. Medical Expenses (0“8) 3.6 “.l 3.8 7. Personal Business (056) .9 1.0 2.“ 8. Transportation (061) 6.5 8.9 11.9 9. Recreation (072) 5.8 6.5 8.7 10. Education (086) .5 1.0 1.5 11. Religious & Non-Profit Organizations (090) 2 .3 .3 12. Foreign Travel (093) 3 7 3.9 5.“ Source: Calculated from original data in work sheets at the Bureau of Economics and Social Planning, Puerto Rico Planning Board. differing magnitude in all other commodity groups. This does not mean that all the individual non—food items experi— enced increases in their relative importance. Actual con- sumption of some non—food items declined, as will be seen in the cases of items with negative income elasticities, during the period. Since Table 6 refers to highly aggregated categories, the declines were over-shadowed by proportionately larger increases in other items within the same commodity group. For example, you may have the category "Recreation" “5 increasing in importance but the sub—category "Admission to Amusements" declining because people are watching more television. Consumer durables increased from 6.7 percent of the budget in l9“7 to 15.5 percent in 1965 while Non-Durables declined. Services increased a little in its relative importance. Based on the experience provided by the period studied and the patterns observed in more developed coun- tries, one may expect that the future pattern would probably be a continued decline in Non-Durables while the increments in Durables would be less than in the past and increases in Services would be larger than in the period l9“7-1965. Major Results of Statistical Analysis This section presents and discusses results of the statistical analysis for the main commodity groups (Table 7) and the individual consumption items (Tables 8 and 9). The results of the test5 for the significance of the differ- ence between the two sub—periods (Table 10) are also pre- sented and discussed. Detailed results for the ninety- eight individual commodities appear in Appendix A. The values of the coefficients obtained from the regression analysis are presented with their standard error and sta- tistical level of significance. Coefficients of {I 'For a detailed discussion of the statistical test used (t-test) and its formula, see: Freund, John E., Modern Elementary Statistics, Third Edition (Prentice— Hall, Inc., Englewood Cliffs, New Jersey, 1967), pp. 250—257. “6 .opom Eoem pcosmmwflo hapcmofihficwfim poc u .Loupmn so Ho>oa scooped o>Hm one pm open Eosm pcoaomgfio mauCMonchHw .Loppoo so Hm>ofl ocoosoo so» one we open Eopm ucmsomhfio zfipcmofiMficwfim seoam "*3. ".3 Aoao.v sea. segmwo.v mew. Amem.v mmm. **Amaa.v Hoe. Amem.v mmm. aooa>som .m Aeoo.v mam. *egoma.v ems. Aema.v mes. **Amoo.v mom. **Aama.v sew. noaoossoncoz .m gamo.v moo. eegsea.v see. Amso.v com. : *egaom.v msm.a *sgmms.v mam.au moaossso .H Ammo.v med. Amam.v mmm.: ssgeam.v Hem.an eegoom.v omo.m sensmm.v mom.sn Ho>ese cmaosos .ma Asmo.v 3mm. *Amaa.v cos. Aoom.v Ham. *egamm.v Hmo.a segmam.v mme.mu .wso eacosducoz a .maaom .aa coaeoosom .oa Azmo.v 2mm. *Aoom.v mas. Awmm.v mom. *shmam.v mmw. *Aomz.v mmm. : cofiomosoom .m Amao.v mam. Amsa.v see. Aaom.v ohm. segmwa.v mmm.a segmem.v mes.mn coaemesoonssse .m mmocfimsm Hmcomcom .5 Ammo.v mmm. Amom.v owm. Aaom.v mas. Aoom.v mms. Amem.v Hma. u osoo Hooaooz .e Aomo.v wmm. xxAmsH.v wmm. Aowm.v mom. I **Aomfi.v mam. Am::.V mmm. I mCOflpmsooo oeso: .m Asao.v mmm. *xAHQH.V mmm. *sAwom.v own. **Amma.v mzm. Ao::.v mmo. wcfimzom .: Ammo.v mma. ssgoam.v mom. segmoe.v mam.mn *egsem.v HoH.H segomm.v oem.mu osoo Hocomsod .m Aemo.v Hem. *Aeam.v sem. Amem.v mmm. I eegaom.v moo. Asao.v mam. u measeoao .m geoo.v as. Aoeo.v moo. Amom.v omo. **Aeoo.v ems. segmma.v oao.a ooooooe s ooom .H z .w< oofism mm \znmHpmHmm oEOocH psmumcoo mEmpH .mwmalmzmfi ”masoso mpHUOEEoo Cam: w .mo0fi>som .moaomssmlco: .moaemcso Lop moHSmom :ofimmosmomnl.s mqm" 6“ lrivate schcrls-profit .07 ’l.818) 89. Private schocls-ncn profit (wages only) LC? (2.389) 90. Religious L non-profit organizations .63: (.5C9) 91. Feligicus bodies .352 (.669) 92. Non-prrfit organizations .313 (.315) 93. Foreign Travel -1_871 ( _3gu)|l 9L. Transportation -1,368 ( .L;3)¢¢ 95. “ther expenditures -2.u7u ( .uab)|0 Personal consumption expenditures 96 Duratle commodities 203 (.632) 97. Non-durable Commodities :75 f .157) 98 Service: :g; ( ,Quz) ' I significantly different from zero at the 10 percent level or better. " I significantly different from zero at the 5 percent level or better. Blank - not significantly different from zero. TABLE lO.--Statistical Significance of Difference in Coefficients between Period 19fl7-1955 and 51 lost-1765. Relative Non-AgN Item Constant Income Prices /AgN 1. Food & totaccc 2. Food produced & consumed on farms 3. Food furnished gcv't & private employee: h. All other food & beverage consumption 5. Dairy products 6. Meat & meat products 7. Food & vegetable products . 8. Alcoholic beverages 9. Tobacco products 10. Clothing & accessories 11. Shoes & other footwear 12. Shoe cleaning & repair 13. Standard clothing issued to military personnel 1“. Other clothing 15. Laundering & dry cleaning 16. Accessories % luggage 17. Jewelry i watches 18. Personal care 19. Toilet articles & preparation: 20. Barber shops & beauty parlors ' ' 21. Housing 22. Owner-occupied dwelling: 23. Tenant-occupied dwelling: '. ' 2“. Household operations 25. Furniture ' 26. Refrigerators, washing i sewing machines 27. Misc. electrical appliance: 28. Lighting supplies 29. Cooking & heating equipment 30. China glassware & tableware utensils " 31. House furnishings & equipment 32. Cleaning 3 polishing preparations 33. Stationery & writing supplies " " 3“. Misc. household paper products '* " " 0. 35. Fuel (except gas) 36. Tools " 37. Misc. repair services 38. Household utilities 39. Electricity “0. Gas “1. Water '0 “2. Telephone " 5' 0. “3. Telegraph, cable & wireless 0| uu. Postage " ' us. Domestic services excluding practical nurses “6. Parcel post imports " " “7. Moving expenses 52 U8. Medical expenses “9. Drug preparations & sundries 50. Physicians 51. Dentists 52. Misc. health services 53. Private & non-profit hospitals 5“. Accident & health insurance, net payments " " ' ' 55. Funeral & burial services ‘ 56. Personal business 57. Misc. bank services 58. Expenses of handling life insurance 59. Legal services " * 60. Other 61. Transportation 62. User—operated transportation ' * ' 63. New Cars *' " " " 6“. Tires, tubes, batteries & accessories 65. Automobile repair, greasing, washing 66. Gasoline *' 67. Oil '* '* ' 68. Automobile ins. net payments 69. Purchased-local transportation " " 70. Publicos & taxis " " 71. All other 72. Recreation 73. Admission to amusements 7h. Informal recreation 75. Books & maps 76. Magazines & newspapers " " 77. Toys, sport goods, etc. 78. Radios, televisions, etc. 79. Pianos & other musical instrument: 80. Flower seeds & potted plants ' 81. Clubs & fraternal organizations 82. Photographic studios ** P” 83. Hotels & restaurants 8“. Miscellaneous 85. Net payments for lottery tickets 86. Education 87. Public schools 88. Private schools-profit 89. Private schools-non profit (wages only) 90. Religious & non-profit organizations 91. Religious bodies 92. Non-profit organizations * ' 93. Foreign Travel “' " " " 9k. Transportation '* " " ' 95. Other expenditures ' Personal consumption expenditures 96. Durable commodities 97. Non-durable commodities 98. Services ' = significantly different from zero at the 10 percent level or better. " . significantly different from zero at the 5 percent level or better. Blank = not significantly different from zero. 53 Determination (R2) and the Standard Error of the Estimate are also presented. When the analysis is undertaken using twelve aggre— gated commodity groUps for the whole time period, the income elasticities seem reasonable in all cases. The coefficients are significant at the 95 percent or better level in all cases except expenditures for medical care. The fact that the income coefficient for medical care expenses does not appear statistically significant is con- sidered to be a reasonable result. Expenditures for medi- cal care is probably a function of several other variables not included in this study, perhaps level of education, level of income, not in the present but in some past years where the quality of nutrition and other medical services were critical to the future healthy development of the population, a variable indicating the age structure of the population and probably others. Another probably signifi- cant reason for the lack of relationship between income and medical care could be the provision of free medical services by the government to a substantial part of the population. It would have been desirable to analyze each commodity individually so as to select for the regression analysis the set of explanatory variables and functional relation- ships most appropriate for each case. Limitations of time and other resources prevented this researcher from investi- gating alternative variables and functional relationships. 5” it was decided, as indicated earlier, to use one functional relationship (double—logarithmic) and one set of variables for g1; commodities. However, it can be demonstrated, as in the case of Medical Care expenditures, that our selection will not provide the best estimates in all cases. More detailed analysis will be necessary for some commodities if adequate projections are to be made. income Elasticities for Ninety-Eight Individual Commodities and Commoditngroups The regression results for the income variable were arranged in three columns according to their values, nega- tive, greater than zero but less than one, and greater than one (Table 8). Examining the column with the negative income elasti- cities, it can be noticed that the values of the coefficients are not statistically significant in all cases except "Photographic Studios" where it was significant at better than the 10 percent level. These results indicate that the variable income does not serve as a good explanatory varia- ble for these commodities or services. That is, there is little statistical relation between income and the expendi- tures for the items involved. The items with negative income elasticities were examined to see if it was reasonable for them to have a negative income elasticity. For some, like Shoe Cleaning and Repair, Domestic Services, Moving Expenses, Admission 55 to Amusements and Photographic Studios, it was considered that a negative income elasticity was a reasonable result. The reason for the negative income elasticity would involve a combination of increases in relative prices and technolo- gical developments that make substitution possible and economically feasible. The substitutes would include tele- vision for admission to amusements, private automobiles for moving expenses, appliances such as washing machines, vacuum cleaners, dish washers, etc., for domestic services, and easy to operate cameras for photographic studios. In the case of shoe cleaning and repair, the effect of signi— ficantly higher prices was probably the main determinant. A shoe shine cost five cents in 19“? and 25 cents presently (a 500 percent increase). The largest number of items was found with income elasticities positive but less than one. In this category, we found all Food items with the exception of Alcoholic Beverages, most Clothing and Accessories, Housing, most Household Operations, and most Medical Expenses. Other expenditure groups showed a mixed pattern as in Personal Business, Recreation, and Education. There were thirty-two items with income elasticities greater than one, predominantly in Household Operations, Transportation, Religious and Non-Profit Organizations and Foreign Travel. The income elasticity obtained for Durable Commodities appears reasonable. The same is true of Non-Durable Commo- dities. For "Services," however, it seems to be somewhat lower than one might expect but not unreasonable in View of the fact indicated in Table 7 where it can be seen that Expenditures for Services grew at a very slow rate from 19“? to 1965. Relative Price Elasticities Economic theory postulates that there is an inverse relationship between prices and the quantities purchased. That is, the "correct" sign for price coefficients should be negative. Seventy, out of a total of 95 items, had negative relative price elasticities. This high proportion of items, approximately 7“ percent, with negative relative price elasticities confirms what one would expect from the postulates of economic theory. Furthermore, out of the twenty—five cases with positive elasticities, only five showed that the coefficients were statistically significant at the 10 percent or better. The five items included Owner Occupied Dwellings, Tenant Occupied Dwellings, Cleaning and Polishing Preparations, Fuel (except gas) and Photographic Studios. The two items relating to Housing can be explained in terms of a rapidly upward shifting demand function plus the necessary income to back this demand up. The result is that purchases would continue to increase from year to year even with price increases. 57 The remaining three items could have positive price elasticities for the same reasons indicated above or perhaps others. They are rather insignificant and no special effort is considered necessary to explain them further. Coefficients for Non-Agricultural Employment to Agricultural Employment Ratio This variable was selected as an explanatory variable, as indicated earlier, to represent the structural changes which have occurred in the economy of Puerto Rico during the period studied. The coefficients for this variable are not summarized in this chapter but appear in Appendix A. The statistical results indicate that in most items, approximately 60 percentaof the total, the coefficient was not statistically significant from zero at the 10 percent level or better. In the cases where it was significantly different from zero, a larger proportion was at the level of 10 percent rather than the 5 percent level or better. These results seem to indicate that, on the whole, it was not a very useful variable in explaining changes in the patterns of consumption of many individual items. Never- theless, at the level of aggregated main commodity groups, it is significant for such groups as Clothing, Personal Care, Housing, Household Operations, Personal Business, Recreation and Religious and Non-Profit Organizations. It was also highly significant at the next higher level of 58 aggregation, that is, in the cases of Durables, Non— Durables, and Services. On the other hand, this variable showed to be of no statistical significance, at the 10 percent level or better, in several important commodity groups such as Food, Medical Care, Transportation, Education and Foreign Travel. It could, perhaps, be argued that the result for food is rea— sonable if one assumes that the pattern of food consumption does not change very rapidly due to habits, customs, etc. But it would be more difficult to explain, or rationalize, the results obtained for medical care, transportation, education and foreign travel. The results obtained for this variable are rather mixed and difficult to interpret in terms of certain patterns. Nevertheless, it seems to be of value as an explanatory variable when working with aggre— gated commodity groups. Coefficients of DeterminationLRe) As in budget studies, regressions where the R2 increases dramatically as observations are grouped due to the fact that variability within each income group is reduced and the average observations fall closer to the regression line, the same happens in the case of time series data where the data is more aggregated than in the case of "grouped" budget studies. In time series, the data is aggregated for society as a whole and then simply divided by the number of persons to obtain the per capita expenditures. This process 59 does not allow for variability in expenditures due to reasons of income, place of residence, education, etc. Therefore, a very high coefficient of determination (R2) in the time series regressions should not be considered as an index of the explanatory power of the variables in question to explain changes in the consumption of the item or commodity group under consideration. Difference in Consumption Patterns Between 1937—1955 and 1956-1965 One of the original hypotheses of this study was based on the assumption that due to a very rapid rate of economic growth in Puerto Rico, during the period studied, the patterns of consumption and therefore the regression coefficients would be significantly different between the earlier and the latter parts of the period. To test this hypothesis, the original data was divided in two groups, 19U7-55 and 1956565. Regressions were run for both periods by means of the dummy variable technique. The following equations were used: 1. To calculate coefficient for first period (19“7-1955) log C=a+bl(Dl log Y)+bz log PX/Po+b NonAgN/AgN 3 2. To calculated coefficient for the second period log C=a+bl(D2 log Y)+b2 log Px/Po+b NonAgN/AgN 3 where D1 = 0 when Y had a value greater than Per Capita Income in 1955. 60 1 when Y had a value less than or equal to Per Capita Income in 1955. D2 = 0 when Y had a value less than or equal to Per Capita Income in 1955. = 1 when Y had a value greater than Per Capita Income in 1955. The use of the dummy variable technique to calculate the regression coefficients for both periods also allowed to determine simultaneously, by means of a t-test, if the coefficients for the two periods were significantly differ- ent. This test involved running the following regression: log C=a+b log Y+b2(Dl log Y)+b3 log Px/Po+buNonAgN/AgN 1 if the coefficient for Income would be significantly differ- ent from zero, using a t-test, then the coefficient for Income would be significantly different between the two periods. The same would be the case with the remaining coefficients.6 Appendices B and C show the regression results for the two periods. The statistical significance of the differences appear in Table 10. The results appear to indicate that the difference between the two periods was not statistically significant for the following major commodity groups: Food, Clothing, Personal Care, Housing, Medical Expenses (except Health Insurance), personal business, recreation (except magazines and newspapers, and photographic studios), and for religious flFor a discussion of dummy variables and their use see: J. Johnston, Econometric Methods (McGraw Hill Book Company, New York, 1963), pp. 221-228. 61 and other non—profit organizations. The difference between the two periods appeared significant in a few items such as telephones within the category of Household Operations, for transportation and for foreign travel where the difference was the most significant. In many instances, see Appendices B and C, the values of the coefficients differed considerably from period to period but the standard deviations were also large and the degrees of freedom fewer due to the introduction of the dummy variables. A combination of these factOrs resulted in determining that the actual differences were not significantly different. Further research, increasing the number of observations and reducing the number of explanatory variables, would probably show the coefficients for a larger proportion of items to be significantly different for the two periods. However, based on our findings, we must conclude that for most items the coefficients were not significantly different between the two periods. Summary The nature and extent of changes in the patterns of consumption in Puerto Rico, based on the time series data, aDpear in most cases reasonable. It was hypothesized that, because of the rapid rate of economic growth, the elasti- Cf’Lties between the period l9U7-l955 and 1956-1965 would be SiRnificantly different implying a rather abrupt change in 62 the patterns of consumption between the two periods. This failed to materialize indicating that changes in consumption patterns have been more gradual than was assumed and, therefore, more in accordance with the experience of other countries and the postulates of economic theory. CHAPTER IV BUDGET STUDIES ANALYSIS OF HOUSEHOLD CONSUMPTION EXPENDITURES This chapter presents the results of the statistical analysis performed on three budget studies conducted in Puerto Rico. As indicated earlier, these budget studies were conducted in 19u1, 1953 and 1963. Results from the 1952 budget study are available but will not be reported because the 1953 results were considered superior in terms of sample size, coverage, etc. The subjects to be discussed in this chapter are, in turn, data, methodology, presentation and discussion of results, for main commodity groups from the three-budget studies based on grouped data; presentation and discussion of results from 1963 budget study, based on statistical analysis of individual observations for all commodities whose expenditures were estimated on a yearly basis; signi- ficance of difference between coefficients for different sub—groups (urban, rural, San Juan, Rest of Island, Wage Earners, Non—Wage Earners, and families with annual income of less than $3,000, families with yearly income greater than $3.000) within the 1963 budget study; and, finally, 63 6“ comparison and discussion of results obtained from the 1963 budget study, based on grouped data and on individual observations. ‘Review of Data Sources This section provides information regarding the scope and other pertinent details necessary to evaluate the quality of the data used and therefore of the results pre— sented in this chapter. JQAI Budget Study This study was carried out Jointly by the Works Projects Administration (WPA) and the Puerto Rico Department of Labor with the purpose of providing weights to be used in a new Cost of Living Index for Wage Earners' Families. It covered H,999 wage earners' families. Interviews were conducted fiwmn.lanuary 19A1 through May 19u2. The sample and final results were divided into four Efiusgraphical regions: Puerto Rico, Urban, Rural Lowlands (CKDastal sugar cane producing plains) and Rural Highlands (ixéntral mountainous coffee growing areas). Original data frcnn this study were lost. There remains only the final Pepcortl and a few unpublished tables in the files of the Depzartment of Labor. The estimates of income and expenditures ‘ 1Alice C. Hanson and Manuel A. Perez, Incomes_and Iflflfiiflditures of Wage Earners in Puerto Rico (Puerto Rico: Department of Labor, 191:7). are presented on a weekly basis whereas in subsequent budget studies data was presented on an annual basis. 1952 judget Study A total of 1070 wage earning families were interviewed by the Department of Labor to update the weights assigned to the different items composing their cost of living index and also to include new items or drop some which had lost relevance in the budget of a working family. The original data was not kept and data used in this study were taken from the final published report2 plus unpublished tables in the files of the Department of Labor. This Budget Study contains five geographical break- downs: San Juan Metropolitan Area, Rest of Island, Urban, Rural. The incomes and expenditures are adjusted and pre- sented on an annual basis. A decision was made to exclude .this source of data for the 1953 Budget Study had greater coverage. 1253 Budget Study For the first time Non-Wage Earners were included in the sample. Two volumes of results were published: one War the wage earnersj and another containing the results * 2Puerto Rico, Departamento del Trabajo, Negociado de lurtadisticas del Trabajo, Ingresos y Gastos de las Familias OhINeras en Puerto Rico en el A30 1952 (Puerto Rico: Depart- amerux)del Trabajo, n.d.I: 3Puerto Rico, Departamento del Trabajo, Ingresos y GTStcm de las Familias en Puerto Rico: 1353, Vol. I (Puerto RICCD, Departmento del Trabajo, 1960). 66 of both groups together representing all families.“ A total of 25MB families were‘interviewed of which 556 were non-wage earners (approximately 22 percent of the sample). Original data was not kept and data used was taken from the two published volumns and unpublished tables. The unpublished tables were used exclusively in this study because the published tables expressed expenditures in terms of percentages rather than in absolute amounts. The geo- graphical areas covered were the same as in the 1952 budget study . 1963 Budget Study The original data for this study is kept on magnetic tapes which were made available to the author. Much more emphasis will be placed on this study than on the previous ones because of the greater statistical significance obtainable by working with the complete set of observations. The sample included wage and non—wage earners' fami- lies and the total number of families interview was 2,662. The geographical areas were the same as those used in the 1953 Budget Study. Following are the sub-groups for which results were obtained: “Puerto Rico, Departamento del Trabajo, Ingresos_y Gastos de las Familias en Puerto Rico: 1953, Vol. II (Puerto Rico, Departamento del Trabajo, 1§60). 67 Number of Families 1- Puerto Rice 260? 2. Urban 1677 3. Rural 985 H. San Juan 107M 5. Rest of Island 1588 6. Annual Family Income less than $3,000 1373 7. Annual Family Income greater than $3,000 1289 8. Wage Earners 1163 0. Non-Wage Earners 1U99 The information requested from the selected families was contained in three questionnaires or schedules. Schedule A requested general information regarding the family to determine whether it met the sample requirements and could be interviewed using schedules B and C and also whether it was willing to be interviewed. Schedule B recorded expenditures, savings, debts, and other information regarding the wealth of the family. It is a very detailed questionnaire consisting of U8 pages and approximately 700 items of consumption. It resembles rather closely the format and extent of detail of the U. S. Bureau of Labor Statistics "Survey of Consumer Expenditures," which had a Schedule B of 59 pages. Schedule C provides 18 pages for the detailed description of expenditures in food, alcoholic beverages, items of personal care, tobacco and household supplies for the week previous to the interview. It included approximately 350 items. In summary each family interviewed had to provide estimates of their expenditures and other household financial details included in the above-mentioned 68 schedules which totaled approximately seventy pages and covered expenditures for an estimated one thousand items. The method for estimating expenditures in the survey was to break up a commodity group into its various compo- nents. Clothing for example was broken into clothing for men and boys, for women and girls, and for babies. Within these categories, further separations were made so as to include 60 clothing and footwear items for women and girls, approximately 60 items of clothing for men and boys, 27 items for,babies, and 13 items related to sewing materials and the maintenance of footwear and clothing. All these items of clothing and footwear total approximately 160. The author considers this degree of detail might lend to serious errors in estimation. It is extremely difficult, for anyone, to remember the number and prices of suits made of wool, the number and prices of suits made out of rayon and dacron, the number and prices of suits made out of other synthetic fibers. In the area of shirts, you must remember the number and prices of dress shirts made out of cotton and dacron, those made of other materials, the number and prices of sport shirts made out of cotton, those made out of dacron and cotton, those made of other materials; and the number and prices of polo shirts. In the area of stockings, the number and prices of stockings bought dur- 1ngthe last year made out of cotton, those made of rayon, Hlose made out of nylon and those made out of other fibers. 69 These examples provide an idea of the type of information required in these questionnaires. The author considers these details unnecessary. Perhaps even damaging to the estimates of all other items in the study because, as a result of this extreme detail, the interviewer may decide that some or a large number of questions are or look silly; he then may decide that the whole thing is not worth a serious effort even for the more reasonable questions. It appears that much more pretesting and determination of realistic interview recall possibilities are needed to improve the quality of the data. The situation described above is considered serious for the interpretation and evaluation of data obtained from Consumer Surveys in the United States where, on the average, the population has better record keeping attitudes (if for no other reason than the income tax laws). It is much more serious, in the opinion of the author, in a country where the level of education is lower and the record keep- ing attitude is not as developed as in the United States. The U. 8. Bureau of Labor Statistics reports that in the case of the U. S. surveys, "inaccurate reporting is a source of error despite continued research in schedule (kfitign and intensive training of the interviewers. Such buiccuracies result from memory errors, misunderstanding of a question or reluctance to answer it, and incorrect mfilries by the interviewer. Although the Bureau of Labor 70 Statistics and the U. S. Department of Agriculture have accumulated substantial knowledge about such reporting errors and will continue research in this field, these errors cannot be quantified satisfactorily."5 The author considers that this problem is much more serious in Puerto Rico, for the reasons indicated pre- viously. The author regrests not to have been able to interview or accompany an interviewer while questioning a number of families to get much improved understanding of how the data was obtained so as to better judge its valixlity. In the 1963 Budget Study non-response or partial response was approximately 12 percent.6 This compares very 7 favorably with the rate of 22 percent reported as non- response for the 1960 Survey of Consumer Expenditures con— ducted in the United States. Houthakker reports a rate of MA percent8 for the British surveys conducted in 1937- 1939, but the method employed in Britain involved keeping records of expenditures for four different weeks throughout 5U.-S. Department of Labor, Bureau of Labor Statistics, Handbook of Methods for Surveys and Studies, Bulletin No. lh58 (U. S. Government Printing Office, Washington, D. C., 1967), p. 61. A 6Worksheets, Puerto Rico, Department of Labor. U. S. Department of Labor, Bureau of Labor Statistics, 0 . cit., p. 61. H. S. Houthakker, The Analysis of Family Budgets, 71 :1 yxear which, of course, increases the probability of' *(erapping out of the survey. Statistical Model The results presented in this chapter are based on tric: single equation least squares model as in the case of 1;fiee time series results. The data was also transformed irifzo logarithmics so that the coefficients would provide nmezisures of elasticity. The model included only two inde- Ifiearudent variables: total family expenditures and family :1 1 1:26). Total family expenditures was selected as the explana— TiCJIfiy variable because, as explained in Chapter II, it is 811::umed to provide a better estimate of consumption by télbting account of future expectations with regard to 11fi<3ome as indicated in Friedman's permanent income hypo- thesis. Family size was considered as an explicit variable irhstead of implicity by transforming per family data into Dear capita data, because of the economies of scale involved if) the purchases of many items as families increase. To tr‘ansform the data to a per capita basis would imply a Druoportional relationship between family size and expendi- tLLPes which has been demonstrated does not exist. This Gilestion is discussed in more detail in Chapter II. 72 The actual equation used is: log Ci = a¥b log Family Total Expenditures % b? 1 log Family size where log C1 = Family consumption of ith good. log a = constant term b1 and b2 = regression coefficients (elasticities) The regression results for 19u1 and 1953 were per- formed on grouped data, on the basis of income, and weighted by the number of families in each income group. In the case of the 1963 budget study, it was possible to make regr ssions on a grouped basis and using individual observations. The regressions using grouped data were performed to compare the 1963 results with those obtained from the 19Ul and 1953 budget studies. Regressions using individual observations were made and results compared with those derived from the grouped data. After all the regressions were calculated, two sets of t-tests were performed. One to determine if the coeffi- cients were significantly different for a given commodity at different points in time (l9Ul, 1953, 1963). The second set determined if the coefficient for a given commodity was significantly different at the same point in time for different sub—groups (urban—rural, San Juan- Rest of island, wage earners-non wage earners, and families with less than $3,000 annual income-~families with yearly income greater than $3,000). 73 It should be mentioned that, in making t-tests to determine if coefficients are significantly different between certain sub-groups, the comparison involves two groups with different income and family size means. When testing between urban and rural or wage earners and non— wage earners, the procedure used did not keep income and family size constant. Keeping these variables controlled would provide a better test of the effect of urban or rural residence or of any of the other compariSons. Otherwise, you get the effect of all these variables and you might be testing not only urban vs. rural but also families with a given average level of income (rural) with families having a substantially higher average income (urban); families with a given average family size (rural) compared with families having a smaller average family size (urban); families with a given average level of educa- tion (rural) compared with families having a higher average level of education (urban) and so on. The important point is that significant differences may appear when comparing patterns of consumption of urban and rural dwellers but these differences should not be strictly considered as resulting only from the effect of place of residence. A further methodological point concerns observations having a zero value. Since the data were transformed into logarithmsznuiit is impossible to obtain the logarithm of zero, it was decided to assign a small but positive number 7D to any observation with a zero value. The number used was 0.00U9. Relative Importance of Main Commodity Groups In order to get an idea of the nature and direction of changes in the patterns of consumption during the period studied, this section discusses the changes in the relative importance of twelve aggregated commodity groups. The most dramatic change, in terms of its absolute value, is the decline in the relative importance of food expenditures (Table 11). Wage earners' families spent rela- tively less for Medical Care in 1963 than in 19Ul. This is probably due to the provision in the interior of free medi- cal services of rather high quality. The same is also true for Recreation, Reading and Education. Alcoholic Beverages and Tobacco, Personal Care, and Gifts and Donations remained approximately constant throughout the period. The other items such as Housing, house furnishings, transportation and personal insurance recorded various degrees of incre- ments in their relative importance within the budgets of wage earners' families. In the case of Puerto Rico families, the changes from 1953 to 1963 may be summarized by indicating that Food, Clothing and Gifts declined; Alcoholic Beverages and Tobacco Products remain approximately constant; and the remaining items indicate increments of varying degree. 75 TABLE ll.--Relative Importance of Main Commodity Groups: 19u1, 1953 and 1963 Budget Studies. ' Wage Earners Puerto Rico* Item 1931 1953 1963 1953 1963 1. Food 60.2 51.7 39.8 42.5 32.8 2. Housing 8.8 8.8 10.3 12.3 13.9 3. Clothing 8.1 18.2 12.9 13.8 11.7 A. House Furnishings 2.2 5.7 8.2 6.2 7.1 5. Transportation “.1 3.9 8.6 6.8 11.1 Alcoholic Beverages & Tobacco N.D 3. 3.6 2.7 7. Medical Care 5.0 2 2.6 3.0 .5 8. Personal Care 3.0 2.u .8 9. Recreation, Reading & Education 5.9 2.2 3.8 3.3 5.2 10. Personal Ins. N.D. .9 3.0 1.6 3.5 11. Gifts & Donations 1.3 1.8 1.5 2.7 2.5 12. All Other 1.5 2.2 2.7 2.7 3.3 Average Family Income 350 1188 2280 1717 3273 Average Family Size 5.6 5.3 5.6 5.0 5.1 38“ 1186 2376 1725 32u1 Average Family Total Exp. Source: Report 5 -B, p. 1. Puerto Rico, Department of Labor, Special Economic Studies Division, Income and Expenditures of the Families, Puerto Rico, 1963, Report h-A, p. 1 and *The 1941 budget study included only Wage Earners and for this reason data for this year is not included in the columns for all families in Puerto Rico. 76 The results for both Wage Earners' Families and All Families appear about what you may expect on the eXperience of other countries. Average Income, family size and total expenditures provide the reader with a frame of reference in inter- preting the data presented in the upper section of the table. Changes in Expenditure Elasticities (1911 to 1963) This section presents a discussion of the results derived from the statistical analysis of the data provided by the three budget studies for the main commodity groups. One of the objectives of this study was to determine the nature and extent of changes in the coefficients at different points in time and for different sub-groups of the total population at a given point in time. The basic information to make these comparisons is provided in Appendix D, which contains the actual regression equations, including the coefficients of determination for eleven main commodity groups. A summary of eXpenditure elasticities for the three budget studies appears in Table 12. An inspection of Table 12 indicates that all items, except Clothing and Furniture and Household Operations in the l9ul .Budget Study, are statistically significant at better than the 5 percent level. Examination of the values of the coefficients indicate no major discrepancy from the 77 TABLE 12.--Expend1ture Elasticities of‘Main Commodity Groups from 19H1. 1953 and 1963 Budget Studies. Item Puerto Rico Urban Rural Wage Earners Non-Wage Earners 1. Food 3. 19h1 N.D. N.D. N.D. .7h9 (.082)'| N.D. b. 1053 .632 (.008)" .63“ (.021)" .580 (.0?:)" .673 (.019)" N.D. c. 1963 .6u9 (.OOB)" .h85 (.025)" .5eu (.027)" .587 (.ONO)" .638 (.010)" 2. Clothing 8. 1901 N.D. N.D. N.D. 1.025 (.60“) N.D. t. 1053 1.002 (.0.3)" 1.025 (.020)" .653 (.072)" 1.12“ 0.056).. N.D. c. 1063 838 (.036)" .60 (.oe'n" 833 (.cué)" .871 (.070)" .788 (.052)" 3. Housin: a 10k] N.D. N.D. 3.1. 1.556 (.290)" N.D. b. 1953 1.278 (. _1) 1 10“ (.0133" 5.1. 1.072 (.100)" N.D. c. 1064 1.178 (.0 A)“. 886 (.020)H 1 079 (.050)" 1.050 (.051)" 1.162 (.029)" u. Furniture & House Opr. . 3. 10a; N.D. N.D. : L. 302 (.716) t.t b. 105? 086 (.088)" 1.023 (.100)" 1 :41 (.Oéi)" 1.798 (.175)" 1.3. c. 135: H0 (.036)" .659 Lemma-:1 hour" 1.168 (.17u)" .853 (.029)" 5. Transportation 8. 11L: N.D. H.u. H 2 1.223 (.303)" H L t. 10‘? 1.58 (.C12)" 1 696 (.C(9}" 1 72: (.OFLF" 1.836 (.221)" N I c. 10v} 1.1’0’J (.07F)" 1 ”Hf 5.1%?)" 1.126 (.121"' 2.26“ (.002)" 1.5.? (.090)" 6. Alcohrl & Tobaccc a. 19L; N D. 3 D. J o 3.3. T - t. 1W‘j Pi€ ( O30)" ."70 (.Oi?*" .71? (.082)" 1.051 (.167"' ..‘ c. 101% .628 ’ 03))" 19 (.035)" .583 (.0“?)" .u80 (.138)" .~ (.241)" 7. Medical Care a. 1031 N.D. X.L. ” L. 1.37“ (.811)" .L L. “_3 1.253 (.OG7)" 1.34‘ (.085)" 1.“?0 (.1f5)" 1.769 (.166)" l c. 13*? 1.13“ (.051)" D27 (.CEH"' ‘ 080 (.th)" 1.h38 (.077)" .J '.371)" 8. fersrnal Care .1. 1741 N.D. 1:42. 3: .. .977 (.219)" .1 b. 1%). .781 (.0114)" .7741. (.01‘)" 730 (.091)" .680 (.053)" ' c. 13": 736 (.c19)" .593mom1" “.4 ($37)" .69" 106?)" 1C3)” 9. Recreation a. 11.1 N.D. r:.’. 3.11. 1.23% (.1114)" . b. 19€3 1.500 (.037)" 1.563 (.071)" ..3?- (.“ t"' 1.7L} {.Zh9)" 2.1. c. 1023 1.2143 (.022)" 9141.- (.owr" 1.: ' 6.5171" 1.371 (.131)" 1.199 (.050)" 10. Personal Insurance 8. 19U1 N.D. N.D. ”.1. l.o. 5.1 b. 1953 1.773 (.026)" 1.739 (.06“"' 1.&j' (.03W)" 1.367 I.513"' L .. c. 1063 1.208 (.173)" 1.092 (.065)" 1.333 (.0u8)" 1.122 (.115)" 1.722 1.119).. 11. Gifts 3. 19M N.D. N.D. 71.27. 2.323 (.559)" 74.1.. b. 1053 1.860 (.160)" 2.008 (.276)" 1.853 (.169)" 3.85“ (.671)" N-L- c. 1963 1.6“0 (.067)" 1.358 (.101)" 1.“?2 (.097)" 1.8u8 (.fl22)" 1.6u6 (.057)" " - significantly different N.D. - No Data from zero at percent level or better. 78 coefficients that one may expect from a priori analysis based on the postulates of economic theory. For all families in Puerto Rico comparisons can be made only between 1953 and 1963 because the 19Ul study included only wage earners' families. The results for all families for 1953 and 1963 indicate that the expenditure elasticities for Food, Housing, Transportation, Medical Care and Personal Care remained approximately constant. It is interesting to note that the expenditure elasticity for food remained constant from 1953 to 1963. A possible explanation could be that in the period between the two budget studies quite radical changes occurred in the market- ing of food in Puerto Rico. The added expenditures for additional marketing services could have offset the expected decline in the expenditure elasticity of food. Other commodity groups had slightly larger expenditure elasticities in 1953 than in 1963. These were Clothing, Furniture and House Operations, Alcohol and Tobacco, Recreation, Personal Insurance and Gifts. It should be pointed out, however, that the differences in absolute terms were not very large for any particular item. Perhaps Recreation, Personal Insurance and Gifts could have been expected to have larger expenditure elasticities in 1963 than in 1953. Coefficients for the sub—groups (urban, rural, wage earners and non-wage earners) indicate more variability 79 from one budget study to the next. Table 13 summarizes the results of t—tests performed to testif coefficients were significantly different for various combinations including the same groups at the same point in time. The headings of columns identify the comparison being made. The results indicate that the coefficients were significantly different in most cases. 1963 Budget Study Results The availability of the tapes containing all the information obtained in the budget study of 1963 made 9 and also the possible the analysis of more commodities study of sub—groups of population such as the families with less than $3,000 yearly income. The 1963 budget study is composed of two types of data. One set refers to estimates of yearly expenditures (schedule B discussed earlier) and the other set refers to estimates of food expenditure for the week previous to the interview. A summary of the results will be presented separately for both sets of data. For further details of results from estimates of yearly expenditures see Appendix E and for selected food items results, based on expenditure estimates for one week, see Appendix F. Then the coeffi- cients for a small number of items which were estimated on a yearly and on a weekly basis will be presented in order h . 9See Appendix H for definitions of all items analyzed liltfiiis Study. sumo oz u .o.z .osoN Eopm ucosomofio haucmoflMchfim uo: u xcmam .soppon so Ho>oa osmopod m on» no ones Eosm psoLmMMHU zapcwofimficwfim n ** .pmppon so Hm>ofi psoopoo 0H mnu no case 509% pomsmmmflo zapsmofimficwfim u * 80 Dz * am es ** ** wcoapmooo a sumac .Ha Q2 ** *3 Oz ** DZ 02 .mcH Hmcomhom .QH ** ** s: ** es s: coapmosomm .m ** * *x ** es ** sumo HmCOmpom .m xx ** xx . * osmo Hmowomz .s Q2 Q2 ** Dz ** Oz 92 ooomnoe a Honooa¢ .m xx ** ** ** ** COHpMpsoowcmmB .m ** ex ** ** ** ex xx occapmsmoo omsom .: es es oz xx ** ** mcfimsom .m xx ** *e ** moxpoao .m es ** ex ** ** ** ** ooom .H mzlcoz Hmssm Hmpzm Hmssm Ammvmmma Amzvmmmfi Amzvmmmfi EopH .m> m3 .m> cons: .w> conga .m> cons: .w> .w> .m> Ammvmmma Ammvmmma Ammvmmma m; Hzma Ammvmmma AMBVHzmH Amzvfizma .mososmlosm mSOHLm> pom Ampwmplpv mpcmfioflmmooo CH mososmumflm mo moCmOHchwflmnn.ma mqmoH .osos Eoug ucmsmueao mHBCMloficme no: u xcmam pcoosod m opp be open Eopu pcosommfio AHpCMonchfim u a. *cAmmo.v wmAmwo.v :aAmmo.v teamwo.v esfimmo.v omAwwo.V msAwmo.V ssfiwmo.v zanowo.v xxAmwo.v **AM:O.V saflmwo.v a*Am~o.v sanmo.V mefimwo.v aefimmo.v **Aawo.v msfisso.v ashmzo.v xonmo.v emfiamo.v oeflmmo.v *xAmmo.V .xxnomo.v moo.m 0mm.m H@:.H :ow.m zoo.m mmw.m m©:.m mam.m HHH.N 350.: mam.H swm.a moa.m wom.H wo:.H Ham.H mmH.m mmo.m omm.H ooo.H NmN.H mmo.a omo.H ssm.fi senmzo.v smz. *anmmo.v mos. *xAmmo.v own. mao.v mes. *thmo.v moa. m Ammo.v Hmo. xefizmo.v Hus. mossofiocmoxm Hmfipsm a Hmpmcsm .wm mumxoae stopped toe spasm umz .om .mCH chomsmm .mm Ho>mse swampom .zm. .wto ascetaucoz a monmHHom .mm mcofiomcom a muwfiu .mm mpcoEomss< ou cofimmfiso< .Hm cowuwopoom .om assesses wcfiommm .mm coapmosom .wm coflpmusoomcwse Hmooq oommcohsm .sm .mcmse oos< .om cofipmpLOchmsB .mm wmwcfimsm Hmcomsom .am .mcH spasm: w ucoofioo< .mm momfipcmo .mm mcwuosmsca .Hm mmspo .om osmo Hmofiooz .mH .>som oflomosoa .mH acoEofisom oEo: .NH chauHCLsm .oH .axm efioemwso: cacao .mH mofiofiflfip: oflocomso: .zfi mpfioomm omsox .MH .cao efiocmmzo: .ma mmsfifiamza oofiosooo ocmcoe .HH mwcfififiozo oofiosooo smczo wcfimzo: oLmo Hmcomsom smozooom socoo a mmocm weacsofio muosoosm Ooomooe .>mm ofiHOLOoH< ooom .me ooomlcoz .me proe *xflmwo.v mam.l O H HNM—ZTLDWNQG H coco Loososo H can» mama psm 0 Coco topmoso o coco mmoq EooH .Aocofipm>smwoo Hmzofi>fioch AUBpm pomosm mmma Iumosam> Leone Op mcfioL000< omwcmcs< oofim ounosm Loo nofiufiofipomam ossufiocoaxmul.zfi mqmoH ucoopon m on» an on." sou» unchouuuu nauseouuucuu- I o. .uoupon no Ho>oa ucoouoo ed on» on one» song ucehouuuv haucdouuucuuo I o -.A~oo.v sea. .oAfioH.o mew. .ogmso.v ems. ..gmoo.o osa. ..1m~o.o mam. -.AM~O.U mom. ..nmma.v moo. .onmmo.o ~02. ..Am=o.v ems. notsuaucoaxm Andean . Antenna .hm onflmofl.o ems.“ -Aflofl.o mas.m onamfla.o mww.~ ..Aoaa.o Ham.“ ..1Hmfi.v mMH.~ ..AHHM.V mms.H ..fiomfl.v ems. ..Ansfl.v «mm.~ ..hmmo.v mmo.~ ouoxuue spouses too assesses so: .wm seamso.v woo.~ ekoHH.v Hea.m ooAHmo.v wmm.m o-AQQ~.V anc.~ c-AOMH.V mmm.m o-Aumo.v mflw.m ooflflafl.v mmfi.fl eoflmofi.v mmm.m nonmwo.v oMM.N conduzacH ascoahem .mm cogmoo.v moo." .aamzfl.v sac. ..Aomo.o moo.fl ..ANMH.V som.~ ..isufi.o mms. ..Aemo.v mm0.fl ..Amqa.v 054.: ..Amofl.v mmm. cofioco.v Hes." Hu>ute cuaotom .am naAoso.v oom.m c.1mn~.o “:0.H -.A=mo.v com.“ ..1=HH.V mm». ..Aomu.u umm.m ..nmmo.o Hso.m ..Amma.v smm.m ..Awafl.v :mm.~ ..Ameo.. aow.m acoduauficauuo uuuotmncoz . ==o_Mu~oz .mm u..meo.v mam.~ .oflmsfl.v meo.m ..Aooo.v flsm.m ..AscH.V moo.m ..asma.o amm.m ..Aflmo.v smo.m ..fimfifl.v sma.m ..AHOH.V oom.m -.Amwo.v amo.m acoauacoa . mouse wm con~mo.v omn.~ .nfimsfi.v omo.m oufinoo.v msm.~ .cfium~.v sw:.~ .cfiwma.v meq.m .cAmmo.v mam.~ ..Asm~.v omo.m oofimma.v ooo.m ooAaoo.v wmo.m oucqunza< ca :o—nn«5e< .am .nAmoo.. =m=.~ ..Aon~.o som.m ..Asmo.v mam.~ ..1:oo.v mam.~ ..AHMH.V ass. ..Asoo.v s:m.m ..Ammo.o oms.fl ..Amsa.v mm».m ..Ammo.v mo:.~ couucettoz .om o-Ammo.. smo.~ ..Asm~.o mew.“ ..Aomo.v mmw.m ..Amoo.v 3mm.m ..1sfla.o mom.m ..1mho.v Hum.m ..AmHH.V omfi.~ ..Aasfi.o 5mm.m -.Amuo.v ouo.m Huston»: saunas: .om ooAHoH.v oaa.~ -.1Hnfi.o mHN.H -.Asoo.o mam.“ ..Amsu.v ss=.m ..1mmfl.o ms~.H ..mmoH.o omM.N ..Ammfl.o spa.m .cgsafi.v mam. ..Aeso.v asa.m couuauaem .aw uafimmo.v ooa. : oofima~.v emu. .Aaoo.v so”. u ..Aomfi.o mm». . Aoofl.o Hmo. ..Ammo.v mam. - ..Amma.v MMN.H- ..AoHH.V mmo. ..Aooo.v cam. n couuauboancaue «coca couscotss .sm .camofl.v mam.m ..A=sa.v oo».m c.1coH.v Ham.m ..Aomfl.o Oms.a ..1mn1.v =13.m ..Amsd.o sum.” ..Amflm.v osa.a ..Aoafl.o mam.~ ..Aamo.v ”No.3 coupautonucete caunoaos=< .wN uofiwno.v mam." ..Amo~.v mo~.m -.Aseo.v mza.fi ..A~eo.o woo.a ..1Hmo.v was." ..Auoo.v Hsfl.~ ..Amso.v 0mm.H ..AmHH.V firs.~ ..Amao.v MHo.a coaoaotoancate .mw o-Aooo.v o>m.~ n-Amoo.u moo. onfimso.v ~:O.H ocaara.v wmw.a ouAmeo.v Mme. qummo.v mm:.fi unflawa.v oma.m oofimso.v 0mm. o-nmco.v hmm.H unocunsm Hecouuem .cm eoAon.v ~Ho.m ner=H.V can.~ oofiaoo.v 03m.H ouaor~.V -m.w ounnHH.v and.a .u10HH.V mim.m cansam.v can.“ oonqo~.v mom. oofiono.v mofi.m oocdhnncm Lawson a acveuou< .mn toamocJ mem.fi ueaonaJ mom; nofifiaoQ ~.cm.~ ovomHJ 9:4 unmeafi; arm; u-AooH; sum; 3.23; 0.1.9; eoflmmHQ 3:4 00:50; 00min agony—7.5 .NN o-AooH.v ss~.~ ashom~.V mmo.m ..Amo~.v mmo.m ..Aaua.v mom.~ ..Aezfi.u HsH.N ..HJHH.V era.” .onsom.v mos. ..Ao>~.v wom.a ..Ammo.v woa.H u:¢«u_n»:m .am ooawoo.v Mao." e-Aom~.v ~o~.H achro~.v omw.H oofimm~.v osa. nomcqfi.v Hmw.~ somon.g moo.a o-Amw~.1 mom. ooAmFH.v mo~.~ eoAmno.v Ham.a nflahn .om uoaooo.v saw." u-Aa01.o omo.~ ..A~oo.v sem.m ..Aaoo.v meo.a ..AMMH.V saw.m ..Amm .v amo.m ..Ammo.v moa.~ ..Amwfi.v wmm.~ .oflueo.v mm~.~ peso “suave: .o~ uquoH.. mso.m .caow~.v ooa.~ -.Amofi.v Ham.~ ..Assa.v mam.m ..1m.fi.v one.fi ..Amoa.v som.m ..Awofi.o som.= ..Aom~.v 5mm. ..Asso.v mmo.m nooa>nom caunmaoa .cn coamoo.v sea.” ouAm~1.V use.“ ..Anoo.v mm=.H ..Ammo.v qma.d ..,smo.v was.” ..1nso.v sen.” ..1011.V mms. -.nmflfi.v “mo.H ..Ao:o.v omm.~ accennsvm one: .9” ooAmHH.V hem. ..~omm.v was.“ .saoms.v ems.“ ..Assa.v was. ..A~sfl.o mmm.H ..Ammfl.v mam. Ammm.v 0mm. ..Amom.v ~5~.H .Amao.v coo.” onzuuctsm .o~ n-fimmo.v o-.~ ..aooo.. smo.fi ..Ammo.v moH.fl ..Ammg.v 0mm.a ..1m:o.v mwo.a ..flom .v mmN.H ..Aaao.v amz.fi ..Amao.v woo. -.AHNO.V awn.“ unusuaeconxm vflononao: tango .mn .-.m.o.v ago. c.1omo.v mmo. ..n~:o.o 0:5. ..Ameo.v Ohm. ..Aeno.v out. ..Amqo.v mas. ..Amso.v mos. ..Aemo.o 3am. ..Ammo.v mes. acquufiups ufiozonaoz .qa ucAo-.V fine." .oasms.v was. ..AMH~.V «mm. ..1msfl.v cos.” ..1mmfi.o mam. ..Asfla.o JMfi.H ..Asam.o wso.fl ..Aasa.v was. ..Ammo.v omo.~ assuage page: .m~ -.¢.o.v mmo.H ..Asmo.o 5mm.” ..Am=o.v mmm.fi ..1w~o.o ego.~ ..1ewo.v How.H ..Ammo.v wmo.a ..Aflmo.v owe. c.1mmo.v mso.fl n.1mmo.v who.“ needy-gone euozonso: .wa no.omfi.v can. ..~m~n. msm.a ..AcHH.V mam. ..Amma.v mos. . ..Ammfl.v mas. ..Am«fi.v ems. . Asm~.v new. I ..Amofl.. 3mm.” ..Amoo.v one. nucaaaozo unannooo-u:ucoe .aa emannnono vauasouolnocxo .o~ wcdnao: .m .uaoao.v mas. ooammo.v Ads. .ofimmo.v mes. ..Ammo.v ems. .ofiamo.o one. oofimmo.v sap. o-Aamo.v mam. ..Amao.v mm». ..Amflo.v use. ouao «accuses .a .omo.v mmo. ..Amso.u sow. .ohmmo.v cam. ..AMNH.V mam. u ..1aoo.v amm. Am~o.v mnfl. . ..Amafl.o Nam. u .cflsmo.v mos.“ -.Ammo.v mos. unotuoom Lozoo . noose .p Aamo.v new. ..Asao.v mmfi.fl ..Ammo.v wmo.fl ..Asqo.v mos. ..nsso.v mmm. ..Aamo.o mmm. ..Am:o.v Hmm. ..Amwo.v mmfi.u ..Am~o.v was. ocucuoao .u onaa-.v mom. oofimmn.v mam. oofiamfi.v mmm. AooH.V mmo. n .Amua.v emu. AmmH.v mmo. AmHm.V omo. commom.v mm». Ammo.v Hmo. naustLm oouaaoa .m ..A=o~.v mmo.~ ..Aaon.v cam.” ..Aofla.v maN.H ..Ammfl.o mam.~ ..1mwa.a moo.~ ..Asafl.o HUH.~ ..Amo~.v ems.” ..A~wd.o smm.fl ..Awmo.v sam.~ coouuosom cadenced< .. .050mo.v mmh. o-Asmo.v mam. ooflwmo.v mms. nonmmo.v mm». o-Aoao.v mom. o-Ammo.v Nos. ooammo.v mom. cofiowo.v ooo.a ooaamo.v ass. voom .m .aum voomncoz .m nousunvcoaum flavor .~ nuucham opossum haauv ocean” Aflanv a «so a :4 u a a a . . Io ouaz-coz «no: to uaox anus can Add a a x AAA V a a Awmsw wmwm. “Mme” wmwM. Ada V m s an .suaun pouvan mood aouu co>auon .oauuouuucan onsuaocoaunn-.m~ n4nmH ucoosod m on» as open Eose ocosmwufio haucmOHuficme n so .Louuon Lo Ho>oH pcoopoq OH on» um chow Cosh pcmpmmufio zfiucmofiuficwfim u s *xflmmH.V mom.H *aANOH.V one. stfimmo.V mom. *sfimHH.V 0mm. *xAzmo.V mflm. aefime.V com. 1.1000; Om”... Aqu.V was. *xAmwo.V was. Azmo.V asflmmH.V aaflm4H.V seflmoH.V xaAmoa.V axfimHH.V aefimmo.V esnmmfl.V Amso.V *xnmoa.V *mflmmH.V *xflom . Assn. Amcfl. xxmmoa. **Afimfi. V V V V V exflmmo.V mmH. I mossufiocmoxm Hmfipsm a assocsm .sm smm. I m exude stopped coo .mussa smz .om .mcH chompmm .mm mom. I Ho>mse cwflmsom .zm mm». I .msc osmosmncoz a msofiwfiaom .mm mam.HI mcofioacoo a moose .mm :mm. I mpcmEmmsE¢ Op cofimoHEo< .Hm coflpmosowm .om moo.HI Hostess: mcflummm .mm cofiumozom .wm .mcmse Hmooq oommcocsd .sm smH.HI .mcmte .0u:< .wm woo. I cofipmpLOansce .mm mew. I owmcfimsm assessed .zm mos. I .mcH spasm: w ucotfioo< .mm msm. I mpmwpcoa .mm mam. I mcefiofiomcm .Hm Hso. I waste .om one. I sumo Hmofiooz .ma ozm.mI .>Lom owoooEoc .mH ucoedfison mEom .NH ossuficssm .©H ssm. I .axs oflocmmso: Lasso .mH woflpfiafio: ofiozooso: .zfi mswmdox peso: .ma .LQO Ufiocoaso: .mH *enfimH.V smm.HI onsflflfiosz oowasooo ucmcoe .HH ouswaaosa oofidsooOILoczo .oH mCano: .m *xflomo.V Ham. wasp assessed .m xxflmoo.V oxH.H Loosqoom Lozuo a moocm .N *xfimqo.V pom. scazaoflo .m xxAmmH.V mos. oposoosm ooomooe .m **Am:H.V mwo. I mommpo>mm ofiH020oH< .3 **AH30.V mmz. ooom .m .axm soosnccz .m .axm Hoses .H H coco Loumoso H coco wmoa use 0 cmcp mood EmuH o coco caboose .AuCOwuw>meLo Hmzofi>fitcHV meson uemwsm mwcHIIoosHm> Lawns 0p wcfiosooo< newcwss< oofim oucosd Loo moflufioweosai omfim mafiEmaII.wm mumoa acouuon m ecu an ego" 60L» uconuuuav AHuCGQuuacmun I .- .Luuuon Lo Ho>oa acoouun OH on» an OLON song ucvnouunu AHuCaUqucwnn - . .AMM~.V as“. I Amos.V smo. .Ammo.V sad. I Asma.V emu. I AMHH.V ~30. I H. I MsVH.V mmo. I .Aalo.V 5.”. I Asmo.V mm“. I saunasecuaxi Vaausm . Vanessa .am ooflzo~.V «mm. I ..Aoon.V mom.~I ..AasH.V 23m. I ..Asmm.V 3mm. I ..AV”~.V mam. I m. I ..Aa,m.V mam. ..AHHV.V oo~.HI ..Aama.V smm. I nauseae spouse; soc assessaa so: .om ..mmH.V com. .Hm~.V ow“. “V3H.V mam. ..mooH.V =05. .AmsH.V omM. m. .AWMH.V mmm. ..mowfl.V cum. .oflsoV.V 0:0. eecmtach Vacantea .wm ..Asw_.V one. I Aom~.V spa. AOQH.V “OH. I ..AVHN.V ago. I AaeV.V was. V. I .lamm.V man. I ..AsoV.V him. ..lm_fi.V mOM. I ~e>age cheLcm .am ..Amm_.V «as. I ..Aoe~.V won. I ..A-H.V mos. I ..fi~Tfl.V mas. I ..A5V~.V was. I n. I .ludfl V mam. I ..nsqV.V emu. I ..A$CV.V so». I .uuo sacoLaIcoz . achmVHma .mm ..Amm~.V s”...I ..Amm~.V mo~.~I ..AowH.V som.HI ..AJVH.V mom.HI ..AmsV.V ”Vs.HI .V.HI .Ao H V mam.~I ..Amm .V m.m.HI .onaoH.V MJV.HI ==oVsacoa . ”acne .mm Aao~.V moi. ..AMcV.V mm~.~I ..Am=H.V mom. I .nsem.V mam. I ..AahV.V sea. I o. is.a V msn. I ..Ame .V new. I ..AmVH.V ass. I n»:~Eeaze< 0; ccann_su< .flm Ioaomm; :3. A93; 1;. I o-AnMVQ 03.. no.2; of. :3; 4...... ,4. A H V mwo. I nomowflV :m. .1600; men. :ojfifiobm .OM -..amfi.V was. I ..A:o~.V .mm. I ..RM“V.V mmo.HI .. amV.V ass. I ..A:QH.V sao.HI ._ V Vac."I ..,H4V.V Hem. I ..Apce.V mco.~I Haauusez echcum .om o.noom.V Hmo.a ..Ach.V ”mo. ..AoaH.V mam. ..Rmma.V mcfl.m ..lusV.V ”as. V Nmm.m ..A4m .V auo.fl ..A¢aV.V on." coVseoaem .om u-Aow~.V mso.~ ..A=m~.V as». ..nm:V.V are. ..flson.V Ema. ..AT:H.V has. V mam.fi .ABHH.V aaa. ..A5~H.V are. :ofiuappoamzage Have; eaaaceeza .sm o.noom.V was. I ..Aam~.V ch. I ..A:QV.V msfi.VI ..AVem.V was. I AM»H.V NVJ.HI V has. I ..»NHN.V ssh. I ..AcmH.V sm_.~I .ecage ea_scscu:< .om A=o~.V moo. I “no".V ooV. I AsoH.V Voo. I ..Axoo.V eem. I Amaa.V med. V use. I AOAV.V sac. Amso.V sso. I ccVueuLcancage .mm Aos~.V mm". I loco.V "NH. I ..ANVV.V 00:. I Asom.V moo. I ..AMVA.V mam. I V mma. Ace .V oVV. I ..AooH.V man. I anocunzx Vacoegea .am anamom.v me». I noao=~.V can. I ooflm:~.V nos. I ovonm.V Ann. I nunmr~.V tum. I V ado. oohmoV.V oflz. I ooflmn .V we». I oocaL3=c_ cafinez a accvuuu< .mm ooaom~.V mm.. I ..Asm~.V omm. I AH:H.V sea. I ..Anmm.V ”as. I .iVe..V 7x“. I V moo. I ..AmmV.V w:;. I ..Aoafi.V mam. I nsnascoa .mm A-~.V .oH. I ..A$o~.V one. I ..AseV.V gun. I Ammn.V “em. I ..l:oa.V ale. I V fiVa. ..AsoV.V ewe. I RV=V.V min. I acaau«nsza .VN Aamfi.V mac. I nem~.V 0:0. I ..Amcfi.V «mm. I Asqm.V mam. Azo~.V was. I V mos. ..AOPV.V saw. I Amm~.V Hso. I waste .on unamm~.V azm. I ..~ns~.V was. I -.AV=H.V ram. I ..AnwV.V sow. ..Acm~.V gas. I V mVV. I ..AmeH.V Hos. I ..hmoH.V 03;. I once Aac.co: .oa oofiom~.V Hmo.~I .ufimo~.V oem.~I .afiooH.V MH2.~I .onnmm.V “mm.mI .uhomV.V osm.mI V moo.HI ..A:m~.V mas.HI noAVm~.V oam.~I soun>nom owuayaca .QV 0.3.2; can. nonw:.v «:9. 0.7.2.0; ems. napanQ 3:4 31.34; 00!». V Few. onflaaaV wms. nugmo.v nHm. uczaaacm ~60: .nm -.A.m~.V “no." R~m~.V ass. ..AHo~.V sea. ..flmmm.V VmV.~ ..lma~.V m4“. V mmm.fi AosV.V mom. ..Ame.V com. agapactam .o~ oofioao.V mom. I .Anco.V see. I ..Am:o.V mm. I ..AVQO.V mma. I .ofiowo.V new. I V Hm“. I .olmao.V mmfi. I -.nmmo.V saw. I auuzuaocoaxm uaozenso: cacao .ta ...omo.V no». ..A=mo.V o:o.~ ..Aooo.V Has. ..AOHH.V m m." ..Ammo.V owm. V mms. ..Aomo.V mm . .oloso.V 0mm. uaauVHVu: eaocenao: ..~ Ao-.V moo. ..nooH.V Hma. Ams~.V ago. I ..fiam~.V gas. HMHN.V ”to. V mos. AHVH.V moo. A3=H.V oafl. niazaem 0:30: .MA o-amoo.V om”. .cfimmo.V 9mm. ..Amso.V was. ..Aamfi.V mma.~ ..naoo.V mas. V mms. ..Ammo.V New. ..Amoo.V was. acouuauoao vacconao: .wa ocaam~.V cos. I o-Amnm. mso.~I coAmm~.V Noo.HI ..Amom.V smo.flI .olmaH.V sow. I V mm=.HI ..Am¢~.V mom.aI .ofiflwa.V -~.HI nchVVexa ovaaauuo sauces .VV flwCaa—UID UOwQDUUO LOGIC .OH wcanao: .m ouaomo.v dam. oonhmo.V mmm. noammo.V Ham. oofioao.v oaw. o-Amao.V dam. oofiflmo.v mom. -.A0mo.v ssa. cohaao.V wmm. nonwmo.v Ham. «Lao «aconnom .w ..Aemfi.V on .H ..nmmo.V cam. ..hamo.V mmo. ..,smH.V sow.” ..lmao.V as». ..Am4H.V mw~.~ ..AamH.V mom. ..Mmmo.V mam. ..Ammo.V mos.” “apnoea; Locuo . spasm .s :33; wow. .1304 2:. :33; m3. .120; :3. .16on mmm. =83; mmm. .130; :9... :2..ro 03. olmzoV 3m. 93508 .o 00:. ..Anm~.V mum. ..AAON.V 3mm. ..Aosm.V mam. ..Ama~.V mam. ..anN.V amo. ..Amam.V a-.H .A«o~.V mam. ..AmmH.V mos. nauseous oeuaooa .m .ofihom.V men. I mmm. I ..hamH.V now. I ..Ahnm.V who. I ..fla~m.V won. I -.Amma.V mom. I ..Aoum.V mmm. I ..AmsH.V 3H3. I ..nm:H.V woo. I nouauo>om cuaocoo~< .a ..Aomo.V um. cam. ..Awmo.V mmz. ..A~wo.V NOW. I ..A=mo.V mma. ..Ammo.V mom. ..Amqo.V mmN. ..Amwo.V mom. ..Afl=o.V was. econ .m .anm voomIcoz .n aohsuavcoaxm HauOB .H atoctau apocgqm Asaduoo ooauuoaunuflm ouam naaaamII.s~ mqmoa ocmoLoo m map no open Eosu ucmummmao mapcmOHMchfim u c: .popuon so Hm>oa unmosoo OH 0:» pm osou Sosa ucmpowmfio aHu:MoHuacme u : Aomo.V mmfi. I *.Aomm.HV moo.m oafimoo.V ems. ..Amom.V omm.a mmszpaocmaxm Heapsm o Homecos .sm ..AQMH.V smw. I Amom.V gem. ooflmmo.V ooo.m *.naoo.V mHm.H momxofie stopped goo mussm pmz .om seasoa.V one. ..Amoo.V som. ..Amoo.V omm.m *anmsH.V wow.H .mcH Hmcompom .mm osflmHH.V mom. I ooflmmo.HV oom.o_ .oflooo.V Hoo.fi .ehomm.V mam.a Ho>ose swampom .om oenmoH.V mos. I ..Ammm.V omH.HI ooflmoo.V zoo.m ..Ammo.V wmm.H .wso pfioosmIcoz a msofiwfiamm .mm eonsoa.V mom.HI .eflmom.V moo.HI ..Amoo.V zoo.m sensoo.V ooo.H mcoapmcoa a moofio .mm gonoHH.V oom. I Aooo.V mas. .ogooo.V omo.m *.Aoso.V sso.a mpcmEomzea 0» cosmmHan .Hm sonooo.V mom. .egsmH.V 5mm. *.Aomo.V moo.m .agmmo.V mom.H cofipmosomm .om ssflmoo.V moo.HI Asw:.V ass. I ..Aomo.V oflo.m ganmmo.V msm.a Hostess: wcfiommm .om ssfiomH.V mom.a Amos.V Hoo.HI sofioso.V HHH.N ..AmmH.V oom.m cofipoosom .mm .efisHH.V omo. *.AmmH.V zom.H *ogooo.V oom. I .*Aomo.V mmm. cofiomosoamcose Hmooq ommocopsm .sm esflomH.V smH.HI AoHH.mV VmH.H eagmmo.V osc.: ..Awmm.V oom.m .mcmpe opz< .om Amso.V sso. I Amoo.V zan.oI ..Amoo.V mao.a **Amso.V ooo.a cofipoptoamcmse .mm osfimoH.V mom. I .ofloms.V omo.mI oxAmoo.V smm.a .ehmmH.V Hmo.m moccfimsm Hmcomuma .om *eflmmH.V mos. I Aoss.HV mmm.H ..Aoso.V moa.m **Aoom.V soo.a .mcH pcooaoo< o npawmm .mm **AomH.V mom. I osfimoo.V ooH.HI ..Aaso.V oom.a *.Aomo.V mmH.H momascoo .mm AHoH.V mam. I Aoms.V ems. I oxgmoo.V ooz.~ .eflmmo.V moo.H mcofiosmsca .Hm AmmH.V Hso. I AHoN.V oom. xeAmso.V H:N.H .xflsso.V ems. mmsso .om ..Amoa.V ooo. I Asom.V oom. I .emfioo.V mmfi.m ..nfimo.V omH.H memo Hmoflooz .oa osAHmH.V oom.mI *.Aooo.V moo.mI ..Asso.V mmo.m **Amwo.V owe.fi moofi>tmm osommsoo .mH wsAzmo.V mam. sxfimmm.v wmm. eonmzo.V mmm.H *thmo.v mom. ucoeafisom oEo: .NH teammH.V oom. *cfloom.v ymo. *aAmmo.V ooo.H *3A2:O.V 3mm. mLJpHCLsm .wa *xflmmo.V New. I ..AooH.V mmm.fiI ..Aamo.V smm.H ..Aomo.V Hom.H .axm oaocmmso: swcooI.mH *.Aooo.V omm. Ammo.V zoo. I *.Ammo.V mos. ..Amfio.V sow. mmfipsfifiu: ofiocmmzo: .ofi Aoofl.V ooa. .Aoom.V mmo. I ..Amoo.V omo.a ..Ammo.V mofl.H wsfioomm mono: .MH **Amoo.V mos. .sAmHm.V ass. *.Ammo.V oso.H ..homo.V mew. .sqo ofiocmmoom .mfi **AHoH.V smm.HI Ammm.V Hao. aeflmoo.V ems. .*Aooo.V moo. mwcfiafimzo omfiasooo ocmcme .HH wwcfiaaozo ooHQSUQO Loc3o .oa wcfimso: .m ..Aomo.V How. AoHH.V Hmfl. ..Amflo.V oss. .*Aoao.V oms. memo assesses .m **Amoo.V oofi.H exgmom.V Hom.H ..Ammo.V moH. x.flfiso.V smm. tmmzoooe emcee one moocm .s .egmoo.V som. axflofim.V mmo. *ehmmo.V woo. **Aomo.V mmw. mascooflo .o ..AomH.V mos. ..Aoom.V msH.H Ammo.V Hoo. ..hfioo.V som. moosoosa ooooooe .m *xAmzfi.V mom. I howm.V mom. *afimmo.V sqm.fi *xAHwo.V Hmm. mcwmsm>om ofiaocoUH< .o **AH:O.V mmz. **Aomo.v 5mm. axflzmo.v an». xxnmoo.v mam. 000% .m .oxm ooomIcoz .m .me Hmpoe .H mcofipm>sowno sumo noososo ncofipo>somoo some ooososo EopH Hmsofi>aocH Hosofi>fiocH A.m.aV spfiOfiommHm swam sfiseoa A.m.aV spaoapmoam menusocmoxm pr08 now mcoauo>somno HospH>HoCH use some ooqsoso co oommm .Asoocm osmosm momHV swam masses ocm motopfiocqum mucofiofimmooo scammmpwmm go comHLwQEOOII.mH mqmom oaaocooas 1 mNHm zfifiemm .ummfim .ocooxm usmpmcoo (\J 134 Amos. vamo. seHomH. mes.H Home. ammo. Hmsm. vmmH. nsocsom owe; eoz .HH HeH veeH. ssHmmo. comm. *eHmeo. Seem. esHomm. vmsm.Hu newness owmz .oH Heme. vmsm seHHmo. ammo. seHmmo. VOHm. seHmeH. Vemm.Hu AHHaV oeoHnH so pnom .m Hmmm. Hvoeo. «*HemH. VsON.H ssHmmH. VmHm. u ssHmse. VmHm.m HHHav case can .m Home. VHmm. ssHmme. some. eeAseo. Vsmm. neAmmH. vsHm.Hu HHHav Hesse .e Amos.vaso. ssHmsH. Smem.H Ammo. VoeH. : ssHomm. Seem. HHHav goose .e Hoom.vamo. sshoms. VHom.H seHemm. some. . ssHHHe.HmeH.e HHHavoome “was soooone .m . . . . . . . . HHHav Co ease nnoH Hmse vaso esHsmH seem Hmmm vam Ammm Seem oso ooom ease sooooso .s Ammz.avzmo. mmfiwma. vmmm. msAmza. mem. I **Anmm. Vamm.: AHHm. vmpm.mu mpmcpam mwmz .OH How». mem. **Hm~o. szz. *mezo. mem.H *xAmmH. mem.mu HHHm. vumm.mu HHHHV cash cwm .m Hemp. szm. **Hmmo. vmmz. **Hmmo. vam.H **Hmom. vam.ma HHHmm vwmo mu pan ooom case pmpmmpo .2 ANQN. meo. **HHQH. vmmp. **HHwo. vomm. **HmHm. Vmwm. I HHHmmm OHHmmEOQ HHHH.HVmom. **HmmH. mem. **Hmmo. Vm5H.H **A5Hm. Vm5m.mI mpmcpmm mwmz coz .HH Hnmo.HvHHm. **HmHH. VHHm. **HmHH. VHmm.H **Homm. mem.:I mpmcpmm mmmz .OH Ammo. Vaom. **Ammo. Vmwz. **Hmmo. vmm:.H **HHom. vzmm.mI AHH000 cwH0000 H000.H0053. 00000H. 0005. I 003050. 0000.0 0*HH30. 0050.5I 0000000 0003 002 .HH A003.HVOHH. 0*A00H. 000m. I 0*AmmH. VH3m.H 0*A5om. 0005.0I 0002000 0003 .0H 0000.H005m. 0*HH0H. 0005. I 000300. 0000.0 000050. 0000.0I HHH00 000H0H 00 0000 .0 H003.H000m. 0*AN0H. VmHm. I 0*H3HH. 0005.0 000003. V0HH.0I HHH00 0000 000 .0 H500.H03Hm. 03350H. 00H5. I 0*H0NH. 0H00.0 003050. V0H0.0I HHH00 H0000 .5 AH00.H0000. 0*HH3H. 0H00. I 003000. 0H50.0 003000. 0350.0I 0HH00 00000 .0 0050.H0000. HH00. 0030. I 0000H0. 0000.H *0A0H0. 0300.0I HHH000005 0000 0000000 .0 H000.Hv000. 000000. 0030.HI 0*H0Hm. 0300.0 00005H.H0Hmm.0I 0%MHMm0m0mm0m0MW0mwmw .3 H000.H0000. 00300H. 0000. I 00000H. 0500.0 0*H5H0. VH0H.0I HHH000000 0000 0000000 .0 A303.H000H. 0*H33H. 0000. I 00003H. 0300.H 003003. 0030.0I AHH00 0000 0000 0000 .0 A050.H0H03. 00300H. 0005. I 000000. 0300.0 0*HNH0. 0003.0I HHH0V 0000 000000 .H chHpmNHcmwmo 000000Icoz 0:0 monwHHmm (\J n: mNHm 0HH000 .p m (U H [I] o Foam Qn1m 00000000 147 H000.H000H. 003300. 0000. I 00300H. V500.H 003300. 0000.0I 0000000 0003 003 .HH A005.H000H. 003000. 0000.HI *0AH0H. 0035.0 0*HHH0. 0050.0I 0000000 0003 .0H H355.H000H. 00035H. 0300. I 00HNHH. 0000.0 003000. 0000.5I HHHOV 000H0H 00 0000 .0 H050.H0500. 0*H000. 0300. I 00303H. 0HH0.H 003000. 0000.3I HHH00 0000 000 .0 H005.H005H. 0*HHH0. 0000. I 00HH0H. VO0H.m 00HH53. 0000.5I AHH00 H0000 .5 H000.H000H. *0HH0H. 0000. I 0*AHHH. 0005.H 003500. 0005.0I HHH0V 00000 .0 HH00.H0000. H053. 0000. I H000. 000H. I A000.Hv000.0 HHH000W05 MM00 0000000 .0 . . . . I . . . . I AHHO 00 0000 0000 HH50 H03H0 003000 0000 003003 0300 H .0000 H0000 0 000 0000 0000 0000000 .3 H0H0.H0000. 000030. 0000. 00000H. 0500. 003005. 0000.0I HHH000000 0000 0000000 .0 H305.H00HH. 0*HH5H. 000H.HI 00035H. 0000.0 0*H0H0. 0003.5I HHH00 000 0000 0000 .0 H000.H030H. 00000H. 0500. I 000000. 0000.0 0*HH00. 0050.0I HHH00 00H0 000000 .H 00000HB 0000000 000 00008300 002 H033.H0000. 0000H. 0000. 000050. 0000.0 003050. 0005.0I 0000000 0003 000 .HH H000.H0030. HHmH. 000H. 0*H0HH. 0HOH.m 0*HH50. 0000.0I 0000000 0003 .0H HH33. 0050. HH3H. 0000. 0*HH00. 0000.0 003300. 0050.5I HHH00 000H0H 00 0000 .0 H000.HO0H0. 00000H. V305. 00000H. 0350.0 003550. 0300.0I AHqu 0000 000 .0 0003.H0H00. *Am0H. 0000. 00000H. 0000.0 000003. 0000.5I HHH00 H0000 .5 H030.H0500. 00350H. 0030. 003000. 00Hm.0 000050. 0300.0I HHH00 00000 .0 A000.H05H0. 003000. 00H0. 0*H0H0. 0H00. A500. 0030. HHH000005 0000 0000000 .0 . . . . . . . . . HHHOV 0005 0000 0000 H000 H0030 HO0H 0HOH 0*AH00 0000 H 030500 0000 0I 000 0000 0000 0000000 .3 A000.H0000. 0300H. 0000. 0*HHHH. 000H.H 000003. 0503.0I HHH000000 0000 0000000 .0 H000.H0000. 00000H. V000. 00000H. 0000.0 000003. 0000.0I HHH00 0000 0000 0000 .0 mem.va5m. *3AO0H. 000 . *mflmoo. 00mm.m 0*AmHm. vmmm.wl AHH¢V oon 000030 .H 0000030CH H0c00000 m 00Hm unfimm .pmmHm .000mxm pzmumCOO (\J 0500.00000. 00000. 0000. I 000500. 0050. 000000. 0000.0I 0000000 0003 000 .00 A000. 0500. A000. 0500. 000000. 0050. 000000. 0000.0I 0000000 0003 .00 0000. 0000. 00000. 0050. I 000000. 0500. 000000. 0000.0I A0000 000000 00 0000 .0 0000.00000. A500. 0000. I 000000. 0050. 000000. 0000.0I A0000 0000 000 .0 0000. 0000. A000. 0500. I 000050. 0000. 000500. 0000.0I A0000 00000 .5 0000.00000. A000. 0000. I 000050. 0000. 000000. 0000.0I‘ A0000 00000 .0 0000.00500. A000. 0000. 00000. 0000. 000000.00000.0I A000N0005 0000 0000000 .0 0000.00000. 0000. 0000. I 0000. 0000. 000000. 0005.0- 0ww0mm0momm0M0MW0mwmm .0 0000.00000. A000. 0000. I 000000. 0000. 000000. 0000.0I 000000000 0000 0000000 .0 0050. 0000. 00500. 0500. I 000000. 0500. 000000. 0000.0I A0000 0000 0000 0000 .0 0000.00000. A000. 0000. I 000000. 0500. 000050. 0000.0I 00000 0000 000000 .0 00000002000m Hwflpsm 0 0000000 m 000m 000500 .ummmm .nzmmxm 00000000 (‘1 APPENDIX E EXPENDITURE ELASTICITIES FOR SELECTED INDIVIDUAL FOOD ITEMS (BASED ON A WEEK'S EXPENDITURES) 1149 ILSO *Awmo.v **A~m0.V *xfimmo.v *xAHmo.v **Amwo.v *xfiomo.v **Amoa.v **Ammo.v A220. NOH. mmm. mom. 5mm. I Has. mom. I 3mm. I mow. omo. mMHDMumwm> Smmkm .mH Ammo.v **Ammo.v *onwo.v **Ammo.v **Aomo.v xxflamo.v *onHH.V **Ammo.v Amzo. mwanmumMm> mmo. mow. 5mm. mom. I mam. mmH. I mHm. I 3:0. mmo. Ummwm00%m w Smwhm .3H Ammo.v **AHHH.V **Amwo.v *xAmmo.v *xAmmo.v *Amwo.v **Amafi.v xxfimmo.v A520. wmofish mac. :HN. 0mm. mmm. I Hmm. maa. I 25m. I mam. Nmo. UHSLE UQCCMO .MH *xflmzo.v *xfimwo.v **A::Q.V *xAmmo.V *xfimmo.v **Az:o.v **Awmo.v xxflwmo.v **Ammo. mmm. mmm. N53. mam. mwm. 0mm. mNH. mmm. 0mm. muHSLm Umccwo .NH *xfiflmo.v *xANmo.v **Ammo.v Awwo.v xxfimmo.v *xflmmo.v Amoa.v xxAmwo.v **Ao:o. Nmm. mmw. Nam. H50. wwu. mmfl. 0:0. mow. mam. mpfidhm Swmhm .HH xxflnflfl.v «*ANHH.V **AHNQ.V **A000.V **A:OQ.V AHNQ.V *mAmmH.V **ANOH.V *xflmmo. mpfizpm mmo. HNO.H 3mm. Nmm. I mmm. mwo. l mom. I m©0.H oma. UmmmmoOhm w Qmmhm .0H Aamo.v *xAmHH.V *xfimmo.v *xflomc.v **Ammo.v *xnwwo.v **ANHH.V xxflmofi.v *xfimso. Mme. m:O.H mmm. mmz. I mHN. Ham. I Nmz. I 5mm. :Hm. wwwm .m *Aflmo.v **ANHH.V *xAmmo.v **Awmo.v xxfimmo.v *xflmuo.v *xAmHH.V xxANCH.V Aomo. - mHH. 2mm. mmm. mmm. I mom. mzfi. I Nmm. I HON. Ono. muwwz UmmmmOOhm .m xxflnmo.v AmoH.V *Ammo.v *xflmwo.v Ammo.v *xAmuo.V **Amca.v Ammo.v **A::o. Uoommom mmn. mad. MHH. Nmm. I mac. mas. I mNN. I mmH. was. hmcpo w Smfim .N Ammo.v *xfiwma.v **Ammo.v **Ammo.v *xAmoH.V Amwo.v **Ammfi.v **A3HH.V x*A:mo. mmo. N:N.H mmw. 3mm. I wmm. :30. I mmm. I HOH.H mmm. hhuaflom .w *xnmmo.v *mAmmH.V *xAHmo.v *Ammo.v **Awmo.v Ammo.v *xAan.V xxflmofi.v *aAmmo. mom. 30H.H 0mm. mNH. I mow. NWO. mmz. I hmO.H @mm. Aflmmhmv xhom .m Azwo.v *xfimmfi.v **Ammo.v **A©OH.V **AmOH.V Amwo.v **Amza.v xxfloma.v **Ammo. mOH. mmm.a :NN. mmm. I ::O.H 000. I mmm. I :mm.a mmm. Mmmm .3 *xflnmo.v *xfimma.v **Ammo.v *onmH.V xxflmoa.v **Ammo.v **Amma.v **A:mH.v xxAzoo. mos. mww. Hmm. Cam. I mam. GHQ. I wmm.HI mmm. mom. mummz HH< .m **pro.v xxAHMH.V **Aamo.v *onHH.V *xAmoH.V *xfimmo.v xxfimzfl.v **Aowo. mam. 03>. mmm. 5mm. I 2:0. Hmm. I 5mm. I mom. mposzLm hQHmQ .m **Ammo.v *xfimma.v *xAmmo.v Ammo.v *xfiaoa.v **ANQO.V **ANMH.V **AOOH.V *xAHmo. mEoz mnfimpzo :ma. mow. wmm. HOG. I mmw. mzfi. Hmm. I 0mm. 0mm. COHUQESWCOO Uoom .H AHHmm 0HH0200H mma. I mmH. NzH. I 0mm. I mmo. now. I NwO. I :Om. mom. I fi mUH3hm UmHhQ .NH txAmmo.v **Ammo.v *AHmo.v xxAmmo.v xxAmNo.v **Amwo.v xxAHHH.V xxAmmo.v *gfimzo.v macho mom. I N02. OOH. mos. I mmm. was. I HON. I mHm. MNN. I poom a mCHMumHm .0H mpmcp AHH¢vooom gum“. cmmswmzieg 5%... MEVWMWM 33.. memmgo 2152 «*ANHH.V mmfimoa.v *mfiamo.v *xfimmfi.v *xfioHH.V *mnm0H.v **AmmH.v xxfiNmo.v mmfimwo.v :HN. NNN. ONO. HOO. MNm. :NO. NOO. Nwm. H20. meDwumwm> Smmfim .mH **A:HH.O *wAmoa.v aaAmmo.v *uaama.v *mAHHH.V asAmoa.V **Ammfi.v m*Ammo.v a.AwNo.V mwanmumwm> OON. HON. cow. mom. mmm. mmm. mmm. mmm. mmw. nmmmmOOLm a nmmpm .:H gammHH.v axfiwaH.v **ANOH.V *aAmMH.v mxAzmH.V saAooH.v *xflazH.v **Ammo.v :anomo.v mm0H5h NNm. 3mm. MNz. NJO. Nmm. OmN. mzN. Omm. O20. ufifihm fimccmo .MH *Azmo.v **AmNo.v *Amoo.v Asma.v *Ammo.v AmNo.v AmoH.v Ammo.v Ammo.v OmH. mmH. I MNH. I OMH. :mH. I NmO. NzH. NNO. I OHO. muHSHm Umccmo .NH AmoH.v AHmo.v xtflamo.v xxfimmH.v *xfimmo.v *mAmmo.v *ANNH.O *mAmNo.v moo.v Omo. mmO. I :OH. I Nmm. OOH. I MNN. mHN. OOH. I ONO. mquhm Smmhm .HH **ANMH.V :wAmmH.v **AOHH.V *wfimmH.v AmMH.V **AmmH.v xonwH. *AmOH.v *afimmo.v mpfizpm ONO. HNz. HNm. ONO. mON. OOO. OON. HON. Omw. UmmmmOOhm a nmmkm .OH **Ammfi.v Aomfl.v Aaofl.v .xfimzfi.v AHRH.V *xflmfifi.v .*AN:H.V AMQH.V .*A:mo.v mam. owe. ”Ho. mNm. oma. I me. mow. wmfi. I mam. mmmm .m **AmmH.v xxAmmH.v *xAmoH.v 3*Aoza.v **A:MH.O axAMHH.V *mAmna.v *xflmoH.v xxAmmo.v ONO. NNN. ONN. ONO. zmz. OOO. mwO. NOm. mOO. mummz UmmmmOOLm .O **AmOH.v *mANoH.V xtflwmo. *aAomH.V *xflmHH.y *xAmoo.v **AmmH.v *mAzmo.O *xflmNo.v nooummm HOO. OOm.H mNN.H OmO.H NON.H ONO.H ONO.H MOH.H OON.H hmnuo w EmHm .N **ANMH.V xxfimmH.v *tfimHH.V **HmmH.V *xAmnH.v *xAmm .V xxAHwH.V *xAmHH.V *mfimmo.v I mmn. mmz. mmm. Ham. mwm. 3mm. wmm. mfim. Fm». zppfisom .m **A:mH.V mmfiNmH.v *xfim0H.v *xAmzH.v AzmH.O xxflmHH.v *xnmmfl.v A20H.v *afiwmo.v mmm. NOM. NOM. HOO. OOH. 23m. NOO. OHH. NOm. Acmmhmv xhom .m asANzH.V :*Amma.v **AmmH.v *onNH.v *AomH.V **Aoma.v xxflwNH.V *AmHH.O **AOOH.O Omo. NHm. zmz. Omo.H OmN. me. OmO. HNN. HaN. wwmm .2 maAme.v **AOMH.V *xAmmH.V xxflmmH.v xxnmmH.V **AmmH.V *onmH.V xxfimmH.v *xfimoH.v :MN.H H:O.H NN:.H MNm.H NHm.H NmO.H ON:.H NOH.H MHO.H mummz HH< .m **A:mH.V *xaNma.V *xfimmH.v *xAmNH.V *xAHmH.V xxfiaqa.v xxAmmH.v *xnmoH.V ONO.H NNH.H N:O.H ONm.H :NO. OOm.H H:N.H MNN.H WUOSUOLN hLHmQ .N **A©mH.V AmmH.V AmOH.V xxfiNmH.V AmzH.v **AmHH.V xxmme.V Ammo.v **ANmo.v mEom mnfimpzo mm. HOH. HNH. MHw. OOH. OOH. NNz. NOH. ONm. COHuQEDmCOO 600m .H AHHdeOOm mpmcpmm mmepmm HH< ncmHmH HH< : , HH< coo .,. mmmzucoz mum: A MW ummm :wsw mam Afiflqvfimpso AHHmm 0HH0£00H a :OH.H MNm.H 2mm.H OON.H NON.H OHm.H H:N.H NMN.H Omz.H mquhm UmHhQ .NH *zAOHH.V xxAmoH.V xxAzmo.V *xflmmH.v *xAHHH.V **ANQH.V **AmmH.V *mfiomo.v *mfiNNo.v macho HOO. ONO.H OMO.H HHH.H HmO. OOH.H OOO.H OHO. HNH.H poom w wCHMUmHm .OH AHH4 If] ~L 3 .C (II 4.) U) L: 0 (Q) (\l 158 $05. meo. 13m? meH. 2.330. mem. .2192. Vmoo.mI mpmcpmm 3%: coz .HH Ammm. meo. *3Hm5o. meH. I $.Ammo. mem. 3*AHmm. Vomm.mI mpmcpmm mwmz .OH Hmmm. Vo5o. *Amoo. meH. I **H33o. V553. 3*AN3H. Vm3m.mI AHHHV ccmHmH mo 3mmm .m Awm5. meo. A3mH. VmMH. 3*Hmmo. Vm3m. **AOHm. mem.mI HHHHV swam cmm .m HH5m. meo. *Ammo. V3mH. I .3Hmmo. V5om. 3*AmmH. V5m5.mI AHHHV Hmpsm .5 Hmm5. meo. Am5o. meo. 3*A33o. Vomm. 3*HmmH. meo.mI AHH Ummmmoomm w gmmm rrI m mNHm NHHfimm .pmmHm .Ucmnxm pcmpmcoo (\J Ammo.HVomo. **HmHH. VHmm. *.Hmmo. Vmom. I **Amom. Vm35. I wmepmm mmmz 202 .HH “mom. meH. .*HMOH. V05o.H .IHmmo. V503. .3H3Hm. V3mH.mI mpmcpmm mwmz .OH Aamm. vzmo. **A:mo. vmmo.H *AHwo. vooa. **Amma. vmmm.HI HHH Qmmpm pmgpo Im mNHm 5HHam 4 I U) (I5 H LL] U Q) 0 >4 LL] 49 at) o. Mum ( ) ( [TI (\ r 163 Amoo.Hmeo. *.AmOH. Vom5. *gHmmo. V3mH. I *3HomH. mem.HI mumcpmm mmmz coz .HH “mam. .*AMHH. mem. .3AwOH. Vmom. *3Hm3m. Vomw.mI mpmcpmm mmmz .OH Awoo.HVm3o. *3Hmmo. VNH5. 3*A3mo. V3mH. 3*Amom. mem.mI AHH @meo 3 mpHspm : mNHm mMHfimm .ummHm .wmmaxm pcmpmcoo (\l 164 Ammm. meo. 3*Homo. V5mm. ..Hm3o. mem. *3HmmH. Vmwm.mI mpmsamm mwmz :02 .HH Hmm5. Vm3o. A3mo. meH. ..Homo. VmHm. *.A5mm. VHH5.mI mpmcnmm mwmz .OH Amm5. meo. *Hm5o. VSMH. .*Hm3o. V5H3. 3*HmmH. mem.mI AHHHV ccmHmH mo pmmm .m Homm. V3Ho. 33H3OH. mem. *memo. Vme. 3*Hmmm. mem.mI HHHmm OHHOQOOHHICOZ .Hmnpo a UmprODeHmo H00H.HV030. 33A00H. V000.H 3*Hm00. V300. I 33A0Hm. VHH0. I 0000000 0003 :02 .HH H000.HVH0H. 0*H3HH. V00m.H 0.300H. V003. 3*H03m. V300.0I 0000000 0003 .0H H000.HV000. 03H00H. V05H.H H500. VO0H. 3*H0Hm. V000.HI HHHOV 0c0HOH mo 3000 .0 A0HH.HVm00. *gHm3H. V00N.H 0*H000. VmH0. I A0H0. V050. HHH