ABSTRACT ANALYSIS OF INTRASEASONAL POTATO PRICE MOVEMENTS By.Ivar Kristianslund Potato prices are subject to very wide price fluctuations both from year to year and during any particu- lar year. ‘The intraseasonal price movements vary from year to year and are very difficult to predict. The major objective of the present study was to formulate econometric models that may explain short term fluctuations in potato prices, and to estimate the parameters of these models. The estimated parameters of twelveeconometric models have been presented, one model for each month of the year. Each model has been estimated on the basis of monthly data for the United States as a whole for the years 1952-66. The models all conSist of a supply relationfor potatoes, a demand relation for potatoes, and an identity. Each model was estimated by seven different' _estimation procedures, including ordinary-least,squares, two-stage squares, limited information maximum likelihood, three—stage least squares, and full information maximum 11- kelihood. A unique feature of the present work is that monthly sales data for potatoes for each state and for the United States as a whole for all of the sample period Ivar Kristianslund have been estimated by the writer and used in the models. This estimation was based mainly on shipments and unloads data and on production and sales data for each seasonal crop, by states. The use of the estimated monthly sales data made possible the estimation of price elasticities of supply and demand for potatoes for each month of the year. Several other parameters were also estimated, among these coefficients showing the effects of changes in production of the various seasonal crops on supply in particular months. Various economic models of the potato market were discussed, and the ones that were finally estimated were a result of a compromise because several of the desired data were lacking and others were deficient. The models may therefore be subject to specificatioh errors and errors in variables that may have biassed the results. Several of the estimated coefficients were not statisti- cally significant different from zero, but in some cases the corresponding variables were still retained in the models on the basis of economic reasoning. The results presented in the following should therefore.not be con- sidered as final facts, but as preliminary insights that ought to stimulate new research in accordance with the principles initiated in this thesis. Theestimated demand elasticities had absolute Ivar Kristianslund values that were generally quite low compared to results from most of the earlier investigations. Reasons are given why many earlier results may be biassed. Some of the supply and demand elasticities and some of the coefficients of the production variables in the supply relations had "unnormal" signs. Although single esti- mates may have wrong signs and magnitudes, heavy evidence is presented, both from economic reasoning and from earlier works, indicating that many of the "unnor- mal" results of the present work may be normal after all. If this is true, very important policy implications emerge. More research is therefore needed to test the results obtained. Several suggestions for research along theSe lines have been set forth in this thesis. ANALYSIS OF INTRASEASONAL POTATO PRICE MOVEMENTS By Ivar Kristianslund A THESIS. Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1971 ACKNOWLEDGEMENTS A number of people have assisted the writer during the research that lead to this thesis and in the preparation of the manuscript. This help is greatly appreciated by the writer. Dr. Lester V. Manderscheid served as the writer's academic advisor and chairman of the thesis committee. He has given much good advice and has effected useful contacts with various key persons. He also spent a large amount of time reading through a preliminary outline and the final draft of this manu- script. Some improvements suggested by him have been incorporated in order to remove linguistic errors and make the text clearer at certain points. In June 1965 the writer spent some days in the United States Department of Agriculture in Washington, D. C. and had very useful talks with several economists and statisticians. Among the many who gave the writer useful information and suggestions the following will be mentioned in particular: Dana G. Dalrymple, Don Kuryloski, Olman Hee, Will Simmons, Fred Waugh, Tono Rojko, Forrest E. Scott, E. J. Holcomb, Robert V. Akeley, and Stephen J. Hiemstra. The use of the crucial ship- ments and unloads data in the present work would have ii been impossible without the wonderful cooperation of Mr. J. M. Saylor of the Agricultural Marketing Service. In August 1965 the writer visited Idaho and Washington and benefited from conversations with people connected with the potato industry. Among these, Mr. Clarence E. White provided useful data that were originally planned to be included in the econometric analyses. The writer has also received valuable data from various officials in Maine. In August 1965 the writer presented a seminar on the subject of his thesis research in the Department of Agricultural Economics, MSU and received valuable criticism from the faculty. Dr. William Ruble is the creator of the marvelous standard programs that have been used to esti- mate the models, and he also gave the writer instruction on particular problems. The remaining programming was. conducted by Mrs. Arlene King, Mrs. Mary Merillat, Mrs. Jackie Musell, and Mrs. Laura Flanders. The accu- rate and patient cooperation of Mrs. Flanders while the writer was in Norway was invaluable. Several typists participated in typing various parts and drafts of this thesis, and thanks are due to them all. The present work was made possible through the financial support of W. K. Kellogg Foundation, The iii Department of Agricultural Economics, Michigan State University, The Agricultural Research Council of Norway, The United States Government, and The Agricultural College of Norway. The writer especially wants to thank Professor Lawrence L. Boger, Chairman of the Department of Agricultural Economics, for his encou- ragement. Finally, the writer wants to thank his wife, Leikny, for her great patience and understanding during the preparation of the thesis. Although many people have made contributions, the writer assumes full responsibility for possible errors. iv TABLE OF CONTENTS Chapter P350 I. INTRODUCTION . . . . . . . . . . . . . . . l The problem 1 Importance of the Problem 2 Objectives of the Study 2 General Review of the Potato Economy A Procedure 7 Basic concepts 8 Aggregation of utilizations 9 Geographical aggregation 10 Time unit 10 Type of model and estimation procedure 13 Sample period 13 II. SOME PREVIOUS STUDIES 15 III. MAJOR DAT SERIES AND THEIR SOURCES 2A Data from the United States Department of Agriculture 2A Shipments and unloads data 2A Shipments data 2“ Unloads data 30 Price data 33 Farm retail spreads 3A Utilization data 34 Data on diversion of potatoes to starch, flour, and livestock feed under govern- ment programs 36 Data on production, sales from farms, and related magnitudes 37 Data from Other Sources 39 EXports and imports by months 39 Income and population data 40 Rail conversion factors A1 Various price index numbers from the United States Department of Labor A2 IV. ESTIMATION OF MONTHLY SALES DATA ON THE BASIS OF SEASONAL SALES DATA AND MONTHLY SHIPMENTS DATA . . . . . . . . . . . . . . . . . . . . A3 V. VI. VII. Introduction Description of Method Evaluation of the Estimated Data THE SUPPLY OF POTATOES DURING THE YEAR . . The Question whether Potato Yields and Potato Production are Endogenous The Effects of Production on Marketings The Determination of the Time of Harvest and Marketing Potatoes to be marketed shortly after harvest Storage potatoes Conclusion The Influence of Home Use for Food, Seed; Feed, Shrinkage and Loss, and Their Variations Foreign Supply of Potatoes Stocks THE OFF FARM DEMAND FOR POTATOES DURING THE YEAR . . . . . . . . . . . . . . . . . . . Demand for Seed Potatoes Demand for Feed Potatoes Demand for Potatoes Fresh for Food Demand for Potatoes for Processing Demand for Potatoes by the Government for Diversion Purposes Demand for Potatoes for Storage TUE MODELS . . . . . . . . . . . . . . . . . Some General Considerations Farm-Retail Spread Relations The Economic Model Functional Form and Statistical Assumptions Identification vi Page too 7‘- A- \l H xrq 5&4 HF4 Q W KO CD 03 O.) k ) \‘l G\ l—’ {EMU C) O \O ‘1‘ ‘O \J U) \1 Chapter Page VIII. THE ESTIMATED STRUCTURES . . . . . . . . . . . IOB Estimation Procedures 108 Criteria for Choosing between Submodels 110 Tests of Significance 113 Introductory Remarks 115 The Supply Relations 116 The Demand Relations 126 The Farm-Retail Spread Relations 135 The Reduced Form Relations 136 Testing for Possible Serial Correlation in the Disturbance Terms 1A1 IX. SOME IMPLICATIONS AND USES OF THE RESULTS AND SUGGESTIONS FOR FURTHER RESEARCH . . . . 1AA Evaluation of the Results 1AA Comments on Elasticities Obtained in the Present and Earlier Investigations 1A6 Some Implications of Low Elasticities of Supply and Demand, Some of Which Rave Unexpected Signs 15A Some Suggestions for Further Research 156 Appendices . . . . . . . . . . . . . . . . . . . . . 161 RIBIJICDGRAPTTY o o o o o o o o o o o e o o o O D ' ' ° 229 vii Table 10. 11. LIST OF TABLES Estimated Supply Relations by Months for the Period 1952-66. January-March . . . . Estimated Supply Relations by Months for the Period 1952-66. April-June. . . . . . Estimated Supply Relations by Months for the Period 1952~66. July~December . . . . Estimated Demand Relations by Months for the Period 1952-66. January-December . . . . . Farm-Retail Spread Relations by Months for the Period 1952-66. Estimated by the Method of Ordinary Least Squares . . . . . . . . . . . Reduced Form Relations Corresponding to YSE bv Months for the Period 1952-66 Estimated Directly by the Method of Least Squares. . . Potatoes for All Purposes Except Diversion: Estimated Sales from Farms by Months, United States, 1951-66 . . . . . . . . . . . Deflated Average Price Received by Farmers for Potatoes: United States, by Months, 1950- 66 O O O O 0‘. O O O O O O O O O O O O O O 0 Total Potato Production: United States, by Seasonal Groups, 1951-66 . . . . . . . . Total Quantities of Potatoes Diverted under Government Programs: United States, by Months, 1952-66 . Total Quantities of Potatoes Diverted under Government Programs: By States and by Marketing Years, Crops of 1953-63 . . . . . Net Exports of Potatoes: United States Totals, by Months, 1951-66 . viii Page 167 170 173 179 190 191 195 197 199 201 202 13. Maine Certified Seed Shipments by Months, 1952-66 . . . . . . . . . . . . . . . . . . 203 1A. Imports of White Certified Seed Potatoes: United States Totals (Mainly from Canada), 13:] montk’ls, 1952-66 0 O I O O O C O O O O . TEA/3L; 15. Potato Processing in Idaho Plus Idaho Potatoes Processed in Malheur County, Oregon. By Months. Food Only . . . . . . 207 16. Potato Processing in Maine by Months, 1961-6; 206 17. Frozen French Fried Potatoes, End-of-Month Cold Storage Holdings: United States Totals, 1956-66 . . . . . . . . . . . . . . P97 1?. Potatoes for All Purposes Except Diversion: Calculated Shipments .8 Percentage of Sales from Farms, by Months and by States or by Groups of States, Years Beginning 1951—66 . 226 LIST OF APPENDICES Appendix Page A. Results of Econometric Analyses . . . . . 161 B. Som Data Used in the Analyses . . . . . . 194 C. Further Details Regarding the Estimation of Monthly Sales Data for Potatoes. . . 208 CHAPTER I INTRODUCTION The Problem Potato prices vary considerably not only between years, but often also from month to month within a given year. These intraseasonal price movements are a major ob- ject of the present study. Usually they follow different patterns in different years. This being the case, the following question arises quite naturally: Given a year with certain attributes or characteristics; what will be the resulting price pattern? If the ideas of causal and interdependent rela- tionships between potato prices and other variables are introduced, the question may, briefly speaking, be split into two new questions: (1) What are the major variables affecting potato prices during the year? (2) What are the quantitative relationships between changes in each of these variables and changes in potato prices? The first question can be analyzed by the combined use of economic theory and general information on or knowledge of how the industry operates. For an analysis of the second ques- tion, econometric theory and econometric methods are additional tools of great usefulness. l Importance g£_the Problem If the two questions raised in the preceding section can be answered with some precision, this gives a basis for making relatively well—founded predictions of the price pattern that is likely to occur in a parti- cular season. Such predictions are of considerable interest to potato growers and to others who are directly involved in the potato business. The availability of good price predictions or similar information will give them a basis for choosing a proper timing of their opera- tions and transactions during the season. This means that they will be able to make a more efficient adjustment to the particular economic situation at a given point in time. The likely results of this are not only higher and more stable incomes to farmers and other groups, but also benefits to the rest of society. Some knowledge of the economic structure by which potato prices are determined can be of considerable help in selecting a good program when governmental programs such as diversion programs are suggested. It can also be useful in predicting the effects of various other structu- ral changes when such changes take place. Objectives 3: the Study The major objectives of the study were: (1) to formulate econometric models of the Unites States' potato economy that could explain potato price fluctuations from period to period within a year and (2) to estimate the parameters of these models. A more subordinate objective was to investigate whether data from the Unites States Department of Agriculture on rail and truck shipments of potatoes could be used with benefit in analyses like the present. It was the intention that the models with their estimated parameters should provide some basic knowledge needed to make predictions of potato prices at the farm level for various periods of a year. Still, since the investigator expected to solve only part of the predic- tion problem, he felt free to include in his analysis. some variables for which no values are available early in the season. The very basic step in predicting potato prices is to learn how they are determined. Variables that are thought to be important must therefore be included until more is known about the price determina- tion. The models were formulated with the aim that they should increase our understanding of how some key variables pertaining to the potato market are deter- mined. Thereby they should also give some guidance for choosing between policy measures. The more detailed limitations on the objectives of the study will be clar from the section on procedure. General Review of the Potato Economy The following paragraphs contain some background information on the potato sector that seems relevant for a proper evaluation of the ideas, arguments, and results to be presented later. Potatoes are grown commercially in every state of continental United States. In fact, however, the bulk of the production is highly concentrated to particular areas within certain states.1 The most important states in terms of potato quantities sold, namely Idaho, Maine, California, and New York, accounted for 52% of total potato sales from the 1962 crop. Together with the next six states, the last of which was Michigan, they made up 77% of total sales.2 Harvesting of potatoes takes place somewhere in the United States in every month of the year. Because of differences in harvesting time and in storability of the potatoes, the group of states supplying the market is different in different parts of the year. 1For an excellent colour map of principal potato producing areas in the United States, classified by six seasons, see August E. Kehr, Robert V. Akeley, and Geoffrey V. C. Houghland, Commercial Potato Production, Agriculture Handbook No. 267, Agricultural Research Ser- vice, U. S. Dept. of Agriculture (Washington: U. S. Govern- ment Printing Office, July, 196A), p. 10. This publica- tion also contains other useful background information. 2Percentages computed from data in U. S. Dept. of Agriculture, Statistical Reporting Service, Crop Reporting Board, Potatoes and Sweetpotatoes: Estimates hy States and Seasonal Groups--Crops g: 1962 and 1963 Washington: August, 196A7: p. 11. The U. S. Department of Agriculture classifies the various potato crops according to the time when the bulk of the crop is usually harvested. There are six seasonal groups: Winter, Early Spring, Late Spgihg, Early Summer. Late Summer, and Fall. It may aid the memory to lump the early and late crops together and note that the bulk of the Winter, Spring, Summer and Fall crops are harvested during the first, second, third and fourth quarter of the calendar year, respec- tively. The early crop is harvested during the first half of the quarter while the corresponding late crop is harvested during the second half. In most states one or two seasonal groups of potatoes are grown. For North Carolina and Texas the number is three. In California potatoes belonging to five different seasonal groups are grown at the present.1 Since there is some overlapping and variation in harvesting time, the seasonal classification is neces— sarily somewhat arbitrary.2 This is especially true for 1The seasonal groups of potatoes grown in each state are listed in Reginald Royston, Oakley M. Frost, and Frasier T. Galloway, Potatoes and Sweetpotatoes: Usual Dates for Planting, Harvesfing, and Marketing 9y Seasons, in Principal Areas, Agriculture Handbook No. 127, Crop Reporting Board, Agricultural Marketing Service, U. 5. Dept. of Agriculture (Washington: U. S. Government Printing Office, June, 1957), p. 5. This publication has been a most valuable source of information for the pre- sent investigation. It contains dates for each individual state. 2Ibid., p. 3. the distinction between Late Summer and Fall potatoes. The consumer has no way of distinguishing between these two crops, and the grower has a definite possibility for varying the harvesting time for the Late Summer crop in response to economic stimuli.l Potatoes from the Fall crop are marketed through- ,out most of the year. The usual marketing periods for the other crops, with the exception of some Late Summer potatoes, generally follow shortly after harvest.2 The Fall crop is by far the most important crop. In 1958 the percentages of total production accounted for by the various seasonal categories were as follows: Winter 2.2, Early Spring 2.1, Late Spring 10.3, Early Summer 6.2, Late Summer 13.5, and Fall 65.7.3 Potatoes are used fresh for food (5A,3%), for processing (23,9%), for seed (7,9%), and for liVestock feed, shrinkage and loss (9,2%). The percentages shown in parentheses refer to production in 1962. Minor quan- tities were also used for non—food industrial uses (starch) or were exported. Almost half of the processed 1The content of this statement has been pointed out to the writer by several officials of the U. S. Dept. of Agriculture. 2Royston, Frost, and Galloway, gp. cit., p. A. 3Pinhas Zusman, "Econometric Analysis of the Market for California Early Potatoes," Hilgardia, Vol. XXXIII (December, 1962), p. 55A. The organization of the present section is partly borrowed from Zusman's work. quantity was used for potato chips. Almost two-thirds of the other half were used for frozen prepared potato 8 products, mainly frozen french fries. Potato utilization as well as production varies from year to year and has also been subject to certain trends in the past. Most remarkable is the rapid increase in processing. In 1951 18 million hundredweight or 9.2% of production was processed. The corresponding figures for 1962 were 6A million hundredweight or 23.9%. Fresh consumption has declined correspondingly so as to keep the level of total per capita consumption approxi- mately constant since 1951.1 Procedure . The present section gives an outlook over the approach that was taken to fulfill the objectives of the study. Some reasons why this particular approach was chosen are also given. 1The figures in this paragraph and in the pre- ceding one were quoted from or computed on the basis of figures from National Potato Council, U. S. Production, Utilization, and Use of Designated Crops. (IWashingtonJ: National Potato Coun011, January, 196A). For a more detailed discussion of trends and changes in consumption, production, marketing and utili- zation of potatoes, see Will M. Simmons, An Economic Study of the U. S. Potato Industry, Agricultural Econo- mic Report No. 6, Economic and Statistical Analysis Division, Economic Research Service, U. S. Dept. of Agri- culture ([Washington]: March, 1962), pp. 1- -20 and pp. 2A- -29. See also U. S. Dept. of Agriculture, Potato Charts and Tables ([Washington]: [1965]), pp 2- 32. Basic Concepts Ample use was made of the well-known concepts of supply and demand. In principle the farm gate was con- sidered the dividing line between suppliers and demanders. Now, some potatoes are used on the farms where they are grown: They are used for seed or livestock feed, they are consumed in households, or they disappear as shrinkage and losses. These potatoes used on farms where grown were not considered as demand in the present analysis, but they were taken into account in that they were allow- ed to affect supply from farms. Supplied quantity in this analysis is identical with quantity sold from farms. Foreign supplies were taken into account in that they were subtracted from exports to give net exports. The demand concept used was demand at the farm level. Potatoes are demanded for several purposes: fresh use, various forms of processing and manufacture, seed, livestock feed, diversion by the government to inferior uses, addition to stocks held by local dealers and others, and net exports. The two last mentioned kinds of demands may be either positive or negative. Potatoes are often bought and resold several times until they are finally demanded for one of the l purposes listed above and thereby removed from the lStock changes excepted. market. Conceptually, however, one may disregard most of these intermediate transactions and concentrate the atten- tion on final demands and their determinants. Changes in the factors determining ultimate demands are then thought to be reflected backwards via the various marketing channels to the farm level. The fact that some potatoes are sold to processors or others according to contract before they are harvested or even before they are planted caused some conceptual difficulties. Since no separate data are available on these sales, about the only thing that could be done, without going into an extensive amount of detail, was to assume that these sales, on the average, did not serious- ly bias the results of the analysis in any particular way. The bulk of the potatoes are sold in the open market. Aggregation of Utilizations It would have been very interesting to have sepa- rate demand equations for various utilizations such as fresh use, processing, seed, feed, etc. If demand elasti- cities are different, as they probably are, estimates of their magnitude would have provided useful information. Since the necessary data on utilization are not available except on a yearly basis, however, most demands had to be aggregated into one group in the models. 10 Geographical Aggregation Potatoes are shipped very long distances: from one end of the country to another. Potato markets in different parts of the country are therefore directly or indirectly interrelated. At the same time, because of the bulk and weight of the commodity and the long distan- ces, transportation costs are considerable, and prices for the same quality of potatoes may differ Quite a bit between regions at times. For some purposes it would have been useful to divide the country into a relatively small number of regions and to treat these simultaneously in one model. The complexity of the market and the unavailability of some crucial data series made this very difficult and time-consuming, however. It was therefore decided to deal with the U. S. as a whole. It is clear enough, how- ever, that as far as supply is concerned, the use of - short time periods in the analyses is to some extent a substitute for geographical disaggregation. Time Unit The relevant choices were to use models based either on monthly data, seasonal data,1 or data for cer- 1I. e. data pertaining to each of the six time periods used by the U. S. Dept. of Agriculture to de- fine seasonal groups of potatoes. See above, p. 5. ll tain aggragates of seasons. The more aggregated models are the easier ones to deal with, but they provide less information than the disaggregated ones. Since the marketing periods for various seasonal groups of potatoes overlap, and especially since Fall potatoes are marketed in considerable amounts during the harvesting periods of most other potato crops, the total production or sales figures that are available for each seasonal group do not usually tell very much about total consumption or purchases of potatoes during each seasonal period. The stocks data that are available for the Fall crop give some information on the rate of disappearance of Fall potatoes during certain periods of the year, but there are two reasons why these data are not well suited for being used together with data on total sales to deter- mine the rate of marketing of the Fall crop from farms: (1) The stocks data include not only stocks held by farmers, but also stocks held by local dealers. (2) The stocks held by farmers include not only stocks intended for sale, but also stocks intended for use on the farms for seed, etc. In sum, the seasonal data were not very well suited for an analysis of supply and demand at the farm 12 level based on distinct time periods like quarters and half quarters. They seemed to be about equally well suited for a similar analysis based on months. No more information on utilization exists for seasons than for months.1 In Chapter IV a method of converting seasonal sales data into monthly sales data by means of shipments data will be described. Provided that this method is basically sound, it seemed like the availability of data did not point strongly in the favour of a seasonal ana- lysis rather than a monthly analysis. Since much informa- tion might be gained by a disaggregation, it was decided to work with monthly data. Type of Model and Estimation Procedure It was decided to aim at estimating what econo- metricians call structural parameters, insofar as the available data would permit this to be done.2 Since the economic relationships under investigation involved more than one endogenous variable, it was clear in the light of 1Except for farm utilization which is of less importance in this connection. 2For a classical discussion of the desirability of structural estimation, see Jacob Marschak, "Economic Measurements for Policy and Prediction," Studies in 'Econometric Method, ed., Wm. C. Hood and Tjalling C. Koopmans, Cowles Commission for Research in Economics, Monograph No. 1A (New York: John Wiley & Sons, Inc., 1953), pp. 15-26. See also Arthur S. Goldberger, Econometric Theogy (New York: John Wiley & Sons, Inc., 196A), p. 365 and pp. 376-380. 13 economic theory that a simultaneous equation's model was called for if consistent estimates of the parameters were to be obtained. Several estimation procedures were compared: ordinary least squares, two-stage least squares, limited information maximum likelihood, three-stage least squares, and full information maximum likelihood. Sample Period The years directly included in the initial ana- lysis were 1952-6A. For the lagged variables, values for 1951 were also used. There are several reasons why years prior to 1952 were not included in the analysis: The price support operations that took place earlier were ended in 1950.1 Still, the acreage of Fall potatoes planted in 1950 was larger than in the following years, however, and this must have affected marketings in the first half of 1951. In 1951, price ceilings were also in effect.2 Another reason for limiting the analysis to the period 1952-6A was that processing played a much more 1For details regarding these operations and their ending, see Roger W. Gray, Vernon L. Sorenson, and Willard W. Cochrane, An Economic Analysis of the Impact of Govern- ment Programs— on the Potato Industry of the United States, North Central Regional Publication No. —A2 (University of Minnesota Agricultural Experiment Station, June, 195A), pp. 39-Al. 2This fact came to the present writer's attention at the reading of Martin 8. Simon, "Forecasting Potato Prices," 1961 American Potato Yearbook, ed., C. Stedman Macfarland, Jr. (New Jersey, By the editor), pp. 29-30. 1A important role in this period than earlier. If years prior to 19A9 had been included, additional data problems would also have emerged. The present investigation was almost completed at the end of 1965. For various personal reasons, the writing was not finished until 1971, however. Most of the experimentation with various models, described later in this work, were based on data for the years 1952-6A. At the end of 1967 and beginning of 1968 all data series were extended to cover the years 1965-66, however. At that time also some of the data for earlier years had been revised by the agencies collecting them, and the data series used in the computations were there- fore revised accordingly. Thereafter the models that had been considered the final ones at the end of 1965 were reestimated in a slightly revised form, on the basis of the updated time series for 1952-66. All the numerical results presented in this work are thus based on updated time series. CHAPTER II SOME PREVIOUS STUDIES In the present chapter a selection of earlier works on potato price movements during the year will be reviewed briefly. Space does not allow a recording of the numerical results from these studies. Attention will be concentrated on the relationships between the earlier studies and the present one in regard to methodology. Differences and similarities in scope and methods will be mentioned, and due credit will, as far as possible, be given to the earlier investigators for adopted informa- tion or ideas. A very early analysis of factors affecting potato prices is found in a publication by HolbrookLWorking.l Since the methodology Working used in analyzing intra- seasonal price movements is representative also for other early investigations, it will be described in some detail. The following quotation is illustrative: "The best way to begin a study of the price of potatoes, is by considering 1Holbrook Working, Factors Affecting the Price 9: Minnesota Potatoes, Technical Bulletin 29 (St. Paul: University of Minnesota Agricultural Experiment Station, October, 1925), pp. l-AO. 15 16 the factors which affect the average price for the entire season."1 Having measured the average effect of five factors which explain most of the changes from one year to the next in the average price for the season, Working pro- ceeds to deal with seasonal change in the following way: A table is presented that shows for each month price as a percentage below or above the average price for the season.2 These percentages are averages based on data for several years. Given a price forecast for the season, a price forecast for a particular month can be obtained by applying the appropriate percentage. The method just described represents a good early attack on the problem of seasonal price variation. Working himself was well aware of its inadequacy, how- ever. The state of the problem at the time of his writing may be summarized by the following quotation: It is not so easy to explain the changes in price during any one season. A general explanation of the causes of the changes can be given but no method has been found for determining what part of each change is due to each cause. 1Ibid., p. A. It should be noted that Working was primarily interested in the price of potatoes in Minnesota for the nine-month season from September to May. At the time of his study, econometric theory and the available data were more inadequate than today. Even though a major idea behind the present work is that Working's statement is wrong in general, it is admitted that the statement may have been more appropriate under the circumstances when it was stated. 2Ibid., p. 25. 3Ibid., p. 2A. 17 Frederick V. Waughl and R. B. Heflebower2 used essentially the same method as Working in dealing with seasonal prices changes. Heflebower, like Working, seems to have felt that much was still to be done: The monthly estimates of prices are less accurate than the estimates of the season's average prices. The movement of prices throughout the season is very 1rregular and prices are often higher in the fall than in the spring.3 The later studies to be mentioned are based on somewhat different methodology. Some investigators split the year into a couple of seasons--usua11y aggregates of seasons defined by the U. S. Department of Agriculture-- and deal with average prices for each of the seasons. Other investigators try to predict a monthly average price, but they deal with only one or a couple of isola- ted months, the price in which they feel is crucial. D. Milton Shuffett analyzed the factors that affect the price of what he called (1) the Early commer- cial crop, and (2) the Late Surplus crop.” Similar lFrederick V. Waugh, Forecasting Prices 9: EEE Jersey White Potatoes and Sweet Potatoes, Circular No. 78 (Trenton, New Jersey: State of New Jersey Dept. of Agri- culture, July, l92A), pp. 16-18. 2R. B. Heflebower, Factors Relating 39 the Price of Idaho Potatoes, Bulletin 166 (Moscow, Idaho: Univer- SIty of Idaho, June, 1929), pp. 6-8. 3Ibid., p. 8. ”D. Milton Shuffett, The Demand and Price Struc- ture for Selected Vegetables, Technical BUIletin No. 1105, Agricultural Marketing Service, U. 8. Dept. of Agriculture (Washington: U. S. Government Printing Office, December, 195A). pp- AA-67. 18 studies on a more disaggregated basis, geographicially, were made by Kenneth W. Meinken.l A study by Dana G. Dalrymple is different.2 The purpose of Dalrymple's study was "to explore a method of statistically predicting average August prices in March, early enough to influence planting intentions."3 Although the statistical results did not turn out very well--another indication that much was still to be done-- the publication serves as a useful reference work on the problem. Dalrymple has summarized the basic difficulties to be solved very clearly in the following statement: It therefore appears that the big problem in predic- ting monthly potato prices will not necessarily be the problem of measuring year-to-year supplies of potatoes, but of measuring the more elusive variation in month-to-month supplies-~greatly complicated by seasonal and economic variations in planting, har- vesting and marketing. This problem is indeed a formidable one and one which will probably need much more study if gonthly prices are to be predicted with accuracy. 1Kenneth W. Meinken, Factors that Affect Price and Distribution of New Jersey PEtatoes, BulletTn 786 (New Brunswick: New Jersey Angcultural Experiment Sta- tion Rutgers in cooperation with Maine Agricultural Ex- periment Station, June, 1957), pp. 17-29. 2Dana G. Dalrymple, Predicting August Potato Prices at Planting Time, Progress Report 29 (Storrs, CERREEtIEut: Storrs Agricultural Experiment Station and the Agricultural Extension Service, University of Connec- ticut, February, 1959. Reprinted 1962 by the Division of Agricultural Economics Programs, U. S. Dept. of Agri- culture), pp. l-A7. Ibid., p. 1. ulbid., p. 15. 19 In a study of the Michigan March price Ronald A. Hagaman followed an approach similar to the one of Dalrymple.l Again, the statistical results left much to be desired. Among the works that will be dealt with here, the next one to appear was a work by Martin 8. Simon.2 Simon was interested in forecasting prices of fall crop potatoes, and he used price series for Maine. He worked with three seasonal periods, namely (1) September-November, (2) December-February, and (3) March-MaY- -As much as four simultaneous equations seem to have been employed in the first period, the corresponding endogenous variables being price, free-market disappearance, government-assisted disappearance, and December 1 stocks. Details regarding model-specification and estimation are not presented. A chart for the second period shows a good fit for the two years (1959-60) immediately follwing the sample period, as well as for the sample period itself. Since the analyses were based upon observations for only eight years, the results, as presented, are difficult to evaluate in a fair way. No results are presented for the first period. For the third seasonal period, actual 1Ronald A. Hagaman, "An October Prediction of the Michigan March Price for Potatoes at the Farm Level," (unpublished Master's dissertation, Dept. of Agricultural Economics, Michigan State University, 1959), pp. l-9A. 2 Simon, pp. cit., pp. 27-33. 20 and computed prices are shown for the years 1955-60. The direction of price change from mid-January to March-May or to May was indicated correctly by the calculated prices or price forecasts for all the years shown (up- wards in all but one year). Very much was left to be desired in regard to indicating the magnitude, however. A report from the U. S. Department of Agriculture, by Will M. Simmons, was an important source of general information on the potato economy for the present work.1 The report also contains interesting results from regres- sion analyses of seasonal, regional and intraregional production-price interrelationships. Several equations are presented, but with the exceptions of the ones for Late Spring and Early Summer, they were all based on data for two or more seasonal groups pooled together. Among the results with implications for the present work was the fact that larger production of Winter and Spring potatoes was associated with higher prices for the Late Summer and Fall crops.2 The analyses considered so far, with the exception of Simon's work, were conducted by single equation techni- ques. In a study by Pinhas Zusman, a comprehensive econo- metric model was employed in analyzing the market for California early potatoes and the interrelationships of lSimmons, pp. 313., pp. 1-83. 2Ibid., pp. A9-53 and pp. 79-83. 21 this market with the rest of the United States' potato market.l Zusman's model consisted of fourteen equations, but the nature of the system required only four to be estimated simultaneously. Only two seasonal markets were distinguished, namely September-February and April- August. The Winter crop, which is small, was disregarded, and some simplifying assumptions were made in order to arrive at certain identities. Consumption data could then be derived from production data and data on stocks, seed use, etc. In spite of the seasonally aggregate nature of Zuzman's study compared with the present one, the study offered several valuable suggestions for the present work. An experience of Zusman, worth noting, is the following: An attempt to estimate a set of simultaneous demand relations at the farm level, for which separate price series [by seasonal groups ]exist, failed to yield reasonable results. It also failed to recog- nize explicitly the locational aspects of the system. In the second part of his study, Zusman analyzes the static and dynamic prOperties of the estimated model. An interesting result is the suggestion that the large lZusman,qpp._eit., pp. 539—668. 21bid., p. 57A. 22 observed short-run fluctuations in prices and quantities are primarily due to random shocks and only secondarily due to erratic changes in exogenous variables. If this is true, one might expect it to be even more so for monthly data. After the present work had been almost completed, 1 Since Hee's work a study by Olman Hee was published. became available so late, it will be commented on only briefly here. Models were estimated for the following cate- gories of potatoes: (1) Late Summer and Fall, (2) Winter and Early Spring, (3) Late Spring, and (A) Early Summer. The models were also tested for predicting ability by means of observations for three years beyond the period of fit. The predictions of seasonal average prices presen- ted for each of the four categories of potatoes were gene- rally not bad. No attempt was made to attack the more intricate problem of predicting prices for shorter periods, such as months, however. The works just reviewed and several other ones have thrown much light on the problem of price variation , for potatoes during the year. Relatively little success 1Olman Ree, Demand and Price Analysis for Potatoes, Technical Bulletin No. I380, Economic and StatIEFicaI Analysis Division, Economic Research Service, U. S. Dept. of Agriculture (Washington: U. S. Government Printing Office, July, 1967). 23 has been experienced in dealing with monthly, or even with seasonal, data, however. The problem of overlapping among seasons, which is a very important difficulty in dealing with short seasonal periods, has still not been solved. Drawing heavily on knowledge gathered by earlier investigators, the present writer attempts to approach this and the related problems in a principally new way. The road to follow was essentially pointed out by Dalrymple when he suggested measuring variations in month-to-month supplies.1 One of the most serious hin- drances for carrying out such a program is the lack of adequate data. The present writer attempts to build a bridge over this hindrance by systematically utilizing the available data on shipments, unloads, and sales of potatoes. lSee quotation above. CHAPTER III MAJOR DATA SERIES AND THEIR SOURCES Data from the United States Department 9; Agriculture Shipments and Unloads Data The shipments and unloads data are dealt with directly in three different places in this thesis. A recording of sources and general description of the data is given in this chapter. In Chapter IV the basic principles and procedures employed in using these data (together with data on total sales of potatoes by seasonal groups) for obtaining monthly sales data for potatoes are described. Appendix C contains some further details regarding the data and their use. Shipments data Monthly data on shipments of potatoes by various means of transportation (mainly rail and truck) are collected and reported for each state by the United States Department of Agriculture in an an 25 annual publication.1 Preliminary data for January through July, 1965 were obtained directly from the Department.2 In 1967 when the data series were up- dated, preliminary data for January through July 1967 were provided in the same way.3 The rail Shipments data through April were final. The shipments data were used very extensively (essentially as weights) in the present work in con- structing hitherto unavailable monthly sales data for potatoes. These sales data were used as primary data in the econometric analyses. Other research workers may be interested in using the same data or in con- structing similar data by an improved method. In the following the shipments data will therefore be dealt with in more detail than the other data used in the analyses. 1See U. 3. Dept. of Agriculture, Consumer and Marketing Service, Fruit and Vegetable Division, Market News Branch, Fresn Fruit and Vegetable Shipments by Com- modities, States, Months, Calender Year, 1966 (Washifig: ton: June, 1967), p. 17. Annual issues for 1950-66 by variouSIy denominated agen01es of the Dept. of Agricul- ture were used in the present work. Some data that are more disaggregated over time and geographically are also available. _ 2Letter from J. M. Saylor, In Charge, Transpor- tation Reports, Fruit and Vegetable Division, Agricul- tural Marketing Service, U. 8. Dept. of Agriculture, Washington, October 6, 1965. 3Idem, letter, November 9, 1967. 26 The major sources for the information given in this thesis regarding the shipments data were the pre- faces to the annual publications and the footnotes to the table in these.1 Since it was difficult to determine, in som cases, whether information given for one year, for a series of years, also pertained to the rest of the sampling period, and since also some other questions remained unanswered, a list of questions was sent to Mr. J. M. Saylor of the Fruit and Vegetable Division. The received letter with answers to these questions served as a supplementary source. A summary description of the data, based on the sources just mentioned, is given below. Where the given information is somewhat uncertain, terms are used that indicate this. The nature of the sources of the shipments data is indicated in the following quotation: Arrangements are maintained with all orginating railroads in the United States to report all ship- ments of certain fresh fruits and vegetables moving in commercial wholesale channels. We can, therefore, assume that rail shipments reported in our summaries are 100% complete. Similar arrange- ments for reporting motortruck movements are not feasible. The Market News Service maintains seasonal shipping point offices in all principal producing areas during the main shipping seasons. Officers in charge report the truck shipments. There are two principal sources for these data. 1U. S. Dept. of Agriculture, Agriculture Market- ing Service, Fruit and Vegetable Division, Market News Branch, pp. cit. 27 The Federal-State Inspection Service reports the quantities of potatoes they have inspected which are scheduled for truck movement. The market re- porter contacts shippers direct for the quantities mOV1ng Via truck which are not inspected. We do, therefore, feel that we have a fairly high percentage completeness for motortruck movements. We are not in pos1tion, however, to affirm the percentage completeness of motortruck shipments for any State.1 Actually, four means of transportation are distinguished, namely rail, truck, truck-to-boat, and rail-truck (piggy back). Rail-truck shipments are re- ported only under rail and have been treated as rail in the present study. The distinction between rail and rail-truck need therefore not be upheld. Truck-to-boat shipments are in some cases included in truck shipments without identification. This is as it should be for the purpose they are used in the present work. In other cases they are reported under rail shipments, but then they are identified by "Boat" or "ET" after the name of the state. In the latter cases these shipments were separated from the rail shipments in the present work and added to the truck shipments. After this had 1Letter from J. M. Saylor, In Charge, Transporta- tion Reports, Fruit and Vegetable Division, Agricultural Marketing Service, U. 8. Dept. of Agriculture, Washing- ton, August 13, 1965. It should be noted that, according to this letter, "no detailed analyses of the shipments data are made by the Market News Service other than the introductory comments carried in the preface of each publication." Certain questions regarding the data can therefore not be answered in an exact manner. Mr. Saylor was asked to give the best information or judgement he could provide. 28 been done, only two gropus, in the following called "rail" and "truck", had to be dealt with. The impor- tant thing to notice is that in no case are rail-truck and truck-to—boat shipments reported more than once. Insofar as possible there are no duplications whatsoever} neither in rail, nor in truck shipments. Potatoes produced in one state and shipped from the' neighbour state are counted in total Shipments only once. Rail shipments represent only carlots moving on initial line-haul waybills. Truck shipments, as well as rail shipments, represent only domestic shipments during the whole period. Imports are reported separately. All the data, as they come from the United States Department of Agriculture, are expressed in car- lots, or for truck, actually carlot equivalents. The weights of the truck shipments are presumably known originally, and these shipments are converted to carlot equivalents by means of conversion factors expressing the number of pounds of potatoes necessary to make up one carlot equuivalent. A set of conversion factors for various groups of states and, in some cases, for dif- ferent parts of the year, were established from January 1, 1950. These were revised, effective January 1, 1960, and were revised again, effective January 1, 1966. The reason why such revisions are necessary is illustrated 29 in the following quotation: Conversion factors are based upon the ”most usual" rail loadings from principal producing areas in effect at the time the factors were established. Rail loadings increased during intervening years as larger rail cars became available and improved packaging and loading methods were devised. Another factor accounting for heavier loadings was estab- lishment of incentive rates made on graduated scales of multiple carload minimum weights, with the per cwt. rate being lower as the carload minimum weight increases. In effect, rail shipments are expressed in a unit that may vary all the time in any way consistent with loading practices. Truck shipments, on the other hand, are expres- sed in a unit that contains a constant number of pounds as long as a certain period, a certain group of states, and (in some cases) a certain part of the year is dealt with. When these specifications are relaxed, the magni- tude of the unit may change quite drastically.z The completeness of the data was mentioned earlier.3 This question has three important aspects: (1) Are any data reported at all for a certain state in a given year? (2) If data are reported, what percentage do the reported figures make up of total shipments? (3) Given that only a part of total shipments are in- lIbid. 2The conversion factors established in 1950, 1960 and 1966 are listed in Appendix C. 3Above, p. 2n. 30 cluded in the reported figures, do all categories Of shipments (shipments for fresh use, seed, processing, shipments from various seasonal crops, and so on) make up approximately the same percentage of reported ship- ments as of total shipments, or are certain kinds of shipments under—represented or not represented at all in reported shipments? The first question can always be answered, and the second one will be dealt with later. The third question is difficult to answer. The following quotation gives part of the answer: Rail and truck shipments include potatoes in- tended for fresh consumption, seed (these some- times are diverted to fresh consumption), and government purchases. Potatoes to chippers are also reported. No shipments to processors or for manufacture have been reported since 1955. Some minor omissions and irregularities in the shipments data and adjustments for these are dealt with in Appendix C. Unloads Data Truck unloads data were used in the present study 1Saylor, letter, August 13, 1965. Similar infor- mation has also been received in letter from Clarence E. White, Agricultural Statistician in Charge, Statistical Reporting Service, U. S. Dept. of Agriculture, Boise, Idaho, September 28, 1965. From this letter we quote the following: "All fresh shipments and all seeds are in- cluded. Apparently 'long haul' shipments to processors (mainly chippers) are included. About the only thing excluded is short haul rail movement to processors. One of the reasons all processor shipments are not excluded is inability to identify them." 31 as a substitute for truck shipments data in cases when the latter were not available. The unloads data that were used are of a similar nature as the shipments data, but they show quantities of potatoes unloaded in cer- tain selected cities by month and state of origin rather than quantities shipped from the same state. Since unloads data cover a limited selection of cities, they are in general less complete in coverage than the corre- sponding shipments data when the latter are available. A question also arises as to how representative they are for the shipment pattern of a given state. For- tunately, the number of included cities is quite large, and the cities are spread all over the United States. One may therefore expect the seasonal pattern of unloads in these cities to be similar to the pattern for total unloads, although there is a possibility, of course, that certain kinds of unloads, like unloads of seed, are not included to the same extent, relatively, in the unloads data as in total unloads. Another reason why unloads data may not be quite comparable with shipments data is the fact that transpor- tation takes time. A few shipments that take place, say, in the end of October are therefore reported as unloads in November. On the other hand, this is compensated for more or less in that some September shipments are un- loaded in October. Since truck transportation mostly is 32 used for relatively short distances, the time lag is usually only a few days, and the possible bias is probably not important. The unloads data were obtained in several different ways. Data for 1950—57 had to be taken from individual city reports and then added together for each state.1 For the years 1958-62 data were obtained directly from the Department of Agriculture.2 Data for 1963-66 were taken from four annual publications.3 In 1965, data for January through July, 1965 were extracted from an IBM run showing potato unloads by cities and states of origin.“ In 1967 when the time series were updated, data for January through July, 1967 1See U. S. Dept. of Agriculture, Agricultural Marketing Service, Fruit and Vegetable Division, 22’ troit: Unloads of Fresh Fruits and Vegetables, 1957, p. 23. Annual issues for 1950-57 were used. Correspon— ding publications were also used for the other cities (about 15—25 cities). 2Letter from J. M. Saylor, In Charge, Transpor- tations reports, Fruit and Vegetable Division, Agricul- tural Marketing Service, U. S. Dept. of Agriculture, Washington, August 25, 1965. 3See U. S. Dept. of Agriculture, Consumer and Marketing Service, Fruit and Vegetable Division, Market News Branch, Fresh Fruit and Vegetable Unload Totals fgr 31 Cities, Calendar Year 1966 (Washington: U. S. Government Printing Office, March, 1967), pp. 35-36. The annual issues for 1963—66 were used. “Saylor, letter October 6, 1965. 33 were again obtained directly from the Department of Agriculture.1 Price Data The price data that were used for potatoes in the present study are essentially monthly average prices received by farmers in the United States as a whole. These prices are reported by the United States Depart- ment of Agriculture. Prices for 1950-55 were obtained from a supplement to Agricultural Prices.2 Prices for 1956—58 were taken from monthly issues of Agricultural 3 Prices. For the years 1959-66, prices were obtained from the annual issues of Agricultural Prices.)4 The latest available revised figures were used in each case. The prices refer to all potatoes being sold in a given month, regardless of the year harvested.5 They lIdem, letter, November 9, 1967. 2U. 8. Dept. of Agriculture, Agriculture Marketing Service, Crop Reporting Board, Agricultural Prices, Janu- ary 1957, Supplement No. 2, Potatoes: Monthly and Season AVErage Prices ReceiVEH bi Farmers, by States and United States, l9A9-56, p. 10. 3See Agricultural Prices, February 1957, p. 19; January 1958, p. 20; and January 1959, p. 20. “See U. S. Dept. of Agriculture, Statistical Re- porting Service, Crop Reporting Board, Agricultural Prices: 1966 Annual Summary (Washington: U. S.-§overnment Print ting Office, June, 1967), pp. 22-23. The annual summaries for 1959-66 were used. 51bid. also refer to the first point of sale.1 Some further details regarding the price data and their comparability over time are given by Dalrymple.2 Farm-Retail Spreads Monthly data on farm—retail spreads for potatoes for the United States as a whole were received directly from the United States Department of Agriculture.3 These data are closely related to the difference be- tween retail prices for potatoes reported by the Department of Labor and prices received by farmers, A reported by the Department of Agriculture. Utilization Data Data on quantities (hundreweight) of potatoes processed from the crOps of each of the calendar years llbid., p.u 2Dalrymple, 9p. cit., pp. 18-19. T 3Letters from Forrest B. Scott, Leader, Marketing Resources and Cost Group, Marketing Economics Division, Economic Research Service, U. S. Dept. of Agriculture, Washington, July 1, 1965, and November 9, 1967. “For further details regarding the farm-retail spreads, see U. S. Dept. of Agriculture, Agricultural Marke- ting Service, Marketing Research Division, Farm-Retail Spreads for Food Products, Miscellaneous Publication No. 7A1 (Washington: U. S. Government Printing Office, November, 1957), pp. 1-95. 35 1956-66 were obtained from the Department of Agriculture.1 Similar data for the crops of the calendar years 1951-56 have been reported by the National§PotatoCouncil.2 These data were converted to hundredweight, assuming that one bushel = 0.6 hundredweight. Still, the data from the two sources did not agree for the overlapping year 1956. Basically, two time series on potato processing for the period 1951-66 were used in the present work. The first one was intended to express total quantity of pota- toes processed, except for starch and flour. The second; one was an expression for quantities of potatoes processed as chips and shoestrings. Both series were constructed by linking data from the two sources together and by using the actual data from the Department of Agriculture for the years 1957-66. Data on total processing for the years 1951-56 were obtained in the following way: For the years 1951-55 it was assumed that data on processing as ships from 1See U. S. Dept. of Agriculture, Statistical Re- porting Service, Crop Reporting Board, Irish Potatoes: Utilization g: 1966 Crop with Comparisons (Washington: September 7, 1967) p. 3. The annual publications for each of the crops of 1960-66 were used. 2National Potato Council, "U. S. Production, Utilization, and Use of Designated Potato Crop," National Potato News, Vol. IV (January, 1957), p. 11. 36 the National Potato Council and the figure for total processing from the Department of Agriculture was used. Data on processing for chips for the years 1951-56 were obtained by multiplying the series of data from the National Potato Council by a constant factor (less than one). This factor was determined in 'such a way that the product for 1956 was equal to the corresponding processing figure from the Department of Agriculture. Data on Diversion of Potatoes to Starch, Flour, and Livestock Feed Under Government Programs Weekly data on total quantities of potatoes diverted in the United States as a whole were received directly from the United States Department of Agricul- ture. Corresponding data, not by weeks but for the whole Fall crop or Late Summer and Fall crops combined, were obtained for each state from the same source.1 The weekly data were converted to monthly data. Data for weeks beginning in one month and ending in another were distributed on the two months in proportion to the number of its working days belonging to each month. 1Letters from E. J. Holcomb, Acting Chief, Vegetable Branch, Fruit and Vegetable Division, Consumer and Marketing Service, U. S. Dept. of Agriculture, Washington, September 9, 1965, and November 9, 1967. 37 Data on Production, Sales from Farms, and Related Magnitudes Several data series related to production were used in the present work. Some were included directly in the econometric analyses, while others provided informa- tion on trends, variabilities, and relative magnitudes of various variables, thus aiding the construction of rele- vant models. . Data on production; quantities sold from farms; acreages planted; stocks; and quantities used on farms where grown for (1) seed, (2) feed, shrinkage, and loss, and (3) household use, were obtained from publications from the Department of Agriculture.1 All these data, except the stocks data, are available by seasonal groups, and by states. The stocks data refer to total stocks of Fall potatoes held by growers and local dealers in the 26 1See U. S. Dept. of Agriculture, Agricultural Marketing Service, Crop Reporting Board, Potatoes: Estimates in Hundreweight by States 1866-1953, Statisti- cal Bulletifi No. 251 (Washihgton: U. S. Government Prin- ting Office, June, 1959), pp. 3-95. Data for 195u-59 were taken from U. S. Dept. of Agriculture, Statistical Reporting Service, Crop Reporting Board, Potatoes, Sweet- potatoes, b States and Seasonal Groups, Crops 9f 1955- 1959, Statistical Bulletin No. 291 (Washington: August, 1961), pp. 5-35. For the years 1960-6U, annual publica- tions were used as sources in 1965. See idem, Potatoes and Sweetpotatoes: Estimates b States and Seasonal Groups--Crop§ pf 1963 and 196A Washington: August, 1965), pp. 3:13. When the time series were updated in 1967, a corresponding annual publication was used to provide data for 1965 and 1966, while revised data for l959-6u were obtained from idem, Potatoes, Sweetpotatoes: By States and Seasonal Groups-—Crops pf 1959-196A, Statistical Bulletin No. A09 (Washington: U. S. Government Printing Office, July, 1967). 38 states where Fall potatoes are grown. These data are available for December 1, January 1, February 1, and March 1, respectively. Total stocks consist of produc- tion less total disappearance to date-l In 1965, indicated acreage and production figures were used instead of the unavailable actual figures for the Winter, Late Spring, and Early Summer crops of California that year. The Winter and Early Spring crops of Florida were treated the same way.2 Data on total sales were also needed and were estimated by assuming that the ratio of sales to production was the same as in the preceeding year. When the time series were updated in 1968 the unavailable sales data were ob- tained by a similar procedure. But this time actual pro- duction and acreage figures were used for the last year for which data were needed.3 The effects on the econo- metric results of the manipulations described in this para- graph are probably negligible. llbid., P. 3a. 2The data were taken from U. S. Dept. of Agricul- ture, Statistical Reporting Service, Crop Reporting Board, Crop Production, United States Crop Summary as 2£ October 1, 1965 (Washington: October 11, 1965), p. 53. 3The production and acreage figures were taken from idem, Crop Production, 1967 Annual Summary, Acreage, Yield, Production, by States (Washington: U. S. Govern- ment Printing Office, December 19, 1967), p. 95 and p. 98. 39 Certain reclassifications between the Late Summer and Fall crops have taken place during the period 1952-66, but these are of little significance for the present work. Space does not permit a treatment of the methods by which the data are collected and prepared.1 Data From Other Sources Exports and Imports by Months Data on total exports and total imports of potatoes by months were used to compute a series of net export figures. These data are reported by the Bureau of Census. 1These questions are dealt with in U. S. Dept. of Agriculture, Statistical Reporting Service, Statistical Reporting Service bf the b. S. Department 2£ Agriculture: Sbbpe, Methods, Miscellaneous Publication No. 967 (Washington: U. S. Government Printing Office, December, 196“), pp. 62-7“. See also Thirteenth National Potato Utilization Conference, Proceedings (Riverhead, New York, 1963), pp. 16-18. 5U. S. Dept. of Commerce, Bureau of the Census, U. S; Exports: Commodity by Country, December, 1966 TWashington: U. S. GovernmEnt PriRting—OffiEE, March 1967, p. “0. Corresponding variously denominated monthly publi- cations from the Bureau of the Census were used for the whole period 1951-66. For import data, see U. S. Dept. of Commerce, Bureau of the Census, United States Imports of Merchandise for Consumption: Commodity by Country bf “— Origin, January, 196A (Washington: U. S. Government Prin- ting Office, April, 1964) p. 16. Corresponding monthly publications were used for each month back to January 1951. The way of recording has been somewhat different since February, 196A. See U. S. Dept. of Commerce, Bureau of the Census, U. S. Imports of Merchandise for Consump- tion, December,—I966 (Washingtan: U. S. Government Printing Office, April, 1967), p. 22. Data for February, 196“ through October, 1966 were obtained from corresponding monthly publications. “0 Income and Population Data The income variable used was based upon disposable personal income. (Seasonally adjusted quarterly totals at annual rates).: The income data are reported by the Department of Commerce.1 As was the case for other data, the latest available figures were always used. The quarterly income data were divided by inter- polated population for the middle of the quarter.2 The interpolations were based upon population figures for 1See U. S. Dept. of Commerce, Office of Business Economics, Business Statistics, 1965, Biennal Edition. A Supplement bb_the Survey of Current Business (Washing- ton: U. S. Government PrinEing Office, August, 1965), p. 7. Earlier Editions were also used: 1955, p. A; 1957, p. A; 1959, p. A; 1961, p. 5; and 1963, p. 7. A source for the later part of the period was idem, Surve bf Current Business, Vol. IIIL (July, 19675, p. 8-2. See also the same issue, p. 9 and the July 196N issue, p. 11. 2Population data for 1952-6” were obtained from U. S. Dept. of Agriculture, Economic Research Service, Economic and Statistical Analysis Division, Food Consump- tion and Utilization Section, U. S. Food Consumption: Sources of Data and Trends, l909-bb (Washington: June, 19555, pT—187. The figure for January 1, 1965, was first taken from U. 8. Dept. of Commerce, Bureau of the Census, Current Population Reports: Population Estimates (Washington: U. S. Government Printing Office, July 16, 1965), p. 2. When the time series were updated in 1967, population figures for January 1 and July 1 of the years 1965-67 were taken from idem, Current Population Reports; Population Estimates (Washington: U. S. Govern- ment Prinfing Office, August 21, 1967), p. 2. U1 January 1 and July 1. Population referred to number eating out of civilian supplies. From 1960 on, the populations of Alaska and Hawaii were included.1 Rail Conversion Factors Quarterly data from the railroads on revenue freight of potatoes originated, by eight regions, expressed in both tons and carloads, were used to derive quarterly rail conversion factors by regions. The data have been published by the Interstate Commerce Commission.2 Interpolations or extrapolations were used for certain quarters since published data were not avail- able. These quarters were: Third and fourth quarters 1951; second and third quarters, 1953; the whole year 195“; and the whole years 1964 and 1965.3 The quarterly figures are subject to corrections of errors in carriers 1Since 1960, national income is defined to in- clude Alaska and Hawaii. For further details see Survey bf Current Business, July 1961, p. 5. The inclusion of Alaska and Hawaii in the population series from the same year on seemed to give the most comparable per capita in- come series over time. These states are not included in the other time series used in this work, however. 2See Interstate Commerce Commission, Bureau of Transport Economics and Statistics, Freight Commodity Statistics of Class I Railroads 12 the United States, Tons bf Revenue Freight Carried and Freight Revenue bf Large Class I Railroads, Fourth Quarter, 1963, pp.?H-22. Corre- spondifig issues were used for other quarters in the period 1950-63. Since 1963, no quarterly publications of this kind have been issued. 3When the time series were updated in 1967, other conversion factors were used for the last part of the period. This is described in Appendix C. A2 reports under correspondence (revised figures are used in annual summary), but this could not be taken into account. Various Price Index Numbers from the United States Department of Labor Six different series of price index numbers from the Department of Labor were used in the present study, namely the Consumer Price Indices for (1) all items; (2) total food at home; (3) cereals and bakery products; (A) meats, poultry and fish; (5) fruits and vegetables; and finally the Index of Wholesale Prices for all commodities except farm products and foods.1 Index numbers with base, 1957-59 = 100 were used for the whole period. Since the published indices for the earlier part of the period did not have this base, they were linked to the corresponding indices with base 1957-59 = 100. The Consumer Price Index for all items was used as a general deflator for all other price index numbers, and also for prices of potatoes and for per capita income. An average index for each quarter was used in deflating income. 1See U. 8. Dept. of Labor, Bureau of Labor Statis- tics, Monthly Labor Review, Vol XC (August, 1966), pp. 117 and 120. Data for the whole period 1951-66 were obtained from various monthly issues. CHAPTER IV ESTIMATION OF MONTHLY SALES DATA ON THE BASIS OF SEASONAL SALES -DATA AND MONTHLY SHIPMENTS DATA 8 Introduction The published monthly shipments data are more or less incomplete, and they are expressed in units that may vary drastically over the years, between seasons, and geographically. Time series of United States totals arrived at by simple additions of these data over states could therefore not be used in the monthly supply and demand analyses of the present work. Instead, monthly sales data were constructed in a different way. Basic- ally the approach taken was to distribute on months the data on sales from farms that are available for seasonal groups. The shipments data served essentially as distri- butive weights. Although this approach appears simple, a large number of problems were involved. Among these were the following: (1) Except for states where only one seasonal category of potatoes is grown (twenty-six states), there "3 AA is in general no way of knowing what percentage of the potatoes shipped in certain months originate from various seasonal crops. (2) Should both rail and truck data be used as distributive weights, and if so, how much mutual weight should be given to each of the two series? (3) How should states and years for which few or no ship- ments data are available be handled? (A) Should all or only part of total sales from farms be distributed on months according to the pattern of shipments? (5) How could unloads data be utilized to make the data more complete? Many decisions had to be made where subjective judgement was unavoidable. In order to secure similar treatment of similar cases and to make it possible for other research workers to check the procedure followed, certain guiding principles that seemed reasonable were established. Space does not allow a discussion of the various alternative approaches that might have been followed. Neither is it possible here to give the reasons for all the decisions that were made. The present chapter con- tains mainly a description of the procedure followed. Description of Method The estimation of monthly sales data was carried out for individual states in most cases. Only after this 45 had been done were the state data added up to United States totals. The monthly data for a few relatively unimportant states were very poor. These states were not treated individually. Instead they were pooled together with other states having similar production conditions.1 The problem of overlapping of marketings from different seasonal groups in certain months was dealt with by pooling together sales data for all seasonal groups of potatoes grown in each state.2 Thus yearly rather than seasonal sales data were used. The crops were not always added up for calendar years, however. For each state the year was defined in such a way that all or most of the potatoes shipped during any single year could be assumed to have originated from the crop of the same year. In many states there are some months when no potatoes are shipped. For such states the choice of definition of the year was easy to make. In other states potatoes are shipped in every month of the year. In a few such cases, and especially for California, the defini- 3 tion was of necessity somewhat arbitrary. 1For details, see Appendix C. Table 18 gives the state groupings. 21f total reported shipments make up approximately the same percentage of total actual shipments in every month of the marketing period, this method will give a satisfac- tory result. 3The year used in each case is indicated in the second column of Table 18 in Appendix C. Calendar year was used whenever satisfactory, since this facilitated the treatment of the unloads data. A6 Except for the information that can be derived from the shipments data themselves, little is published on the relative importance of rail and truck transporta- tion in various states or for various categories of potatoes.1 It was therefore decided to use both truck and rail data and to give each of the means of transpor- tation weights according to their importance in the re- corded data.2 In cases when truck shipments data were not re- corded, or when these apparently were incomplete for part of the year, truck unloads data were used as a sub- stitute if they were available.3 The cities for which un- loads data were looked up were the same for all states and for all months of any year, but in order to utilize as much as possible of the available information, they were allowed to vary from year to year.“ In general, there was 1For a treatment of this question, see Ivon W. Ulrey, Fresh Potato Transportation to Larg_ Markets from Five Major Producing Areas, Marketing Research Report No. 687, Marketing Economics Division, Economic Research Ser- vice, U. S. Dept. of Agriculture (Washington: November, 196A), pp. 1-31. 2I.e. the rail and truck data were simply added after having been converted to a common unit. 3The term "year" is used in the rest of this chapter to designate a year as defined for the state or states in question. ”Even though the reports for the same cities were used in all months and for all states, this does not necessarily mean that data were actually recorded for all months and all states. A7 an increase over time in the number of cities included. In 1957/58, the number of cities could not be kept con- stant because totals rather than individual city figures were used for 1958. Certain adjustment then had to be made.1 Some minor adjustments of the movement data were made for other reasons: mainly in order to correct for irregularities in the reporting of the data or to provide some convenience in the handling of the data. In a few relatively unimportant cases it was also necessary to use data for other years or for other similar states since no data were available for the years in question.2 When the various adjustments and substitutions had been carried out, the truck data were multiplied by the official truck con- version factors, and the rail data were multiplied by quarterly rail conversion factors derived from railroad 3 statistics, and the two series were added. The next decision to be made was whether all or 1For further details regarding the use of the un- loads data, see Appendix C. 2The adjustments, etc. mentioned in this paragraph are further described in Appendix C. 3The truck conversion factors are listed in Appen- dix C. The rail conversion factors are described in Chapter III and in Appendix C. According to Saylor, letter, August 13, 1965, use of conversion factors derived from railroad statistics give the most reliable tonnage figures for rail movements for each year. A8 only some sales should be distributed on months accor- ding to the shipments pattern. Potatoes diverted to livestock feed are to a large extent used on farms where grown or on nearby farms. In most cases, these potatoes 1 Also are probably not included in the shipments data. diversions to starch and flour are probably excluded from the shipments data in most cases. Shipments for manufacture have not been reported since 1955.2 Before 1955 diversion operations were rather limited,3 and it is likely that potatoes for starch and flour were to a large extent excluded from shipments in that period too.“ Altogether it was believed that the best result would be obtained by assuming that potatoes for diversion purposes were excluded from the shipments data during the whole period. Total yearly diversions were therefore subtract- ed from the yearly sales data, and the remaining sales were distributed on months according to the shipments 1This argument is also used in James Harold Cothern, "The Importance and Impact of the 1955 and 1956 Government Potato Diversion Program on the Potato Indu- stry" (unpublished Master's dissertation, Dept. of Agri- cultural Economics, Michigan State University, 1957), p. 50. 2Above, p. 30. 3See table 7, Appendix C. "Up to 1955, footnotes to the tables on shipments show yearly rail shipments for manufacture for Idaho, Maine, and a couple of other states. “9 pattern.1 When monthly data on sales including diversion were needed in the econometric analyses, the monthly di- version data were added to the estimated monthly data on other sales. Since appropriate data were lacking, it was not found practical to apply similar procedures for other categories of sales. In effect then, all sales except diversion were assumed to be distributed on months propor- tionally to total shipments. Evaluation bf the Estimated Data It has been shown elsewhere in this work that the shipments data have certain deficiencies. .Accordingly, one might argue that the reliability of the estimated monthly sales data is somewhat doubtful. Various ad- justments and substitutions were made to improve the data, however. For instance, in some cases when the shipments data cover only a part of the season because the shipping point offices have been closed during the rest of the season, unloads data have been used. 1This procedure also assumes that potatoes diverted to livestock feed on farms where grown are regarded as sold for statistical purposes. An inquiry sheet sent from the U. S. Dept. of Agriculture to farmers in 196A indicates clearly that this was the case for that year, and it was therefore assumed that it was true for the whole period. See U. 8. Dept. of Agriculture, Statistical Reporting Service, Disposition b£ 1963 Potato Crop and Acreage for 196A (June:I96A). 50 In some instances the manipulations that have been undertaken may seem drastic. In evaluating these cases it should always be kept in mind that the ship- ments data have been used only as distributive weights. Their relative distribution on months rather than their absolute magnitude is therefore the thing that counts. The principle followed was to estimate monthly sales data for all states, even if doubtful weights had to be used for a few states. Since the final figures of interest were U. S. totals, this procedure was deemed more satisfactory than to disregard these states comple- tely. The yearly sales data for these states are pro- bably about as good as for other states. In any case these states account for only a very small percentage of total potato sales. The three most important potato producing states, Idaho, Maine, and California, are covered with seemingly very good rail and truck shipments data every year. These three states accounted for A2.2% of U. S. potato production in the years 1958-62. Quite satisfactory movement data are also available for several other im- portant states such as North Dakota, Minnesota, Washing- ton, Colorado, Wisconsin, Oregon, and Michigan. The ten states mentioned made up for 71.A% of U. S. potato pro- duction in 1958-62. Even several of the less important states are covered with satisfactory movement data. 51‘ The states for which the data for one or more years were really deficient, namely Arkansas, Georgia, Illinois, Iowa, Kentucky, Mississippi, New Hampshire, New Mexico, Oklahoma, Rhode Island, vermont, and West Virginia, accounted for only 2.2% of total U. S. potato production in 1952-62. The estimated monthly sales figures for the U. S. as a whole are shown in table 7 in the beginning of Appendix B. A question one might raise is the following: What percentage of these estimated sales were actually recorded as shipments by the U. S. Department of Agri- culture? This question is partly answered in table 18 in the end of Appendix C. The figure shown in the table for each state and year expresses the yearly sum of the distributive weights as a percentage of the correspon- ding total recorded sales, excluding diversion. As has been described elsewhere, the distributive weights in- clude rail shipments and in addition either truck ship- ments or truck unloads. In most cases the percentages therefore give a pretty good picture of the yearly coverage of the movement data. In some special cases the distributive weights have been constructed in a somewhat artificial way by using data for other states or other years, etc. In such cases the percentages are of little value, and they may even exceed 100. The pre- 52 ceding section of Appendix C should therefore be consul- ted when table 18 is being studied. For example, the percentage of 1A0 for Oklahoma in 1966 is probably due to the fact that the distributive weights were construc- ted by adding the movement data for Arkansas to the Oklahoma data. The reason why some percentages for Alabama and Arizona exceed 100 may be inappropriate rail conversion factors. Loading practices may vary somewhat within regions. If the rail conversion factors that were used are proportional to the actual ones, the estimated sales will not be affected significantly by this error, however. A crude check of the overall coverage of the shipments data can be made in the following way: In 1963 rail and truck shipments together for the U. S. as a whole amounted to about 250,000 cars. Assuming the weight of each car to be A5,000 pounds, this equals 112,5 million hundreweights. Since total sales that year amounted to about 2A5 million hundreweights, the coverage was about A6%. The actual coverage is con- siderably higher, since we have not included in the 250,000 cars the truck unloads data that have been utilized. The estimated sales data are not equally re- liable for all years. The data for 1957/58, for example, were adjusted in several cases. Most of these 53 adjustments affected only the truck unloads data, how- ever. The number of cities covered by the truck un— loads data has been increased over the years, and the re- liability of the data has therefore increased also. To some extent a similar development has taken place for the shipments data. Since crop year was used in several cases, some data for 1951 and 1967 had to be used even though the econometric analysis cover only the period 1952-66. Some of the data used for 1967 were preliminary. The data for the last part of 1966 is to some extent influen- ced by this, but the effect is probably small. In conclusion, we must admit that the estimated sales data have certain deficiencies. However, the data are believed reliable enough to be capable of producing several meaningful econometric results. The data play a crucial role in the type of models explored here. If such models prove to be of considerable value, there are several ways of improving the data in the future.1 1Some suggestions for improving the data are set forth in the last section of Chapter IX. CHAPTER V THE SUPPLY OF POTATOES DURING THE YEAR As mentioned previously, the concept of supply employed in the present study is supply from farms at the farm gate, so to say. In order to understand which factors affect supply during a certain period, it is necessary to rea- liZe what alternatives the farmers have to selling the potatoes in the market during that period. In the following, we shall discuss these alter- natives with a view to sorting out variables that may be useful as "supply shifters" in the estimated supply func- tions. In the course of doing this, we shall also discuss whether the variables in question may be considered ex- ogenous or must be treated as endogenous variables. As a guiding framework for our discussion, we_ shall assume that the farmers act as profit-maximizing economic units. 1See above, p. 8. 5A 55 The Question whether Potato Yields and Potato Production are Endogenous A natural starting point in discussing supply of potatoes is the acreage planted of each seasonal crop. If we make the very reasonable assumption that the seasonal classification of any field of potatoes is predetermined at the time of harvesting and marketing, it follows that the planted acreage of any seasonal category of potatoes is predetermined during the market- ing period of that crop. What we are interested in, however, is whether also production can be considered predetermined during the marketing period. There is a possibility that farmers may leave some acreages of potatoes unharvested for economic reasons. If we use production figures based on total acreage (as has been done in this study) rather than on harvested acreage, we do not have to worry about this possibility, however. Another possibility that must be considered is that farmers, during the marketing period of a crop, may influence the yield of that crop by better growing prac- tices, in response to higher prices. The present writer is of the opinion that this possibility is not important enough to take into account when we deal with prices for periods as short as months. Still another possibility of interest is that 56 farmers may be in a position to influence yields by vary- ing the time of harvest. For certain crops that usually are harvested before maturity, such variations no doubt take place in many cases. Again, since we deal with prices for periods as short as months, yields of the crops harvested during these periods may, as an approx- imation, be assumed to be unrelated to prices during these periods. Yields are namely influenced by several other factors, among which we also find prices in other months. It is also somewhat doubtful whether the record- ed yields and production figures actually reflect vari- ations in yields due to variations in harvesting time. The published figures may more nearly reflect some kind of potential yields that would have materialized under the assumption of a "normal" or "average" harvesting time. In summary, we may assume that the recorded total production of any crop of potatoes is unrelated to the price of potatoes in any single month during which the same crop is marketed. Whereas this is assumed for production, it is important to note that the marketed quantity of any crop during any month of the marketing period of that crop may very well be influenced by the potato price that month. This is true both for storage potatoes and other potatoes since larger or smaller parts of the available acreages of the latter may be 57 harvested during any month of their potential harvest- ing period, depending on prices. The Effects bf Production bb Marketings We have now arrived at the conclusion that during each month of the year there is available on farms a fixed quantity of potatoes that potentially may be marketed. Other things equal, one would usually expect more potatoes to be marketed when this quantity is large than when it is small. The available quantity that potentially may be marketed is related to the pro- duction estimates for all crops that may be marketed during the month in question. Our discussion thus provides a rationale for including total production of certain seasonal categories of potatoes as variables in the supply function for potatoes each month. For the Fall crop, estimates of total stocks may also be used as a substitute for production estimates. The stocks data have the handicap, however, that they also include stocks held by local dealers. If we can assume that farmers hold a constant percentage of these stocks, the data can still be used.1 'lIn Chapter VI it is shown that the estimated sales data can be interpreted in two alternative ways. According to one interpretation, the estimated sales data are regarded as sales not only from farmers, but also from other shippers. If this interpretation is adOpted, the fact that stocks also includes stocks held by local dealers ceases to be a problem. 58 In most cases one would expect large production figures to be associated with large marketings, but there are several exceptions to this. It should namely be kept in mind that the estimated monthly data on marketings in- clude marketings from more than one seasonal crop. As an example, let us consider marketings in January. One would expect a large Fall crop to be associated with large marketings in January. But will also a large Winter crop be associated with large marketings this month? Not necessarily. The Winter crop is planted from August through the first half of January. During that period the size of the current Fall crop is pretty well known. If a large production of Fall potatoes is anticipated, growers will possibly plant less Winter potatoes than usual. Since marketings in January are dominated by the Fall crop, marketings will still be high, and as a result a small Winter crop will be as- sociated with large marketings in January. This hypothesis implies a negative correlation between Fall production and Winter production. In the period 1952-6A the correlation coefficient between the two variables was 0.02, however. We may therefore conclude that the above hypothesis is of more theoreti- cal than practical interest. This may be due to a scar- city of alternatives to Winter potato growing. Another hypothesis implying a negative relation- 59 ship between, say, Winter production and January market- ing should also be considered. According to this hypothe- sis the Fall and Winter crops play roles that are the opposite of what was the case above. Fall potatoes are marketed from the time when the Winter potatoes are planted till long after the last Winter potatoes have been marketed. Producers of Fall potatoes are therefore in a position to follow the develop- ment of the Winter crop and to take this development into account in their marketing plans. It is reasonable, then, to expect that when a large Winter crop is expected, pro- ducers of Fall potatoes will tend to limit their marketings during the marketing period of the Winter crop, and rather increase marketings before January and after March. True enough, a large Winter crop will lead to increased market- ings of Winter potatoes, but when producers of the much larger Fall crop restrict their marketings, an "overcompen- sation" may very easy occur so that a large Winter crop will be associated with small total marketings in January, February, and March. The discussion above pertains to the Fall and Winter crops, but it is clear enough that similar relation- ships also exist between other crops. We may therefore conclude that the signs of the coefficients of the produc- tion variables in the supply functions may be either posi- tive or negative. What is the actual sign is an empirical question. 60 So far, we have shown that an approximately fixed quantity of potatoes is available for possible marketing each month and that total production of the various seasonal categories of potatoes are among the variables determining actual marketings. In order to determine which other variables should be included in the supply relations, we must consider what alternatives the farmers have, other than marketing the potatoes dur- ing the current month. We will first consider potatoes that have not yet been harvested. Later we shall con- sider potatoes in stock. The Determination 93 the Time b: Harvest and Marketing Potatoes to be Marketed Shortly after Harvest Usually, potatoes are harvested when they are mature or when the potato field has tobe cleared either because the field is to be used for another crop, or be- cause cold weather or snow is expected. There are situ- ations, however, when harvesting before maturity is con- templated and when such factors as price and yield ex~ pectations are decisive. In some instances determination of the appropriate harvesting time is of considerable economic importance to the grower. The decision may also be very difficult because prices, as well as yields and quality, may change rapidly. The following theoretical analysis is intended 61 to illustrate the principles that may be operating when a profitemaximizing grower determines when to harvest a potato crop. Even though it is impossible to follow the grower's decision-making process in detail, it may be possible to set up a theoretical model that fairly well accounts for the principles underlying his decision. Let us consider an arbitrary field of potatoes that is of such a homogenity, magnitude, etc. that it will be harvested as a unit. At the time when the potatoes have reached a stage of development at which the possibility of harvest- ing them comes to the grower's mind (in the following called "today") we suppose that he has estimated several functions of time each covering the interval from the present till the point in time when the potatoes obvious- ly have to be harvested if losses are to be avoided. The dependent variables in these functions are the following: (1) Expected yields of each quality or grade of potatoes from the field in question. (2) The grower's price ex- pectations for the corresponding qualities and grades of potatoes. (3) The expected net increase (or decrease) in total costs attributable to the potatoes from this field from now on, if they are to be harvested on some future day rather than today. It does not matter how the grower obtains all this information. Of course, he will use his experience 62 and notes from earlier years, all kinds of information available from experiment stations, and his own judge- ment. He will probably take yield samples from his field at various points in time. He will certainly also have to take into account the opportunity costs of his labor force, transportation equipment, etc. in other enter- prises. On the basis of the functions listed above, the grower can compute a new function showing the expected net increase or net decrease in profit that he will experience by delaying the harvesting to some future date. This net increase is expressed as a function of all potential future harvesting dates (including today). If the last mentioned function has a maximum to- day, the grower will decide to harvest the potatoes to- day; if not, he will blgb to postpone the harvesting till th day when this function has a maximum. As time (p goes on, however, the grower may obtain various new info- rmation that leads him to revise one or more of the estimated functions underlying the profit increase function mentioned above. Accordingly, the planned harvesting date may also be changed. Expectations may even change in such a way that the grower some day finds himself in a situation where he concludes that profit would have been maximized if he had harvested the pota- toes in the past. Rapid price declines or unnormal 63 weather conditions are probably the main reasons why he might have such an experience. To make the picture com- plete, however, we must add that the grower may also ex- perience larger profit increases by postponing the har- vesting time than he had originally expected. iThe possibilities mentioned in the last part of the preceeding paragraph show that the analysis so far is not quite adequate except for growers who look only at the expected value of the profit increase function above and do not pay any attention to its variance. One possible way of modifying the analysis for growers with a different attitude towards uncertainty would be to apply a set of discount factors in order to convert each of the future values of the profit increase func- tion to values comparable with certain profit today. Storage Potatoes The problem that has been discussed above is in principle identical with the problem of optimizing the marketing rate of potatoes from storage for a single producer. Potatoes in storage usually deteriorate both in quantity and quality as time passes away, but at the same time prices are often increasing. We may therefore assume that the point in time at which a given lot of storage potatoes is sold is determined in a way similar to the point in time at which potatoes to be sold direct- a. 1y from a given field are harvested. A more detailed discussion of the economy of potato starge will be found in Zusman's work.1 We see no reason to repeat his discussion here. Conclusion Our treatment has revealed that expectations play an important role in the supply of potatoes both from the field and from storage. Unfortunately, we have no appropriate variables to represent the expectations themselves. All we can do, at best, is to include in the supply functions variables on which the expecta- tions are based. Clearly enough, only a very tiny selection of the relevant variables can for practical reasons be included. The price of potatoes in the curr- ent period is one such variable that probably has some effect on expectations. One may for instance expect that when prices are high, ceteris paribus, growers do not expect further significant price increases and vice versa. In addition, prices of potatoes must also be considered b priori to have a direct influence on market- ings. The price coefficient of an estimated supply rela- tion therefore measures the combined effect of expecta- tions associated with price, and the direct price lZuzman, bb. EiE': pp. 576-79. 65 response. Another variable that was used in the present in- vestigation as a possible variable associated with expec- tations is the price increase for the relevant crop during the preceeding season. This price increase is the grower's immediate experience regarding price increases for the crop in question and one might assume that this experience forms a basis for his expectations for the current season. In the following this price increase variable will be labeled the lagged price variable. The value of the lagged price variable for a certain month, m, in an arbitrary year, t, was formed as the price in- crease from month m in year t-l till a particular month m' later in the same marketing season. The month m' may be- long to year t-l or to year t. The price increases were computed on the basis of deflated average prices paid to farmers for potatoes. The months, m' corresponding to the various months, m are listed below. For January, February, March, and April m' was chosen to be May of the year t-l. For May, June, July, and August, m' was chosen to be June, July, August, and September, respectively. The latter months all refer to year t-1. For September-December m' was chosen to be May of year t. The various factors associated with yield expec- tations and cost expectations were not taken into 66 account in the empirical analyses of this work. Varia- tions in yields due to variations in harvesting time are accounted for indirectly insofar as prices are assumed to affect marketing rates which in turn are assumed to affect individual yields. The effects of interest rates on marketings were not taken into account since it was believed that inter- est rates have varied too little during the sampling peri- od to warrant their separate inclusion in the supply func- tions. It seems reasonable to expect that there is some dependence between marketings in various months. If the quantity of potatoes marketed early in a season is unusu- ally large, one might expect, ceteris paribus, that quantity of potatoes marketed later in the season will be unusually small. A variable, labeled as a lagged supply variable, was designed to allow for such effects. The lagged supply variable for a particular month was con- structed as the sum of the supplies in certain earlier months up to the month in question. In the following the additions will be described for each month of the year. The lagged supply variables for January-June were formed by adding up supplies for the preceeding months, starting with August and ending with December-May, re- spectively. For July and August the additions started with May and ended with June and July, respective- 67 1y. The lagged supply variables for September-December were constructed by adding the earlier supplies, star- ting with June and ending with August-November, respecti- vely. The Influence b: Home Use for Food, Seed; Feed, Shrinkage, and Loss, and Their Variations In addition to the possibility of selling the potatoes early or late, growers also have the alternative of not selling the potatoes at all. First of all they may leave potatoes unharvested. This may occur when prices are low compared to harvesting costs. Among the relevant variables determining the quantity of potatoes left unharvested are the price of potatoes, the size of the crop, and the harvesting costs. Our approach is to let the supply of potatoes be a funct- ion of (among other variables) the price of potatoes and of total production, including production left unharvest- ed.l Thus the only important variable not included in the analysis is harvesting costs, but these costs do not vary violently from year to year. The trend in this variable may be assumed to be taken into account by the trend variable. 1In the final analyses after the time series had been updated, total production less home use for food was used instead of total production as a product- ion variable. (See later). 68 Shrinkage and loss are probably related to prices of potatoes, to the size of the crop, and possib- ly to time. The potatoes are likely to have better care when prices are high. When the crop is small, only the best storage space and the best facilities for harvest- ing, transportation, etc. need to be used. Shrinkage can then be kept at.a minimum. Over time the methods and facilities utilized in handling potatoes may have improv- ed. Again, since our approach is to let supply be a function of total production, price, and a time variable, shrinkage and loss are automatically accounted for inso- far as our model is correct. The quantity of potatoes fed on farms is not recorded separately, but is lumped together with shrink- age and loss in the statistical publications. The quan- tity fed is probably in some sense, and to some extent, a residual. Except when feeding of potatoes is planned in an irreversible way or when the potatoes fed are of inferior quality, feeding occurs when the profit from feeding potatoes exceeds the profit from selling potato- es for other uses. Accordingly, one would expect the quantity fed on farms to be a function of such variables as market prices of potatoes, prices of other concen- trated carbohydrate feeds, and possibly of an index of the number of certain kinds of livestock animals in the potato growing regions. Since feeding of potatoes is 69 actually not very important in most of the U. S., and since the approach outlined involves several problems, only the price of potatoes and a trend variable have been included as variables describing feeding of pota- toes in the present analysis. Consumption of potatoes in households on farms was regarded as predetermined in this analysis. The writer believes that potato growers' households usually consume the quantities of potatoes they like to consume regardless of potato prices in particular, and also in- dependent of other prices, incomes, yields, etc. The possible gains from consuming less are very small. Pota- toes are a very cheap food at the farm level, and commer- cial potato growers are generally not poor. The consequence of this is that these potatoes ought to be subtracted from total production in forming the production variable. Since it was assumed that the change in consumption on farms can be approximately de- scribed by a linear trend variable, this was not actual- ly done, in the initial analyses. It was done in the final analyses after the time series had been updated, however. The use of potatoes for seed on farms where grown is probably to a large extent determined by the same factors as the use of seed in general.1 The distri- 1For a Discussion of the demand for seed potatoes, see Chapter VI. 7O bution of total seed requirements on home produced and purchased seed is partly determined by habit and is also subject to gradual change over time. The price of pota- toes and a trend variable should probably take care of a substantial part of the variation in home use for seed. In addition an acreage variable to be described in Chap- ter VI may be used as an explanatory variable for cert- ain months. Relative to total seed consumption for the crop in question, home use for seed plays a very modest role for all crops except the Fall crop. Foreign Supply bf Potatoes The foreign trade in potatoes is small and rela- tively unimportant. Essentially it amounts to some border trade and seed imports from Canada. Even though it actu- ally represents an oversimplification, net exports (or imports) of potatoes were treated as a predetermined variable in the present analysis. Stocks There exists a separate supply of potatoes from stocks, as well as a special demand for potatoes for storage. The supply of potatoes from stocks has been discussed already, and the demand will be treated later. CHAPTER VI THE OFF FARM DEMAND FOR POTATOES DURING THE YEAR Demand for Seed Potatoes The quantity of potatoes utilized for seed depends on the planted potato acreage and the quantity of seed potatoes used per acre. The yearly variations in planted acreage depend on several factors among which prices of potatoes may be of considerable importance.1 The relationship is such that high potato prices may in- duce increased plantings and increased demand for seed potatoes. The quantity of seed potatoes used per acre for a certain crop has to be within certain limits for biologic reasons, but there is also some room for vari- ations in seeding rates in response to variations in potato prices. Smaller or bigger seed potatoes may be used, and the seed may also be cut or not. For certain seasonal crops there has been a gradual increase in the use of purchased seed due to in- 1According to Simmons, bb. bib., p. 56, "actual planted acreage in the late crop States corresponds closely with acreage estimated on the basis of previous year's price". 71 72 creased use of certified seed and other changes in grow- ing practices. Almost all seed potatoes that are sold come from the Fall crop. For each seasonal crop of potatoes the total quantity of potatoes used for seed and the quantity used on farms where grown is known.1 When we deal with month- ly models, the matter is complicated by the fact that we do not know at which rates the seed is shipped through- out the marketing season for each crop. The shipment pattern is probably to some extent fixed from year to year, but there are certainly also irregular variations in seed shipments. On the basis of information regarding the usual planting time for the various crops, a variable was con- structed to express the demand for seed potatoes in vari- ous months.2 The variable was constructed as the sum of the acreages of certain seasonal crops, the plantings of which were assumed to take place partly during or shortly after the month in question. The crops whose acreages were added to form an acreage variable are listed below by months. For January and February: Early Spring. Late 1U. 8. Dept. of Agriculture, Agricultural Market- ing Service, Crop Reporting Board, bb. cit. '3 “Planting dates are listed in Royston, Frost, and Galloway, bb. Elfi- 73 Spring, and Early Summer. For March and April: Late Spring, Early Summer, Late Summer, and Fall. For May and June: Early Summer, Late Summer, and Fall. For July: Late Summer and Winter the following year. For August, September, and October: Winter the following year. For November: Winter the following year and Early Spring the following year. For December: Winter, Early Spring, and Late Spring, all crops being the crops of the following year. The actual shipment pattern of potatoes for seed is also reflected more or less through two other monthly sets of data that are available, namely imports of certified seed potatoes (mainly from Canada), and shipments of certified seed from Maine. These data are shown in Appendix B, Tables 13 and 1A. Neither of these series are representative of total seed shipments, but they might reflect pretty well the shipment pattern for Fall potatoes. This is especially true for shipments from Maine, which is a very important Fall producing state. Imports from Canada may be somewhat misleading as an indicator of seed shipments, because potatoes im- ported as seed sometimes are used for other purposes. When the data series had been updated in 1967, a variable intended to explain demand for seed was con- structed on the basis of Maine certified seed shipments in the following way: For each month a variable was 7A formed by taking actual shipments as a percentage of total shipments during the crop year July-June. The primary data that were used are shown in Appendix B, Table 13.1 Demand for Feed Potatoes In the models of the present work, potatoes fed on farms where grown have the effect of decreasing supply, while potatoes fed on other farms are taken in- to account as a category of demand. Potatoes are used for feed mostly in the West. As mentioned when home use for feed was discussed, the variations in the demand for potatoes for feed are pro- bably related to variations in prices of potatoes and prices of other concentrated carbohydrate feeds. Also an index of the number of certain kinds of livestock animals in certain regions might be used to explain feed demand. There is certainly a limit to the quantity of potatoes that can be profitably fed to livestock. Pota- toes for feed are also cheap and bulky and thus unable to bear high transportation costs. For these reasons 1In Table 13 data for certain months such as May and June are not recorded separately. The combined data were treated as if they pertain to a single month, and the percentages obtained were used directly as common percentages for the months in question. For the months January and May-June data were lacking for the year 1952. The aritmetic means of the corresponding percentages for 1953 and 195A were used to replace the lacking data. 75 the number of livestock in the growing areas may be of some importance. Most likely, however, the relevant in- dex number does not vary very much, and besides, the possible effects of such variations are probably over- shadowed by other influences. In the estimated models it was assumed that the price of potatoes, and possibly also a trend variable, are the only variables related to feed that can be pro- fitably included. The variations in prices of feed con- centrates were supposed to be too small to influence the total demand for potatoes appreciably. Demand for Potatoes Fresh for Food This demand can be divided into two categories: potatoes demanded by the growers' own households and off farm demand. As mentioned earlier, demand on farms was regarded as being predetermined in the final models. The off farm demand for potatoes is derived from consumer demand. Variables that determine consumer demand are therefore relevant in the demand equations. In addition, certain variables pertaining to the opera- tion of the marketing system may need to be included in order to explain demand at the farm level. According to economic theory, one would expect the demand for potatoes to be related to such variables as prices of potatoes, prices of various substitutes for 76 potatoes, and consumers' incomes. Some of the most important substitutes for pota- toes are probably macaroni, spaghetti, noodles, rice, and various vegetables, notably the kinds suggested in an earlier publication.; Since appropriate monthly data on prices of sub- stitutes for potatoes were not available, it was decided to use various index numbers from the Bureau of Labor Statistics. But these indices are very crude approxima- tions to the variables of real interest. The Consumer Price Indices for cereals and bakery products and for fruits and vegetables were used to represent the substi- tutes already mentioned. In addition the Consumer Price Indices for meats, poultry and fish were used. As an alternative to all of these price indices the Consumer Price Index for total food at home was used in some ana- lyses. . Because of the very limited number of observa- tions it was impracticable to include a whole series of different potato prices in our model. The main price variable that was used for potatoes is a weighted average of prices received by farmers (from Agricultural Prices). As already mentioned, the price variables used for other 1See U S. Dept. of Agriculture 'Potato Prefer- ences Among Household Consumers, (Miscel ahefifig'PUEIIEE- tion Number 667 (Washington: August, 19A8), pp. 18- -2l). 77 commodities were price indices of retail prices. In order to take the costs of marketing pota- toes into account, farm-retail spreads were included as a variable in the demand equations. Farm-retail spreads are, however, very complex magnitudes which depend on profits as well as costs. The sign to be expected for the coefficient of the farm-retail spread variable is therefore somewhat ambiguous. _If, say, farm-retail spreads are high because the profits earned by marketing firms from selling potatoes are high, one would expect retailers and dealers to do more to sell potatoes in competition with other goods. If such situations are prevalent, the sign of the coefficient may be positive. On the other hand, if spreads are high because costs (other than profits) are high, prices of potatoes will be high without this being compensated by greater eff- orts among marketing firms to sell potatoes. The result could then be a negative coefficient. A detailed knowledge of marketing conditions is required in order to find out what is the typical situ- ation in various months. It is well known that market- ing costs have accounted for substantially more than one- half of the retail price during our sampling period. According to Simmons, it is also a fact that "marketing costs such as grading, washing, packing, transportation, and wholesale and retail handling costs change slowly 78 over time, and generally change little in the short run."1 Quality may also be an important demand shifter. If some differentiated kind of potatoes such as, say, Idaho potatoes are of significantly higher quality and price than the rest, and if the relative supply of such potatoes varies much from year to year, then the use of an average price may tend to give the result that a high price is accompanied by a large quantity demanded, in the estimated demand relations. The reason for this is that the effects of quality differences on prices and demand are not recognized in such a formulation. One way of dealing with this difficulty is to include, as a variable in the demand function, the percen- tage of production which is Idaho potatoes. This quality indicator or demand shifter may then explain, say, a high demand in spite of high prices. In the general case, when we have several different qualities, we might use some weighted percentage. Since there is such a host of different qualities of potatoes and since quality specifications differ for various uses, such a procedure was not attempted in the present work. The available data are also deficient for this purpose. One difficulty in using, for instance, lSee Simmons, bb. cit., p. 3A and p. 17. 79 the percent of total production which is Idaho potatoes is that in some seasons Idaho potatoes are not produced, but supplied out of stocks, and the percentage of Idaho potatoes purchased may not be proportional to the per- centage produced. If we use purchases instead of pro- duction we still have a problem since purchases are en- dogenous. Deflated per capita disposable personal income was used as an income variable. (Seasonally adjusted quarterly totals at annual rates). This variable was assumed to be predetermined. According to the follow- ing quotation from Fox, this should be a safe procedure. "With reference to a system centering on the supply and demand curves for potatoes, disposable income might as well be treated as a predetermined variable."1 A change in tastes for potatoes may have oc- curred over time. In order to take this into account to some degree, a time variable was included in the demand functions. This variable may also partly take care of possible changes in buying and storing habits and other characteristics of the market. The use of a linear time variable is of course only a first approximation, but with the limited number of observations and the many lSee Karl A. Fox, Econometric Analysis for Public Poligy, (Ames, Iowa: The Iowa State College Press, 1958). p. 15. 80 potentially useful variables to be tried out, this was considered sufficient. The demand for potatoes for fresh consumption could profitably have been eXpressed on a per capita basis since this variable no doubt is approximately pro- portional to pOpulation. In this way we could have eliminated pOpulation as an eXplanatory variable in the demand functions. But the complete model also contains supply and other quantity variables which are linked to- gether with demand by means of an identity. If demand had been expressed on a per capita basis, it would have been inconvenient to express the other quantity vari- ables differently. The use of per capita quantities may not be good theory for all of these other quantity variables, however. There was also another reason for not deflating demand with pOpulation. The population trend had been approximately linear during the sampling period. Since time was included as an eXplanatory variable in the demand equation, it was anticipated that it would not be necessary at all to retain pOpulation as an explanatory variable in the same equation, and so it also turned out. 81 Demand for Potatoes for Processing1 The demand for potatoes by processing firms is complex, and in this section we only intend to separate out some key variables or indicators that can be used to approximately explain this demand. Unfortunately, very few pertinent data are available, especially on a month— ly or seasonal basis. Some monthly data on potato pro- cessing in Idaho and Maine are shown in Tables 15 and 16. These data seem to indicate that there is no extreme seasonality in potato processing. The price paid by processors for potatoes is obviously a variable related to processing demand, but unfortunately prices paid for potatoes for processing have not been available.2 The price received by farmers for all sales should be a workable approximation for these prices, however. An average price for all sales is especially relevant when the corresponding demands are treated together, as was the case in this thesis. Another group of variables that would seem to be relevant are the prices the processors expect to receive for the finished products. An index of current 1The various technical, biologic, and economic aspects of potato processing are ably described in William F. Talburt and Ora Smith. Potato Processing (2d. ed.; Westport, Connecticut: The Avi Publishing Company, Inc., 1967), pp. 1-588. 2Simmons, bp. cit., p. 27. 82 prices received by processors for their products might be used as an indication of these expectations. Processing costs might also be important, espec- ially costs of labor and of raw materials other than potatoes. Costs of equipment, machinery, and buildings will probably affect demand very little in the short run, although they will affect replacement of worn equipment and the speed of starting up new plants in a period of expansion of processing. As long as there is unused capacity, only variable costs are important. Since most processing activities have been rapidly expanded during our sampling period, capacity has probably been fully utilized, and even costs of more durable means of pro- duction may have influenced short run demand for potato- es to a limited extent. Unused capacity may have existed in certain seasons, regions, or kinds of plants, but it is virtually impossible to obtain meaningful data on this. The data on stocks of frozen French fries reflect to some extent the state of the market for this product from 1955.1 Since most cost items other than the cost of the potatoes are of such a nature that they are common for this industry and several other industries (labor costs, transportation costs, costs of buildings, admini- 1These data, for the United States as a whole, are listed in Appendix B, Table 17. 83 stration, etc.), there might be some doubt whether an index of processing costs would turn out to be signifi- cant in the demand relation for potatoes for processing. The prices of such general items tend to move together with the general price level in such a fashion that other observable movements in these prices are compara- tively small. The proportions in which these items are used in various industries must of course also be taken into consideration in addition to their prices. The proportions tend to be roughly similar in broad groups of industries, however. Therefore the corresponding costs also tend to be similar for groups such as food processing industries. Since these costs tend to move together for competing industries, and since they will tend to be passed over to consumers through higher prices, the present writer would not expect them to significantly influence the demand for potatoes by processing firms. Prices and costs are not the only variables determining processors' demand for potatoes. The demand for the finished products, 1. e. the quantity that can be sold at each price, is another determinant. Stocks of processed products may be an indicator of deficient de- mand, but probably an unreliable one, since stocks can also be built up to meet an increasing demand or can have other functions. Among the other possible demand shifters, con- 8h summers' income, prices of close substitutes for processed potato products, and costs of marketing from the processing firms to the retail level might be considered. Consumers' income is presumably important, but it may be difficult to separate the effect of this variable from other causes such as changes in tastes and eating habits. A time trend may therefore work just as well. Marketing costs affect the prices consumers have to pay and therefore the quantity they demand. It is a question, however, whether they change so much that it is worthwile to include them. In a study like the present, one can afford to include only a few of the most important variables because of the few observations to our disposal. Prices of substitutes may be important. The question is whether we can find any substitutes that are close enough and important enough to really affect the demand for processed potato products. in this connect- ion one may also ask whether fresh potatoes and processed potato products substitute for each other to any sub- stantial degree. The answer to this question has impor- tant policy implications for potato growers. It would seem that, fortunately for potato growers, fresh potatoes and potato chips are not close substitutes for each other at the retail level. For other processed potato products the situation is different and there probably is consider- 85 able substitutability. An important factor that directly affects processors' demand for potatoes is the quality of the potatoes that are supplied. Since so very few pertinent data are available, it is very difficult to take this factor quantitatively into account, however, The conclusion of this discussion is that there are several variables that probably affect processing demand, but very few data are readily available, and relatively little is known a prior'. A thorough analysis of this demand on a monthly basis will require very mmch work and is best left to a separate analysis. In the present work the only explanatory vari- ables used to explain the demand for potatoes for proces- sing are (l) the deflated average price of potatoes re- ceived by farmers for potatoes, (2) a time trend, and (3) two alternative time series or variables relating to an- nual processing activities. The first of the alternative variables mentioned in the preceeding paragraph has been denoted as the total quantity of potatoes processed and the other as the quan- tity of potatoes processed as chips and shoestrings. The two time series are described in Chapter III. The two time series relate to the crops of a series of calendar years. In applying each of these annual series in the models for particular months, the following procedure 86 was adopted to take into account the fact that potatoes are stored: For the months January-April the processing figures were used with a one year lag. For May and June the aritmetic mean of the lagged and the current proces- sing series was used. For July-December the current processing series were used. The rationale for including these variables in the demand functions is, of course, that when processing activity on an annual basis is high, there may be a tendency for the demand for potatoes for processing in a single month also to be high. The processing variables were assumed to be pre- determined. This assumption is probably approximately fulfilled. Demand for Potatoes by the Government for Diversion Purposes In some years during our sampling period a government program has been operated by which certain quantities of potatoes have been diverted to inferior uses such as starch and feed.1 The prices and quality requirements for such potatoes have been determined by the government. If the government had had perfect knowledge, the fixing of prices 1The quantities diverted during the sampling period of the present work are shown in Appendix B, tables 10 and ll. 87 would have resulted in a quantity that would have been known beforehand. In such a state of affairs, it would make no difference whether we say that price is deter- mined by the government or we say that quantity is So determined. Since knowledge is less than perfect, the fix- ing of prices is not equivalent to fixing quantities. The difference between these procedures amounts to an uncertainty as to the exact quantities resulting from certain prices. For our purpose it seems satisfactory to regard the diverted quantities as fixed by the govern- ment. Another question is whether these quantities can be regarded as truly exogenous. In setting up the program, the government of course takes into account the state of the market. If the government changes the program during the marketing season, we may say that the government not only affects the market but that it is also affected by the market as far as diversion programs are concerned. To specify in the model how the govern- ment is affected is practically impossible, however. We must therefore be content to treat diversion as an exo- genous variable. Demand for Potatoes for Storage The demand for potatoes for storage and the supply out of stocks is very complex and difficult to 88 handle. In the models of the present work, the way of dealing with stocks was dictated by the particular lay- out of the investigation and of the available data. It will be recalled that the data on sales of potatoes are not actually recorded data, but rather data constructed on the basis of certain proportionality assumptions. The fact that the data were constructed leaves some room for alternative interpretations of them. Especially in regard to stocks, a couple of alter- native interpretations are possible. One interpretation is the following: Since the data on sales are based on shipments, and since the group of shippers includes not only farmers but also local dealers and other owners of storage potato- es, the estimated sales data must be interpreted as sales from farmers and local dealers, cooperatives, etc. Our former assertion that the farm gate is the dividing line between supply and demand must therefore be modified to the extent that potatoes are stored by local agencies other than farmers. It might be more appropriate, then, to say that the shipping points are the dividing line between supply and demand. According to this point of view, farmers and various local agencies participating in the storing of potatoes with a view to later shipment are treated as a single group, the suppliers. More specifically we may say: The potatoes held in storage 89 by farmers and the potatoes held in storage by local storing agencies are all regarded as potatoes for poten- tial supply. Trade between farmers and these local agencies is not dealt with at all in the models. The supply functions of both these groups are treated as a unit. If the importance of each group as suppliers is uniform over time, or is changing according to a linear trend, this interpretation and the models that have been used are in complete harmony. Contract pricing has been practiced to some extent for potatoes. This phenomenon may cause distur- bances in a model based on the assumption of free market pricing within each time period. If the above mentioned interpretation of the sales data is adopted, contract pricing ceases to be a problem to the extent that it takes place between farmers and other local shippers. An alternative interpretation of the estimated sales data is the following: The farm gate is upheld as the dividing line between the suppliers (the growers) and the rest of the market. The fact that the shipments data on which the derived sales data are based also in- clude shipments from local dealers is taken into account by assuming that shipments from farmers are always rough- ly proportional to shipments from other agencies. This interpretation seems natural when we recall that the sales data for each crop that were distributed on months 90 according to shipments pertain to sales frgm_f§gm§. But this does not exclude the first interpretation, for sales from farms must be the basis for sales from ship- ping points in any case. Each interpretation has something to say for‘ it, but since sales data are based on shipments data and shipments are made not only by farmers but also by others at the local level, the first interpretation seems preferable. According tothis view, supply in the models is regarded as a supply from growers and other local owners of potatoes. These local owners of pota- toes (cooperatives, dealers operating farms, etc.) are often closely related to the farms. Their behavior as far as storing and shipping of potatoes is concerned, can be assumed to be very similar to that of farmers since they are performing a function that is very often performed by the farmers themselves. There are of course differences between the two groups that may cause dif- ferent behaviour, but these differences are probably not very important. When we adapt this point of view, the demand for potatoes for storage by local shippers does not enter into our models as demand at all. On the cont- rary, since these shippers and their suppliers are treated as a single group of suppliers, their demand for potatoes for storage is taken into account in the supply 91 function. According to this interpretation the demand for potatoes for storage is regarded as demand only when the demanders belong to one of the following groups: (1) local dealers or consumers that are not shippers and that do not sell to shippers, (2) dealers and consumers in the terminal markets. In the following, the expres- sion "demand for storage" will then include only demand for storage by these. In this way a large proportion of the more speculative storage demands are excluded from consideration. Stocks of potatoes are held by the demanders just mentioned either for speculative purposes or in order to meet future demands. The latter reason for holding stocks is probably the most important one. To some extent the demand for potatoes for storage is a derived demand. In the context of a model to be estimated on the basis of monthly data, the demand for potatoes to be stored by marketing firms for a few days or weeks is determined partly by the same variables that determine demand for current use. Changes in marketing practices and market structure may also affect storage demand. Such changes are difficult to measure, but a time trend may account for some of the effects of these changes. The more speculative demand for potatoes for 92 storage over a longer period must be explained by vari- ables that are related to future price expectations. Costs are also important, but these are to a large extent fixed costs or variable costs that are either stable or increasing over time in such a way that they can be approximately described by a trend variable. Specula- tive storing of potatoes over periods as long as a couple of weeks or more probably take place mostly on farms or nearby. The demand for potatoes for such storage is thus an internal phenomenon that will not be considered as demand in the present context. CHAPTER VII THE MODELS Some General Considerations The conceptual framework of the present investi- gation is a simultaneous equation's model containing separate demand equations for potatoes for (1) seed, (2) feed, (3) fresh food, (fl) processing, (5) diversion, (6) exports, and (7) stocks. It is clear from the preceed- ing discussion that the total market for potatoes is very complex, and that monthly data for an analysis where these demands are treated separately are lacking at the present. The necessity for simplification of the models is thus obvious. The lines along which the conceptual model was simplified will be explained in the following. In earlier econometric investigations of the potato market, exports and imports of potatoes have been ignored alltogether. In the present work exports and im- ports were recognized, but they were treated as exogen- ous variables. In the mOdels, exports and imports were combined to a single variable called net exports. The monthly net exports are sometimes positive and sometimes 93 9“ negative. They were explicitly recognized in the models as a separate source of demand. Also government iiversion of potatoes to infer- ior uses was treated as a separate exogenous source of demand. Diversion is actually different from net ex- ports in several ways. While net exports remove pota- toes completely from the market, diversion results in the potatoes being used within the country, either for feed or for starch. Diversion may therefore, to a very small extent at least, affect commercial demand for potatoes. In the models of this thesis net exports and government diversion were both treated equivalently, namely as exogenous demand, but they were not combined to a single variable.1 Since separate monthly data are lacking, the following demand functions were aggregated into a single demand function: Demand for (1) seed from farms other than where grown, (2) feed on farms other than where grown, (3) fresh food, except demand on farms where grown, (M) processing, and (5) storage, as defined above. The quantities of potatoes demanded for each of these purposes have not been separated, since this is 1Whether these variables are combined to a single variable or not makes some difference when simultaneous equations methods are used for estimation. 95 practically impossible. It was recognized, however, that the various demands have their separate demand determi- nants, and these have, as far as possible, been included as explanatory variables in the aggregated demand func- tion. Data are available only for very few explanatory variables pertaining to each type of demand. Also, only a limited number of explanatory variables could be included in the aggregated demand function due to the short sampling period at disposal. For both these reasons, some of the explanatory variables in the ag- gregated demand function were chosed so as to pertain to more than one kind of demand. An example of such a variable is the price of potatoes. Ideally, different price series should be used to explain the demand for seed, feed, food, and processing. Separate price series are not available, however. Instead, the average price received by the farmers was used in the aggregate demand function as a common indicator of prices for the various kinds of potatoes. Since this price actually is an average price for approximately just these demands, the procedure probably is sound. Moreover, this price is also the appropriate one to use on the supply side. The price variable is probably a relevant explanatory variable in all the demand equations. Another example of such an explanatory variable 96 is the time trend. The coefficient of the trend vari- able may be different in different demand functions. It may even be positive in some functions and negative in others. A priori one would expect a time trend to be present in all or most of the demand equations. The use of a common time variable in the aggregated demand func- tion takes this possibility fully into account. The coefficient of the time variable in the aggregated demand function will be a weighted average of the coefficients of the time variable in the individual demand equations. Farm-Retail Spread Relations As mentioned earlier, the only price of pota- toes that was used in the entire model is the average price received by farmers. ‘This price is appropriate in the supply function. It is also suitable in the demand functions for seed, feed, and processing. The demand at the farm level for potatoes for fresh use is derived from consumer demand. In the cor- cesponding demand function the retail price should there- fore possibly be included. Since all demand functions mentioned above were conceptually added up to an aggre- gate demand function, both the price received by farmers and the retail price probably ought to have been included in this function. The two prices are correlated, how- ever, and their difference is roughly equal to the farm- 97 retail spread. Instead of letting the retail price appear together with the price received by farmers in the aggregated demand relation, it was therefore decided to substitute the farm-retail spread for the retail price in this relation. One would eXpect a priori that the farmfretail spread is an endogenous variable. The farm-retail spread is determined in a very complex market involving supply and demand for marketing services for potatoes. This market again, is related to the supplies and demands for other products and marketing services. It was not our purpose to explore and describe econometrically this complex structure. Only a very crude farm-retail spread relation was formulated with the aim to take care of some major interrelationships. In setting up the farm-retail spread relation, a publi- cation from the United States Department of Agriculture was of much help. In this publication the farm-retail spreads are also defined and described in detail.1 The Economic Model The first economic model that was constructed 1U. S. Dept. of Agriculture, Agricultural Marketing Service, Marketing Research Division, Farm- Retail Spreads for Food Products, Miscellaneous Publica- tion No. 7M1“TWashington: U. S. Government Printing Office, November, 1957), 98 for estimation consisted, for each month, of four relations containing four endogenous variables. The four relations were (1) a supply relation, (2) a demand relation, (3) a farm-retail spread relation, and (u) an identity. The variables used in the following_are defined in the beginning of Appendix A. For easy reference, the symbols are also explained briefly in the text. Variab- les denoted by X have been assumed to be exogenous. The subscript m denotes month number (m 01 for January, 02 for February, . . . , 12 for December), while the sub- script t signifies the number of the year (t = 1952, 1953, . . . , 1966). (The subscript m is omitted for variables that do not refer to a particular month). The superscripts serve to identify the variables.1 The four endogenous variables that were includ- ed in the first model to be estimated were the following: (1) Total domestic supply of potatoes, Yig, i. e. estimat- ed shipments plus government diversion, X38. (2) Total domestic commercial demand for potatoes, YQE, i. e. NX estimated shipments minus net exports, X (3) Deflated FR mt' monthly average prices received by farmers for potatoes, mt' farm-retail spreads for potatoes, Y (A) Deflated 1For one variable, XEq, the symbol q was used in the superscript to denote quarter number (q = l, 2, 3, A). 99 PP mt' The identity eXpresses the fact that total domestic supply equals total domestic commercial demand plus government diverSion plus net foreign demand. The identity is common for all months and can be written as follows: SU _ DE GO NX (VII.1) Ymt - Ymt + th + th. Also the farm-retail spread equation was common for all months. It can be written in the following way: (VII.2) YFR = fFR(Y DE PP mt Y WH TE mt’ mt’ X Xt )° mt’ The argument runs as follows. The farm-retail spread, YEE, might be related to the quantity of pota- toes moving through the marketing channels that have DE produced the spreads. Ymt is an approximate expression for this quantity. The reason why Y3: is only an ap- DE mt also includes potatoes for seed, proximation is that Y feed, processing, and stocks. Potatoes for these pur- poses are only to a limited degree involved in the farm- retail spreads. Another reason is that there is a time lag which might be disturbing. The potatoes do not al- ways move the entire way from the producer to the con- sumer during a single month. PP Ymt’ the price of potatoes received by producers, was included in the farm-retail equation because this 100 price often is the basis for calculating various market- ing costs, such as interest, insurance, and certain marketing margins. The wholesale price index, ng, was included as an indicator of various costs that are too numerous to be mentioned individually. We may mention such broad groups as package, transportation, and equipment. The index in itself is probably also correlated with labor COStS . TE t 3 into account possible linear changes in farm-retail The time variable, X is included to take spreads over time. Next, let us turn to the supply relations. \ Some explanatory variables are present in the supply relation for every month, namely (1) the price of potatoes, YES, (2) a lagged price variable, Xifi, (3) a lagged supply variable, Xi%, and (A) the time vari- TE mt' month contains a group of explanatory variables that able, X In addition, the supply function for each are particular for that month or for some adjacent months. This group of variables consists of a produc- tion variable for each seasonal crop that is usually marketed partly during the month in question. In deciding which crop to include each month, Agriculture 101 Handbook No. 127 was very helpful.1 The crops that had production variables includ- ed as explanatory variables in the supply relations for. the various months, are the following (the symbol of each production variable is given in parentheses). January and February: Fall (XEL) and Winter (xgfi). March: Fall, Winter, and Early Spring (XEP)- April: Fall, Winter, Early Spring, and Late Spring (XEP)- May: Fall, Early Spring, and Late Spring. June: Fall, Early Spring, Late Spring, and Early Summer (XEU)- July: Late Spring, Early Summer, and Late Summer (X%U)- August: Late Spring, Early Summer, Late Summer, and Fall (new crop). September and October: Early Summer, Late Summer, and Fall. November and December: Late Summer and Fall. The supply function for January or February looks like the following: PP XFL WR TE XEE’ XSL XAE SU - f mt’ t ’ mt’ mt)' SU( mt ’ (VII.3) Y Y The Supply functions for the other months look the same, except that the production variables are different. The variables in the demand functions for the various months were common for all months. These varia- bles and the reasons for including them are listed in the following. lRoyston, Frost, and Galloway, pp. cit. 102 The price, Yii, of potatoes received by farmers was included as an explanatory variable in the demand function for obvious reasons. This price is the one that supposedly is important for demands at the farm level. The farm-retail spread, Yifi, is another variable that sup- posedly affects demand since it reflects costs and pro- fits imposed on the potatoes on their way from the farm to the consumer. The next three explanatory variables included in the demand functions were the Consumer Price Indices for (l) cereals and bakery products (X55), (2) fruits and vegetables (Xiz), and (3) meats, poultry, and mt ° A time trend was included to take care of pos- fish (X sible linear changes in demand over time. An acreage AE mt’ ation of the demand for potatoes for seed. This variable variable X was included to contribute to the explan- was an index based on the acreages of the potato crops the seed for which is usually shipped partly during the month in question. Finally, a variable related to the demand for potatoes for processing was included. This variable (XEP) was essentially the total quantity of potatoes used for processing during the year in question. The demand function for an arbitrary month then can be written as follows: DE mt PP FR PB PV PM TE : f mt’ mt’ mt’ mt’ th’ Xt DE(Y AE TP) (v11.u) Y mt’ xt . , X 103 Functional Form and Statistical Assumptions Very little knowledge is available that can be used to decide which functional forms the structural equations have. The choice of assumption on this point therefore was directed to a considerable degree by practical considerations. First of all the reader is reminded that the use of non-linear functions, like n'th degree polynomi- als, reduces the number of degrees of freedom. Since the sampling period is very short, such reductions ought to be avoided, if possible. Most econometric analyses that have been under- taken have been based on relations that are linear either in the original variables or in their logarithms. Once a decision has been made to use one of these functional forms, the choice between the two is really not a big problem since these functional forms seldom produce significantly different results. i In the present work it was decided to work with functions that are linear in the orginal variables. "Linear equations give results which, when translated into total value-supply curves, make more economic sense at the extremes than do the results obtained from logar- lOu itmic equations".1 This argument has some force in the present case since the supply of potatoes vary ex- tremely from year to year, but actually the choice is probably not important. Another reason for not using logarithms is that the logarithm of a negative number is not defined. Net exports are sometimes negative. Furthermore, if loga- rithms had been used, also the identity (VII.l) would have caused some trouble. A compelling argument for choosing to use linear functions can not be given. In any case, however, linear functions can be regarded as approximations to the true functions for values of the variables within certain limits. The statistical model for January can then be written as follows: SU _ SU (VII.5) Y - 801 + B SUPPYPP SUFL FL SUWR WR 'Olt xt 01 Olt * B01 Xt * 301 SUTE TE SUPL PL SUSL SL * 801 X1: + 801 x01t + 801 x01t SUAE AE SU * 501 Xcit + 6on 1Richard J. Foote, Analytical Tools 32; Stud in Demand and Price Structures, Agriculture Handbook No. 1E6, Agricultural Marketing Service, U. S. Dept. of Agricul- ture (Washington: U. S. Government Printing Office, August. 1958). p.37. 105 DE DE + DEPP PP DEFRYER DEPP BP (VII°6) YOlt : B01 B01 Yoit + 801 dlt + 801 XOlt + 88?an 83?an + 88$”sz + area + BEETPXEP + Egit mm at = a; + + area are. FRTE TE FR + R01 Xt + EOlt SU DE GO NX T : (V"I‘8) YOlt YOit + X01t + XOit' In equations (VII.5) - (VII.8) the subscript 01 is the value of m for January. The B's are constant coefficients. The first two letters in the superscript of a 8 indicate which variable is on the left side of the equation, and the next two are identical with the superscript of the variable to which the coefficient belongs. The estimates of the 8's will be denoted by b's with the same subscripts and super- scripts. Each 8 has the same superscript as the left-hand variable. This superscript thus shows to which equation an e belongs. The e's also have time-identifying sub- scripts identical with those of the X's and the Y's. 106 The following assumptions were made regarding the model (VII.5) - (VII.8). The figures ESU (t = 1952, 1953, Olt , 1966) are values of 15 different but identically distributed random variables EOEt (t = 1952, 1953, . . . , 1966).1 Each of these random variables has an expected value, E(eSU ) e ual to zero and a finite variance Var(eSU ) —Olt q ’ ’ —Olt ° The random variables corresponding to various values of t are mutually uncorrelated. Formally, we may write (VII.9) E.mwm u m.mmmm mHmm H Homsmm.ov HommHo.ov Hma.Hmmv Aw.mmamv mm.H mm.o mmeem.on momHH.o Hm.mmm - o.mmmm mamm-H H AHsHmm.ov AsseHo.ov Hmm.msmv Am.smmev mm.m Hm.o wmwsm.o- mmMOH.o oo.mHHH- N.NMHm mmHH H Hmmmmm.ov HemHHo.ov “om.we=v Ha.mmmmv mm.H mm.o smmmm.o- mmeHH.o sm.ssm - o.mmoe mama H Ammowm.ov AeHHHo.ov Hmo.msmv Ho.oewmv ww.H mm.o eeasm.o- smeHH.o mo.omm - m.ossm mHo H o o o5 max Hex may stop ooeooe HES 0 mm peepmcoo coapme .o: oo wpcoHoHooooo -Howm coco: mcoLMZIhnmzcmw .oe-mmmH ooHnoo we» too wepoos so mooHooHon sHoosm oouaeHomm--.H mqmee 168 Amm.mmmv Hmmmwm.ov HmHHHo.ov Ha.mmsmv mm.H :m.o mmmmo.o momeH.o mm.=Hm m.ma~mu mHmm m Hesomm.ov AsmOHo.ov Hm=.=mmv Hm.mmmmv om.H em.o Hm30H.o HezeH.o m:.Hmm m.meei mHo m Hmsmos.ov Amsomo.ov H=>.mmmHv Hm.NOOOHv om.H mo.o mmoom.o mommH.o sm.HmmH m.omms- Hsz m Amsmos.ov Amsomo.ov Aes.mmmHv Hm.moooHv om.H mo.o mmomm.o mommH.o so.Hme m.omms- HzHasm m Hmomsm.ov HommHo.ov Amo.emev Am.mmHmv mm.H mm.o smosH.o- mHmoH.o :H.o - m.mHmm mHmm N Ammem.ov Hmmeo.ov Ham.omev Am.Hmmmv mm.H mm.o :OHsH.o- OHNOH.o ms.o o.mmmm mHmmuH m Ammmmm.ov HmmmHo.ov Hmo.ssev Am.ommmv me.H ww.o oonH.o- meOH.o mm.aH s.msam mmHH m Hemmmm.ov AmmmHo.oV Hmm.msev Am.mmmmv mm.H ww.o momom.o- SNOOH.O mm.oo . =.mmmm mqmm m o o as o m max Hex me» stop oooooe Hsv m use mco coapme .o: co moooHoHecooo o o -Homm eoooz ooschcoouu.H Hemae .< xflo unmoa< do monogamoe who so oocfiwaaxo one moonpoe coopwEHuwo mo modem: som omzoaa nEo mcooumfl>osoom on» com canonsm> come now new: news: new mHooEmm one .co>Hm one .o .oflpmflpmum cowpmzucaepso one can .mm .COHmeHEoopoc oHoHpHsE ho ucoHon nmooo one measaoo osp unma on» CH .mposnc osmccmuw o>Hpomomon one one mucofioam IMoOo one Amos: mononucmamo do mopswfim one .cwflm muflamsvo one no puma one on mmw Spas coupes: mo npcoe some now coopmaoh madman on» dogs omocxoe can so oopmfla moanmfiam> one no mucofiofimmooo one use Shop pompmcoo on» mzonm oanmp ones smooHé - 383.8 Hones: 3.0mm; . nm.H no.0 0m00H.0 emonH.0 no.000 o.ommmu nzHo m AmmooH.00 Ano0H0.00 Aon.0mnv A0.0momv ”w nm.H no.0 0m00H.0 emonH.0 no.000 s.ommm- HsHauo m l Ao00no.00 HmHHH0.0V Amm.mmmv An.0m0mv om.H no.0 ss00m.0 omnnH.0 mm.ms0 m.000n- mHmm m Amomnm.00 HanH0.00 Hoo.n0mv H0.o0ou0 om.H no.0 o0o0m.0 0annH.0 mn.mh0 0.o00n- mHmm-H m Amsoom.0v HucHHo.00 H0e.HomV AH.muoo0 0n.H no.0 0HHm0.0 0H0nH.0 0m.nom H.momm- mmHH m o o. as c m max qu mmw some connoe AEV m pampmcoo coapme .oc oo mocoHoHooooo . -Homm specs ooncHocooan.H when? 170 Hmm0mH.00 Ammmom.00 HHmom0.0v A00.mHn0 A0.ommn0 mm.H no.0 0o0s0.0 0noo0.0 0mHmm. ma.0m0- n.s0n 0H0 m . . . nommmo.0v Homm00.0v Hom.HnoV Hm.mHnnV om.m 0o.0 . . . maHmm.0 onomn.0 0a.mm0H H.nonHH- nzHa n . . . nommoa.ov m Asmmo0.ov Aam.Hns0 Am.mHnn0 om.m 0o.0 . . . monm.0 m snoon.0 0s.mm0H H.nonHH- HzHa-m n . . . . NO0H0.00 _ AHmon0.0v Amo.mon0 Hm.0nnmv o0.m mo.0 . . . Hn0m0.0 _ 0oo0n.0 0a.mmn 0.00no - memo n . . . Hm0000.00 m Aoo0o0.00 An0.00m0 Hm.mmmmv 0m.m mo.0 . . . umomm.0 I 0o0Hn.0 nH.omn H.ommm - mHmm-H n . . . nonmoo.ov 1 Hnmmmo.00 Amo.Hmm0 A0.Hamm0 00.m No.0 . . . 0momm.0 0H0mn.o on.on0 m.0oso . mmHH n . . . Hn0m00.00 A0o0m0.00 AHo.00mv Ho.omnmv no.m mo.0 . . . 0mm00.0 mH0mn.0 no.0on m.0HHo . mHmm n om.~ mo.0 . . . em0mo.0 0mH0n.0 0m.nHm m.mm0o . 0H0 n ex on o: _ as 0 on ma on. Hs> a man soon ooeooe H20 c t, 0 made 0 ammo » pampmcoo cowums .0: r m a H o ufipmm coco: mocsmaHHsc< .monmmma coated one now menace he mcoapmfios madman comeHpmmnt.m mqm<9 171 Amman.0V . . . HHmnm0.0v Amo.o0mv H0.nm0nv nH.m ns.0 0mm0H.0 . . . momm0.0 mn.0on- o.moo0H mmHH 0 HommmH.0v . . . HOHnN0.0v Hmo.n0mv H0.0m0nv NH.m ns.0 ommmH.0 . . . momm0.0 mm.on- m.HomHH memo 0 HnnnHH.0v . . . Ho0mm0.00 Hon.mmmv Hn.0mmv m0.m mo.0 mommH.0 . . . Hmon0.0 mm.o0m- 0.msmmH moo 0 o o o o o o o a o o o o o o o o o o o quHm m Amnomm.0v H0m000.00 HHN000.0V A0H.omoHV Am.mmo00 00.H no.0 HmmHm.0 no00m.0 HonHm.0 o0.on n.0o00H- Hszum m AsmomH.0v An0mon.00 Amm0m0.00 HHn.omnv An.mHmnV 0m.H no.0 n0SMH.0 H00om.0 0oonm.0 no.0mo- 0.0mm - mHmm m HommmH.0v Hwommn.0v A0n0m0.0v Am0.Hnnv Hm.mmmnv 0m.H No.0 mommH.0 n00oo.0 00mnm.0 ns.om0- 0.0mm - mHmm-H m mn.H No.0 0ono0.0 m0om0.0 omoom.0 mm.ono- o.mmm - mmHH m HoonH.0v Amonom.0v Honmm0.0v Ao0.mnnv Ho.Hnmnv 0m.H .mo.0 0mms0.0 0o0no.0 m0mmm.0 Hn.nm0- n.on mHmm m o o o- as max mmx HEX mmw Shop wonpmfi AEV 0 mm pompmdo scenes .0: oo moooHoHooooo H 0 -Homm coco: vmzcflpcooui . N mamfiw 172 menses cocoooxo no: name can now: a Econ assoc mafia wanna one .houSQEOo on» so can on» .muasmoh oz 9 .¢ chcooo< we meascawoo on» an cocfimaoxo one escapes coopmsapmo no nose: no“ cohoaoso mcoapmH>oannm one can canmfinm> some now com: made: one maooszm one ImHH m®HQMHhfl> Gnu. MO mHCOHOHMh000 02% CC.“ 509. PCNOWCOU 029 $303” ”HQNQ. @390 .:o>aw mam .o soapmaumpm schematcfinnsn one can .Nm adofiumsoahopoo oaowpase mo pomfioammooo on» mossaoo esp pmca on» :H can mudoaofihwooo one hoods mononpdohmo CH headway one 0» mm» new: coupfinz on names some now coaumaoh hang .mnonno pudendum oefiuooomoh one .omHm soHHoooo be» do pooH on» am one does cmonxoo on» :a cop AmoonH.0v . . . HomHn0.0. Hmo.00ov Hn.0mHov mm.m on.0 mownm.0 . . , omnoH-0 0o.0mm o.H0oH asHm m AmoanH.0v . . . H0~Hn0.00 Heo.00wv Hn.0mHev mm.m nn.0 momnm.0 . . . o0n0H.0 0o.0mm o.H0oH HzHaum o Hmomo0.0v . . . Ammmm0.0v “moynomv Hs.mon0 m0.m ms.0 mmmoH.0 . . . momo0.0 omeHoni_ m.0om0H memo 0 H0mno0.0v . . . Honmm0.0v HMH.momv Hn.nHmmv m0-m mo.0 H0H0H.0 . . . momm0.0 0m.mmnu n.mHn0H mamm-aw 0 o o o as x x M M 003908 AEV 6 mm m4 mm H: mm qwmwwo coauma .0: do muooHoHuoooo o o .nomm coco: ooschcoou-.m mqmee 173 Aomn00.00 HmoooH.00 . . . HHHNH0.0V H00.o0mv Hm.m0m00 0m.m oo.0 o0Ho0.0- mommm.o- . . . 0s0m0.0 m0.mnm m.000HH mHo 0 Ao0nm0.00 AnmmmH.0v Hno0m0.0v . . . Hmm.omnv H0.s0wmv 00.m 0o.0 on00H.0- omomm.0 n00m0.0 . . . mo.mmn 0.000m Hsz a Ao0n00.0v AsmmmH.oV Hno0m0.0v . . . Hm0.omn0 A0.o0mmv 00.m mm.0 sn00H.0- omamm.0 n00m0.0 . . . mo.mmn 0.0000 Hsz-m o “momm0.0v AHnan.0V Ho00m0.00 . . . Ams.oomv HH.m0nmv 00.m no.0 Hmmn0.0- mmmom.0 00mm0.0 . . . mm.omH o.0mm0 memo a Hoom00.0v HmHmmH.00 HHHHm0.00 . . . Hoo.HHm0 Hm.0mnm0 m0.m 0o.0 amon0.0- mmmem.0 mnmm0.0 . . . oH.mmH n.0Hn0, mHmmuq o HmomnH.0V Ho0omm.00 Homon0.00 . . . Hmm.ommv Hm.o0mmv mo.H mm.0 on0o0.0 n0omm.0 anm0.0 . . . oo.mHm m.nmmm mmHH o AMHmmH.00 A000nm.0o Homon0.00 . . . Hmm.o0mv Hm.onHm0 no.H no.0 NH000.0 nmon.0 mmmo0.0 . . . nm.nmm n.mmHm mHmm o A00smH.0v Ammoou.0v Hommn0.0v . . . Aom.HmmV Hn.osmn0 oo.H m0.0 m0o00.0 HNm0H.0 oH000.0 . . . 0mm.mou 0.mmmm mHo o o o o o as eHx pox Hex Hex moo ooeooe Hey c mm cmnmwo cowume .oc no mpdoaofiumooo u p o owpmm pesos whooEoooQazHSh .mwnmmma penned on» non mnpcoE he nQOHuwHos madman copmefipmmuu.m mqmge 17A Hoooo0.0v . . . . . . Amoo00.0v Hoo.oonv Am.HHmnv mo.H oo.0 ooom0.0- . . . . . . Hoom0.0 nn.0mH o.nonH mmHH o noooo0.00 . . . . . . AHoo00.0v Am0.oon0 Ho.oomnv oo.H oo.0 moom0.0- . . . . . . nmom0.0 om.ow n.sommH mHmm o Homoo0.0v . . . . . . Anoo00.0v HmH.oonv Ho.mnmnv .oo.H oo.0 mo0n0.0- . . . . . . o0om0.0 00.0o m.oman mHo o Hmooo0.0v HoHoom.0v . . . HoHoH0.0v Hoo.ooov Ao.Hm0nV om.m no.0 0omo0.0- oomoo.0 . . . oHHo0.0 m0.HHon o.onmnH HzHa m Amooo0.00 AoHoom.0v . . . HoHOH0.0V Aso.oomv Ho.Hm0nv om.m no.0 0omo0.0- oom0o.0 . . . oHHo0.0 m0.HHon o.snmnH HzHoum o AmmHo0.0v AmoowH.0v . . . Hoo0H0.0v noo.momv _AH.om0mv mm.m os.0 oom0H.0- oH0o0.0- . . . so0n0.0 oo.omm o.m00mH mHmm o Aoooo0.00 Ammnom.00 . . . HmonH0.00 Hom.ooo0 Hm.ooan mm.“ oo.0 oomo0.0- oooHn.0- . . . sooo0.0 oom.mn - m.m0omH mammaH o Hn0an.00 AHooom.00 . . . Hoo0H0.0v Hoo.Hoo0 HH.0ooov m nH.m 0o.0 nHuHH.0- omHoo.0- . . . 00oo0.0 mn.ooo- n.moo0H amHHl o A0nom0.00 noommm.0v . . . Hman0.0v AHo.omnv Am.mmomv om.m oo.0 nnmo0.0- moHom.0- . . . noHoo.0 0m.oQH m.oman memo o s7 o o o; o as 0ex sax Has sax om» some ooeoos Hsv e um pompmcoo cowpms .0: c co mocoHoHooooo -Homm coco: GmSCHpCOOII.m mqde 175 HoMMMH.0V . . . . AHonH0.00 Hoo.n0ov no.sooo0 o0.m on.0 o0ooH.0 . . . . ooom0.0 oo.wm u n.0omo memo 0H “anom.0v . . . . HHomno.0v HoH.mommv no.H00oHv m0.m on.0 oommm.0 . . . . MHom0.0 nH.0 . o.HoHs mHmm-H 0H Honoo.00 . . . . AHHoo0.0v Hon.oHoo0 Ho.oooomv oo.H Hs.m- noHno.0 . . . . oomo0.0 om.oHno 0.nnHom- mmHH 0H HoomoH.00 . . . . H0HoH0.0v noo.omov H0.mnoov o0.m on.0 o0moH.0 . . . . oo0n0.0 0m.om . m.oooo mHmm 0H HooonH.00 . . . . noonH0.0v H00.mHov mm.nmnov o0.m an.0 HoomH.0 . . . . m0om0.0 00.omm . 0.nooo moo 0H H0noo0.00 . . . . “nooH0.0v Hom.onov nH.oomov mm.m no.0 onoH0.0- . . . . oo0n0.0 mm.ooo m.oooo HzHe o H0noo0.00 . . . . HnoMH0.0v Hom.onov HH.oomov mm.m no.0 mnMH0.0u . . . . mo0n0.0 mm.oom m.oomo qum-m o Honoo0.0v . . . . HoHo00.0v H0.0Homv oo.H oo.0 momo0.0u . . . . oonm0.0 -Wo.oo0nH mHmm o I Hom0o0.0v . . . . Ammo00.0v Hoo.0onv Hn.HoomV oo.H oo.0 onmo0.0- . . . . o0om0.0 Ho.oH . o.nH0nH mammnH o o e o u , as x xx .x x w 939: 25 o m :4 as do arm mm CMAMWO incense . o: N no moooHoHooooo o o 0 -Homm specs 603:..nucool: . m mamde . . . . . . . . Hmo0m0.00 Hoo.onHm0 Ho.ooo00 oo.H on.0 . . . . . . . . ooHo0.0 oo.Honm- m.0ommH HzHo HH . . . . . . . . Hmo0m0.0v Hom.onHm0 Ho.oooo0 oo.H on.0 . . . . . . . . ooHo0.0 oo.Honm- m.oommH HzHaum HH . . . . . . . . HooNH0.0V Hoo.Hmo0 An.ooomv ms.H o0.0 . . . . . . . . nm0o0.0 oo.H0o . n.0owm mHmm HH . . . . . . . . HH0om0.0v Hoo.oomHv Ho.ooHov oo.H oo.0 . . . . . . . . mmHo0.0 MH.Hoo . o.0oom mHmm-H HH . . . . . . . . HHmmm0.00 Hoo.omomv Ho.onoov oo.H Hm.0 . . . . . . . . ono0.0 0o.omomu o.0monH mmHH HH ,o . . . . . . . . HoomH0.00 Hoo.Hmov H0.0ommv w“ oo.H oo.0 . . . . . . . . nm0o0.0 mo.H0o . m.0oom mHmm HH . . . . . . . . HommH0.00 H0o.omov Hm.moomv oo.H oo.0 . . . . . . . . ommo0.0 om.mnm u o.nm0m mHo HH 0 o o o o o o o o a o o o o o o o o o o o DqHzHfiH OH Hooooo.H0 . . . . . HooooH.0V Hom.onoHVHn.omno0Ha Hn.H Ho.oa ooHoo.0- . . . . . moon0.0- Ho.ono0H- o.o0ooo HzHa-m 0H o o o o as . :Hx 0mx st nan om» shop ooeooe Hsv c mm ucmpmcoo coapme .0: mo mucoHonmooo n aHpmm specs Umzcfiucooll.m m4m<9 177 noo.omomo . . . . Hmnnm0.00 Hm.m0NNHV on.H no.0 . . . . oonH.0 0N.oNom o.nNoHHleHm-N NH . . . . HnHNH0.0v HNo.onov HH.HooNv nn.H oo.0 . . . . m0o0H.0 oN.moo n.NoNH - mHmm NH . . . . HnoNH0.00 Hmo.onov HN.HooNv nn.H oo.0 . . . . o0oOH.0 .No.moo o.00mH lmHmm-H NH . . . . HooNH0.0V HoH.onov Ho.oooNv mn.H oo.0 . . . . m0o0H.0 No.0no om.0oo . mmHH NH . . . . HooNH0.0v Hoo.ono0 H0.oooNV Nn.H o0.0 . . . . mNo0H.0 Hm.Hoo Nn.0no - mHmN NH . . . . HnsNH0.00 Hom.oHov Ho.0ooNv Nn.H oo.0 . . . . oNo0H.0 oo.ooo 0H.Noo- mHo NH o o o o as ea ox Hex Hex on» some oeoos Hev 0 mm 0 ucmumcoo Oprs .0: no mocoHoHooooo -Homm coco: omscfipcooll.m mqm _ a 1..” H. v.,.- ”ac. HH u_ _. _ n‘fiJa .~ I V 12.1.0... I .n“- .h . _ _. .11.; 1 . .i . I. 1 . Lutit - n1. .‘. uh ‘ I. 4 3 .- :vr . . 1.... 3.... _ . . . .. . .. o H. .. I. .e. . .7: . . .I‘QIAOI v If! i . , .1 0.11;! , I . I t v I. 4 ti! '.|| ’{l‘!' I- ‘i‘ v \ i 1 o(.v I .|.A\U.Alli.". c .1 .. 178 .soHSQEoo one so mmoo logo COHponopH one mcHsso oesopcsooco mm: xHHme Hmstch < .mpHZmop 029 .< xHecooo< mo wCHccHwoo on» CH oochHoxo ohm moozuos :oHmeHpmo mo moss: som poonan mQOHQMH>osoom one cow oHomHHm> some now owns mpHcs pom mHooEmm oLB .co>Hw one .o .ooumfipmpm commeICHoLSQ on» com .mm .COHHMQHELopoU oHQHpHse no pcoHonooOo one mcEsHoo 03p pmmH one CH .msosso oncogene o>Hpooomos on» one mucoHonmooo on» Loos: mononpcosmo CH mossmHo wee .cmHm mpHHMSUo one no puma one on man :oHs copoan mH ences some som COHpmHos oHoosm one cogs cwoexoe one CH oopmHH moHanLm> ecu mo mesoooogmooo one cow Esme pompmcoo one msosm canoe ones Iéillii!-:IH!i u , -I, . . . a- . . . . . . Honoo0.0v H0o.oooov Ho.N0NNH0 w:.a :m.o . . . - . . . . . . mmHMH.O ON.NN®M m.:NmHHI QEHm NH s . _ o o. o as < . , o to 0d: _ 0mg Has Han om» soon ooooos Hsv . Ill}! :i9:_:;;.,_::;-:léil:aa . ,; -e,eL|Il!:i!!li cm moo :oHme .0: no mquHOHmaooo p p o lopmmbzpcoz ooocHocoouu.m means 179 m0.N oo.0 . . . oH.mon No.oNo on.0mo - N.oonom- mHo N . . . Hom.nnHv H0N.oNNV Hoo.ooan HH.oonNNV oo.N om.0 . . . om.oHo om.0o Hn.oMNN H.oooN Hsz H . . . Ann.oHNv Hoo.HoNv Ann.H0oov HN.omn0nv Ho.H oo.H- . . . oo.an nm.o 0o.ooso- n.ooooN HzHo-N H . . . Hoo.nov HHO.oMHv Hoo.nomv HN.omnMHv oN.N No.0 . . . mo.o0o no.moN Ho.o0o . 0.xnoNH- mHmm H . . . HNo.oov HHo.mnHv Hno.Nom0 Hm.oooMHv 0m.N No.0 . . . 0o.o0o Hn.HoN m0.oNo . o.ooHnH- mHmm-H H . . . HNo.oov H0o.onHV Hn0.o0nv Ho.onNnHv xN.N No.0 . . . oo.H0o o0.NNm MH.nn0 . N.oNHoH- mmHH H . . . Hoo.nov HoH.oMHv H0o.nxmv Ho.0oanv oN.N No.0 . . . 0o.o0o Ho.ooN 0o.ooo . o.onHNH- mHmN H Hom.nov Ho0.NmHV Hn0.NHmV Ho.HooNHV 0o.N mo.0 o0.o0o no.ooN Hn.Hon . o.N0n0H- mHo H as o as He max max >nx no» stop ooenos may 6 mm m C0 0 m0 pcmumcoo COMNMM SUN”: 0 II o o H Ham 0 wheroooauzum:CMh .wmummmH pooped on» son canoe an mCOHumHoa cacaoc coumEHpmmun.n mqmnx max some mmmwmm WWW N no moooHoHooooo oooomooo -Homm ooooz Umscflpcooun . a mqmnne 185 haw.mamoav . . Amm.flmmv Aom.flw~mv Am.mmo:mmv mm.m mm.:- . . mm.mmm Nn.omma- oa.ao~m m.m~mmmfi qum m . . Amm.ammv Amm.fimpmv Azw.mam0Hv Am.mmo=mmv mm.m mm.:u . . mm.mmm Nn.ooman oa.:o~m m.m~mmma qumum m . . Aqm.mmv Amo.mmav AHH.~m:V A=.Hhomav NH.N m~.o . . ow.mmm mm.mma mm.mam =.moa mqmm m . . Azm.mmav Amm.m::v Amm.mmmav A~.mmam:v Hm.H m:.o . . om.~mm No.50: I um.oomfl H.=Hmmm mqmm-q m . . ANH.:mmV A=N.mmaav Awm.~mmmv A~.mm~:oav mm.m mm.a- . . w:.m=: Fm.mmm . ma.m-m m.omamm mqu m . . Auo.mmv Anm.mwav Amm.mwav AH.omm~Hv mH.m :~.o . . om.omm =~.ma :m.mam m.m~:ma mqmm m . . Amm.mmv Amo.amav Ahm.mzmv Am.mm~mav ma.m m>.o . . am.amm Fm.mfl Hm.mm: m.am:oa mqo w . . Amm.~:v Amm.omv Aom.mmmv AH.mmev ~m.a No.0 . . H©.mom om.o~ . :m.~oa . m.mmmwa qum N pa p- p5 as m¢x may >¢x mmw Sam» nogpme Asv 6 mm pampmcoo coapme .0: no mpcmflOflmmmoo -prm spec: Omscflpcooll.: mqm<8 186 505.0000 . . 500.000 505.00HV 50.Hmmmfiv m0.m m0.0 . . Ho.0mm 00.000 HH.m=m 0.000: . 000 0a . . Am0.0mmmv Amm.0ommmv 500.00m000 Am.0amwammv 00.H 00.00- . . m0.500- om.m0mm m=.mm05a- 0.0mmawa- qum 0 . . Am0.0mmmv Amm.00mmmv 500.00m000 Am.0ammammv 00.H 00.00- . . m0.a00- 00.mmmm m=.mm05a- 0.000000- 02H0-m m . . 500.000 50m.55av 500.0000 50.0000Hv 00.H m5.0 . . Hm.mmm mm.mmm 0:.mma - H.0mmm . 000m 0 . . A5=.mmv 50m.HOH0 Amm.HHmv 50.H=m5av mm.a m5.0 . . 50.0mm 5m.m00 mo.H0H . m.mm00a . mammuq m . . A00.0mv 500.0Hmv 500.05mv Aa.mmmomv m0.a 55.0 . . 55.0mm 00.0mm no.Nmm - 0.0m00m . mqu 0 . . A00.0m0 AH0.0HNV A00.00m0 50.Hmmomv . 00.H 55.0 . . 05.0mm 0H.5Hm H=.:00 - 0.5000m . 0000 0 . . Am5.mmv Am0.HHmv 550.0mmv 50.00H0m0 00.H 55.0 . . 5m.H00 00.000 :5.Hmm . H.00mmm - 000 m we 0 as us m00 000 0000 wmmwmm WWW 0 00 000000000000 00000000 -0000 0000: nmscfiucooll.: mqm00 000 0000 000000 000 0 mm unapmcoo :00008 .0c 00 000000000000 -0000 00002 Umscapcooll . 00 ”.0393. 189 .3mu3asoo m3» 3o mmmo uouq 300pmhmp0 030 030333 vm3m033oo3m 003 003003 00030300 < .mp030m3 020 .3mp3QEou may 30 333 m3» 0303 I33 omcmmoxm was @000 333A 3mm3 m 8030 03300 m300 03033 use .mp03mm3 023 .< 0063mqa¢ mo w30330wmn ms» 30 Um3000axm mhm muonpme 300me0umm no mmsm3 you ummo0afim 0300900>m3nnm m3» 330 m0nd0pm> nomm you @003 00033 330 0000800 mne .3m>0m mam .3 .000000000 3000031300339 m3» 330 .0m .300003050mpm3 m0g0u03e mo p3m0o0m Immoo m3» 0383000 03» p000 map 30 .mnogpm 03033000 m>0pomammp m3» mnm 00300000 ummoo m3» pm333 mmmmnp3mnma 30 mm3300m was .3w00 0000m3dm m3» mo pmm0 m3» on mm» 3003 3mpp033 00 30305 300m pom 30000003 c3msmc mnu 3m33 vmmnxon m3» 30 cm0000 m0nm0hm> map 00 mp3m0o0mmmoo m3» 330 sump 03000300 m3» 03030 @0000 @390 . . . 050.000 000.0000 000.5000 00.000000 00.0 00.0 . . . 000050 00.000 00.000 - 0.00000- 0200 00 . . . .050.000 . 000.0000 000.5000 00.000000 00.0 00.0 . . . 00.050 00.000 00.000 - 0.00000- 0200-0 00 . . . 000.000 _ 000.0500 000.0000 05.000000 00.0 00.0 . . . 00.000 50.000 00.00 - 0.05000- 0000 00 . . . 000.000 000.0500 000.0000 00.000000 . 00.0 00.0 . . . 00.000 00.000 00.00 - 0.00000- 0000-0 00 us a as as x x x w conums 050 0 0m 00 09 >0 mm pcmmmwoo 000005 .03 -0300 0030: 0o mpcm0o0uumo0 Umszfipcooll.= m0mwm ohm mmopsow m sm6.smm omauom 0:6.mfi smo.mm :mm.: :mo.m mwmfi ocm.mfim www.mm mmm.HH amm.:m o:m.: mmo.m mwmfl Hm:.:sa smm.sm mHN.HH mms.ma wmfl.: Ham.m somfi mmfi.mofl mma.mm :mm.mH mom.mm mma.m mom.m moma msm.:os :mm.mm mmm.mH omH.Hm mm:.m owfl.: moms mmm.:om Hm:.mm mom.ma omo.om mzo.: som.: Hmofi Hso.msfi mam.zm emm.:fi mmm.mm mm:.m som.m coma mm:.sma Hms.:m sow.ma :ma.mm o:a.m moo.: mmma swfi.amfi mam.mm soc.:H figs.mm mos.: Hsm.s mmmfi :ms.mma mo~.mm mum.HH amo.sm mo:.s 005.6 smma Hmm.sma sma.mn mmo.HH o:w.am mmo.: oom.m mmofi mmm.m:a 6mm.Hm Hoo.:fi mmm.mm oom.m msa.m mmmfi moo.mzfi 6:0.mm smH.HH smo.mm mmm.m mms.m amma msm.HmH m:m.mm wmo.HH wou.sm Hmm.m Hmm.m mmma moa.H:H mmm.am mca.m mmm.mm :ma.m Hos.m mmofi Hos.mmfi 6mm.fim www.mfi mmz.om mmm.m omn.m Hmma Hamm LoEésm mama pmaezm zapmm _mcfipom mama meHumm mahmm gonna: pdmm muzwflmzfimhficzfl OOO.H ma .masopm ammonwow an .wmumpw omufic: "cofiuozvopm Oumuou Hauoaul.m mqmwav coamnm>fio o o o o o o o o o o o o o o o o o o o o o o o o WWOH o o o o o o o o o o o o o o o o o o o o o o o 0 Woman . . . . . . . . . . . . . . . . mam mm» ooo.H mam.a noma ems mms . . . . . . . . . . mmm.a mmm.a oHH.H mmo.a mo:.H mmma . . . . . . . . . . m: ms: mmo.m soa.m mam.m sow.m mmm.m moaa ozm.: mmm.m sm>.m cmm . . . . . . mmm.a m=N.H mom . . . . Hmma . . . . . . . . . . . . . . . . . . . . . . . . om H . . . . . . . . . . . . we zmm Hom.m moo.m mo:.m mao.m mmma. mm:.m OOH.m som.z :3N.H . . . . . . . . . . sso.a mow.H omn.m mmma mmfl.m mma.m HH6.H mm» 0mm oma oam.a mmH.m mam.m mmz.m :am.a sma.m smma mom.m msm.a ema.a . . . . . . mm can! mom mmm.a mmm.a nmm.H ommfl oo~.H smH.H mm:.a 2mm . . . . . . . . :mm mam . . . . mmmfi . o o o o o o . o . . . mwm NOHAH MMMQH HF . o o o :mQH . o o o . . o c o o o . o o o 0 o o o o o o o . mmmfi o o a o o o o o o o o o o o o 0 o o a o o o o . Nmmfi .omc .>07 .uoo .oamm .m:¢ >H3h mash mm: .LQ< .Lms upon .cmw .Hmmw muzwfimzcopccsz ooo.a ”mammmogo acmECLm>om ocumuaa .mcucoe an .mmompm Lmv:: ooppm>wo mmoumpod do mmfluwpcmzc awooenu.cfi mqmHHv :oflmpm>flo mo mELom Ham cam .AmHHSo new < COflme nwmwomawv mooumuoa mo mmwpommpmo HH< .wonzwma cam .mmma .mmma .HmoH no ongo wapomm .smm meanwopo COHmLm>HU o: mum: ogmze .copm>oo UOflLmQ on» wcflmso omppm>flo who: wmopo Hamm new meESm mama on» 806% mmOpmuoa mace .Umocxoo on» Ca czozm mummz pmocmamo 0:» mo maopo 6:6 mo UOMLmd wcfiuoxgma mnp op gmmmp mama» 05p mo moon 03p CM mopswwm 039m omn.a sow oms.m . . . . osm.m wwm.H on» mm . . . . .6682 mm G w . . . . . . o . . . MH . . . . SQUD . . m MH . . . . . . o 0 o . . o o o o o .H .m . . . 0 MM 0 o . o . o 0 . o . NWH . . o . 0mm HH2.H osm mm:.m . . . . oms.m Ham.H msm.H mmo . . . . .mmso . o o 0 wanm . o o . mMMNAm WNW mNH o o o o o o .me oz . . . . mm . . . . . . . . . . . . . . . . .w .z . . . . Hm . . . . . . . . . . . . . . . . .m .2 mm o o 00H 0 o o o I a mg“ 0 o o o o o o o Dumbo: . . . . Hmm.H . . . . oao.m mHH HH . . . . . . .ccflz . . oms.m omm.m zmm.m .mam.s mm:.m www.ma.mmo.s Hmm.H omm.m 0228: mmm.a . . mmn.m . . . . mmH mmm.m mo:.m mmm.H . . :mm osmuH . . mam an . . . . Nos.a saw wsm mmm . . . . .oHoo ass Hmm com . . . . was mom 0mm om . . . . .ufiamo moms Nomah Hood 0689 mmcs mmos » 2mmfi smog mmmfi ramma mmma 66666 66322026066236 ooo.~ mmcnwmcfl mo moogc .mnmoz mcwpmxsng a: cam moumum am "cExLMOLQ osmECLomom 262:3 cmpgm>mt mwouwuoa ac mmwpflpcmzv Hm60911.HH mqm<6 202 QEH'.DI14HI‘EI .0 \'r_ mom .mpmv hamefipa mo mmopzom pom .HHH 6666860 .Apmnpo .muflnz .mmOpmpOQ mafia @mmm vmflmwunoo .mpflcs .mmOpmpoov mpmv upoaefl ocm Anmmpm .muwnz .mmOpmuoov Mama uponxm Eopm omBSQEoom 3051 omm1 em 1 m as 3mm mmm :mm m: sma H: mm 1 mmma mam1 ooa1 m 1 pm an me mm: ms Hmz1 me1 mom1 mmm1 moma mmw1 som1 mqfl1 ma 1 H :mm man 0: 1 omm1 1m: 1 ma 1 am 1 :wma am 1 06H1 :m mm mm aw: mam :mm on .606 moo am 1 mmmfi mma1 mm 1 mam on as mum on: com Nma Hm N: mm 1 mmma amH1 NH 1 mm mm :OH mzw mom 0mm mm ma m 1 mm 1 Homa Hma1 ma ma mm mHH as: com mam 2mm Hmm mm 00H coma mm Ho 1 mmm Hmm Hmm 23m own Nam m: ms 1 mm 1 Had mmma mm ma Hum omm mam mod mmm mma mmm1 :mm1 mam1 wmm1 mmmfl mmm1 wfim1 omm msm :mH Ham mm: mwm NH mma an mad smma Hom1 moH1 nmm mwm m mmm mmm m 1 mom Hmm1 mma1 :man mmma mom1 Ha owe om: now mam mmm mmm mmm oma 5m 3 1 mmma mm 1 com mas mm: Nam smfl mmm mm mm o 1 H3H1 Hmm1 :mma mmfi1 mmm1 0mm mmm smm Hmm ms: HAN 2H 1 mm 1 ooa1 wmm1 mmma :mm1 mma1 mom Ham mmm sma an: a: 1 00H on: mod mm mmma mm 1 HmH ma: mm: Ham :MH ooa Hmm1 mmm1 OH 1 mom1 maz1 Hmma .omm .>oz .poo .uamm .w3< masw mcsw mm: .Ld< .Lmz .hoom .cmm pmow mpsmflmzcmhnczz ooc.H «oo-HmOH .mcpcoe an .mHmQOp mmpmpm 60662: ”66666606 mo muhoaxm umZII.NH mdmdb _203 TABLE 13.--Maine certified seed Shipments by months, 1952-66a 1,000 hundredweights 1.... ,,93887:???7.1-1-9831113198911.919113; 3?: 1952 I . .91 1,7651 2,8821 1,586 295’ 253 1953 i 915 1 1,700 2,579 1,816 107 118 110 1959 i 399 3 1,830 3,185‘ 1,356; 165 59 7 71 1955 f 290 i 1,130 3,309 2,019§ 389 298 i 99 1956 1 291 5 1,627 2,292! 1,588: 327 302 j 92 1957 3 189 1 1,308 2,792 1,6632 260 295 E 66 1958 j 299 g 1,092 3,100 1,809! 323 95 § 60 1959 g 219 1,950 2,660 1,8561 950 65 E 38 1960 ' 297 : 1,193 2,556: 2,653 309 99 1 28 1961 1 185 g 1,190 3,978‘ 2,191 639 20 17 1962 i 236 3 1,293 2,513 2,991 662 351 91 1963 296 3 1,278 2,627 3,110 705 62 1 29 19(9 232 1 917 2,987, 2,172 960 2 A 19 1965 , 265 i 1,052 2,199 2,079 190 23 3 17 1966 1 256 E 781 3,697 1’8851 395 22 g 20 8Source: Letter from John F. Boyle, Local Represen— tative, Federal State Market News Service. Dept. of Agriculture, Presque Isle, Marketing Service, U. 8. Maine, November 29, 1967. Consumer and .muosd m>oom new Casuaz mLLOQEH mmozaocHo .muose m>oom 6cm swaps: mpaogEH moUSHOCHo .HHH smuamco CH Co>flw mam mooszom mmmse .momo mupooew amuo och mm mmoasom mEmm mcu anm omHaQEoom _ 1 1 _ 1 1 _ . . . a W ma 1 . . m as m m 1 6 1 mm as 1 am om W mm 666a 3mm 1 smam ma 1 . . M m _ m 1 aa m mm amam mmaw mzam mma mmma as: m 6mm“ ma 1 . . 1 . .m m ” am ” mma mma, m . mm M o3a :mma mm m owafl . .m . . _ . .1 . . n ma 1 sm am _ mm om M m: mmma 6mm 1 so m . .1 . . w . .1 . . 1 w as as 1 am aa a moa mmma .9 com H mm . . .. . . pa m . . Ha W mm 1 no W on mm m mma awma 0 mm m m _ a W . . . .. . . O,a H ma 1 om fl am mm M ow coma 2 oma M amam . .16a . .W . . m H m: w om w ama mmaw on mmma oma _ mman ca 1 a 1 . .w . . m m m: _ smaw ama msam nmmm mmma so: _ mm” cm W . . . .woa m w as scam ms ms _ as smma mm: 1 aam a m . . ha _ . . me ~ mma _ mmm" msa moa. mma mmma mom 1 mma mm m . . . . . . ea oa mm _ mo zaa“ mma mmma mmm 1 moa pa M . . . . pm Mom 1 moa mm ms maw omm amma mma 1 mma m m . . . . . . o . om maa om om . oom mmma o 1 msa m . H . .1 . . m 1 oma o m , 2-1,-1-1 ..... 1.1-... 1.- .- 2 1.. . -, w ., .11-ma a, 131mm .063 .>or .uoo ..Qoo .m:< hazw mes-0“ has .amm .aMw .omm .CMn. 1|! -1111L1 11- 11._ I121 1 M11- ..-1 a. 111. -n.1- . p - a -1 - 11:11. 1 1-- 1 :III-~111IL mem. mUCLamzcmanCJc coo. a P mwoummoa .msucoa ho .amowccu Eons macamfiv amLoB mopmpm dosacn "mnolmpoc o o: c..6aLLoo opanz mo musedsa11.za mauq: lit}. a . .HHV .aocooam vcm .LSOHQ .nopmum Q0 mcflmmmoch msp Cw won: mmopmpOQ mczaocfi no: 06 damn one .smma .ma LmnEw>oz 6cm mama .om hash .oxwvH .omwom .mLSpasoflpwa mo .meo .m .2 .mow>pmm wapQOQmm awowpmwpmum .mmpmno CH :mwofipmwpmpm ampspasoflpw< .muwc3 .m mocmamao Eopm wpwppma cfi Um>wmomp mpmc no mammn map :0 nmafiQEoo mm: magma mcem s35 1 £115 amms 1 91m; ammx 1 moms 11 3m.“- mzmk £2 mmm1m - mmm.m msm.m 1 ms22a mma1a mos1m 1 amm1m amm1m mwma amm.m som1m mcz.m amm.a om:.m mmm1m mm:.m ms:.m :mma m1 omm.m mmm.m amm.: ma21a sam.a mmm1m mma.m am01m mmma 21 mmm1a osa2m mms.m mam.m oma1m oom.m mmm.m :mo.a moma mmm.a mmo.m wam.m msm.a mm01m ms11m oza.m mmm.a .aoma o:m.m :ma.m sm 1m acm1a ma:.a m:o.a mma.a smm.a omma ma21a smm. aom.a nom.a mma.a mmm1a wsa1a mwa.a mama amc.a msm sm32a pmmm p:mo.a paw01a comm noes wmma nmam 9mm: pssm pmas pmmw nmam pans . omzs smoa .mmc .>02- 1wwmw wmmm .LQ¢ .Lmz .pmm .cmh . mew wpnuflmzvmpuzzz ooo.a 1oom .mzpcoE mm .cowmpo nzucfioo Q o a Lsmnamfi CH Unmmooonu cmvH m3aQ ochH c mmmoopQ 0pmp0Q11.ma mg 120 C .,—1 (Y) F 1. <5.» 206 .nucoe o» cpcofi flopm mmflomLSO nomcfi mEom sampcoo awe mcflpmfia esp mmsfip uLOQmL mo Qmaam>o cm 0» mza .mcwmz Cw mpommmoopQ 0pmp0Q nwfipm cocmpm cmmopm me map mo m>fim 0p mmw 1Lm>wamn campOQ mo coauomecfl wcHLm>00 mmocmflmexm co Ummmn mum mums mne .mLSpH30flpm< mo .meo .m .3 nmungpmm mewpmxpmz new LmESmcoo .Lomfl>me3m ampmwmm amxoa .xpflx cm> .D .2 Q0 zmmuLQOQ mSp waOLsu vmvfl>opQ mm: COMQmELOQCw mga .swoa .za LmnEm>oz wcm mama .ma Lmnouoo mcwmz .sopwpmo .mamz mo mumpm .mmspazowpm¢ mo .uQmQ .LOpqumaH wcfimfi>mezm .pmpLOQ ospnmo EOLQ mpmupma Ca ®m>fimomp mpmw mo mamas map so wmaHQEoo mm; magma mzem Q mmw mmm mas asm smm mm mmm mom. mzs mam mmm mmm owma amm 3mm cam «no ssa . . mm osa msm mam mom mom mcma mmm mm: mms mam sma mm mwa mnm ms: 3mm mom can :wma amm amw 0mm mam mm mma mma ac: 0mm 2mm amm mam mmma mom maa omm 0mm maa mm am: smm we own an mm moma om ma 0: om . . m sma a: sm mm mm sm aoca .omc .>o7 .poo .mem .w:< zHSh mash zmz .Lo< .sz .nmm .cmw mew mpswflmzumptcsn coo.a mwmuamma .mcpccE an mewmr :w mewmmmoogQ oompomuu.wa mqmow .m .D ”20pmcflswmzv mmoa .mwmwvaom mwMLOpm caoo Hmcowwmm no NLmEEsm .Uamom mcwupoamm Qopo 2m0fi>pmm wcfipLOQmm amowpmflpmpm .mpspazowgw4 mo .meQ .m .3 Ca Mums Eopm omHHQEoo was magma anew woo.z omm.m :oo.m mza.m msm2m moo.m mmm2m mm:.: :mm.: smm.z mma2: mms1m mmma mqa.m mmm.m amm.m :om.a omo.a smo1a :ms.a o:m.m oma.m amm.m osm.m mms1m mmma smm.m mmm.a mmm2a mmm.a sma.a mmm.a wm31m mma1m mmz.m smo.m amm1m mom1m zoma aam2m smm.a moo.a sam.a mmm.a mmm.a mom1m mw01m mom.m s:o.m moo.m mmm1m mmma mam.m mmm2a mmw.a mam1a mom.a oo©1a maa.m ss:.m mam1m mmm1m wom1m :om.a mmma mmw1a mss.a mmm.a 2mm1a m2m1a sm:.a omm.a moa.m s:m.m sma.m mmm1a msm.a aoma omm1a mao.a oas cam mom smm mmm noo.a oso.a mso.a mom 3am omma mos :mm mom sm: mom awe mow =:O.a zoo.a asm mom mos mmma mmo mam ma: own man om: new omm ass amm mmm smm mmma sam mam mam ma: ma: msm ams :am mos mmw ssm mm: smma om: :nm m:m oma zam msm msm on: mmm mmm som mom mmma .omo .>o: .poo .uQmm .w:« hash mean was .LQ< .umz .pmm .cmh .V mew mpcwamzcmpucsn ooo.a mmwnmmma .mHMuou mmpmpm Umufic: ”mwcficaon mwmQOpm caoo npcoEquuvcw .mmOpmpOQ umHLQ cocmnm cowopmuu.>a mam<9 APPENDIX C FURTHER DETAILS REGARDING THE ESTIMATION OF MONTHLY SALES DATA FOR POTATOES 209 Abbreviations In this appendix, the following abbreviations are used: RS TS TU rail shipments data truck shipments data truck unloads data rail shipments RSD truck shipments TSD truck unloads TUD Standard abbreviations are used for states. Truck Conversion Factors The truck conversion factors used were taken directly from the annual publications and are reproduced below. From January 1950 through December l9f9, the following representative factors were used by the U. S. Dept. of Agriculture to convert truck and boat shipments to rail carlot equivalents. Potatoes All size & type pkgs. in pounds Gulf & Atlantic States—-Texas (Rio Grande Valley only) thru Del. . . . . . 30,000 Mont., Wyo., Nebr., Colo., N. Mex., & West--Year around . . . . . . . . . . 36,000 All other States: October thru May . . . . . . . . . . . . 140,0003 June thru September . . . . . . . . . . 36,000' From January 1960 through December 1965 the following factors have been in effect: 1U. S. Dept. of Agriculture, Consumer and Market- ing Service, Fruit and Vegetable Division, Market News Branch, Fresh Fruit and Vegetable Shipments by Commodi~ ties, States, Months, Calendar Year, 1959 (Washington: April, 19607, pp. 2-3. 210 Potatoes All size & type pkgs.--in pounds 1 MAINE . . . . . . . . . . . . 50,0001 1 ALL OTHER STATFS . . . . . . . . . . . . A3,000 From January 1966, the truck conversion factors are as follows: Potatoes All size & type pkgs.--pounds p Fla. & Long Island . . . . . . . . . . . 50,000 l majne . o a o o o o o o o o o o 0 5590002 I All other States . . . . . . . . . . . . A8,000 Rail Conversion Factors r- As described in Chapter III, rail conversion fac- tors were derived from railroad statistics for eight re- gions. The region borders do not always follow state bor- ders, but each state was classified as belonging to a single region on the basis of two maps: one showing pro- duction areas within states3 and the other showing the region borders.“ Only for a few states (Pa., Ohio, N. Y., £yo., Texas, Mo.) was there some doubt about the classifi- cation. West Va. did not have any RS. Miss. was placed 1Fresh Fruit and Vegetable Shipments . . . Calen- dar year, 1960, p. 3. 2Fresh Fruit and Vegetable Shipments . . . Calen- dar year, 1966, p. 3. 3Kehr, Akeley, and Houghland, 99. cit., p. 10. “Interstate Commerce Commission, Bureau of Trans- port Economics and Statistics, Summary Tables of the Sixty- Second Annual Report on the Statistics of Railwa s in the United States for the Year Ended December 31 l9h8 (Wa" sh- ington: U. S. Government Printing Office, 1950 05, p. 211 in a wrong region because it was treated together with La., but it had practically no RS. The various states were classified as follows: New England Region: Conn., Maine, Mass., N. H., R. I., and Vt. Great Lakes Region:- Mich., and N.Y. Central Eastern Region: Del., I11., Ind., Md., N. J., E Ohio, and Pa. Pocahontas Region: Va. E Southern Region: Ala., F1a., Ga., Ky., N. C., S. C., and Tenn. Northwestern Region: Iowa, Minn., Mont., N. Dak., S. Dak., Wash., and Wis. Central Western Region: Ariz., Calif., Colo., Idaho, Kans., Mo., Nebr., Nev., N. Mex., Oreg., Utah, and Wyo. Southern Region: Ark., La., Miss., Okla., and Texas. When the time series were updated in 1967, the question regarding rail conversion factors had to be taken up to renewed consideration. No quarterly publications from which rail con- version factors can be derived, have been published since l963.1 It seemed inappropriate to extrapolate the data used earlier any longer, and a new approach therefore had to be devised. One possibility was to use yearly railroad data and derive yearly rail 1Letter from M. Paolo, Director, Bureau of Accounts, Interstate Commerce Commission, Washington, November 16, 1967. 212 conversion factors from these.1 Another possibility was to use the earlier mentioned truck conversion fac- tors established from January 1, 1966 by the U. S. Dept. of Agriculture. The latter approach was followed for the calendar years 1965 and 1966 for the crop years l965/66 and 1966/67. Since the truck conversion factors are based on the best judgement of the offi- cials of the U. S. Dept. of Agriculture, as well as on the available railroad statistics, it should be safe to use this method. The use of common conversion factors for rail and truck also provided some computational convenience. Since neither of the two alternatiye_sets of rail conversion factors vary between months, the choice between them is actually not a very important or=. For the earlier years, the rail conversion factors used previously were retained. Disregarded Data When calendar year was not used, certain rather time consuming precautions had to be taken in order to keep the number of cities represented in the unloads data constant for any year. Calendar year was therefore lSuch data were available for 196A and 1965 for the U. S. as a whole, Eastern District, Pacahontas Region, Southern Region, and Western District, See Interstate Commerce Commission, Bureau of Accounts, Ergight ngmggity Statistics of Class I Railroads ifl the United States, 22$“ endar Year 1965 (Washington: U. S. Government Printing Office, 1967), pp. 1-17. 213 chosen as often as it seemed satisfactory. In a few case' when calendar year was used, a very small number of car- loads, reported as shipped in the first months of the cal~ endar year, were disregarded. Such shipments belonged to the crop of the preceding year and were therefore not al- lowed to affect the monthly distribution of the current year's crop. TUD were treated the same way when they were used. In many cases the present writer had a suspicion that the movements in question were reshipments from other states. The problem is not important and is mentioned here only for completeness. Even though TU and TS generally had a similar sea- sonal pattern during the usual marketing period, a few cars of TU were often reported as having occurred long before the beginning or long after the end of the usual marketing season for the state in question. Such TU were disregarded. In many cases they had probably been wrongly classified. The following principle was applied: Unloads occurring as much as one or more months before the begin- ning, or two or more months after the end of the usual marketing period for a state as listed in Agriculture Handbook No. 1271 were disregarded. Agein, the problem is a minor one and is mentioned for completeness. Data on TS were disregarded when they were re- lRoyston, Frost, and Gallaway, op. cit. 21H I‘latw d Efiy llsi;a ()U '?U fk\r IWTSIHDIMT Ukzntxlcniecl 1r! Clwiptxar ‘IJ. 'Further Descrfiguihvizof the Unloads Data and Their Treatment In all cases when they could be obtained, TUD were used instead of TSD when the latter were unavailable or inappropropriate. Individual city reports could not be obtained for all cities, and in some cases the avail- able reports did not contain TUE for potatoes. The city reports used for each calendar year are listed below. When calendar ye.r was not used, data for some cities were 1 d‘isregi—lrled for one part if the year in order to keep the number of cities constant for the whole year. i in the folluuing, the cities for which city re- ports were used are listed. An K under a year indicates that the city report for that year was available and that it contains TUD for potatoes by states and months: 1950 lgrl 19 Atlanta g Baltimore X Birmingham Boston Chicago Cincinnati Cleveland x Dallas X X P“ \C) X >< U"! U“- lr-J \O N><><><><><><><><><><><><><><><><>
    <>< ><><\J“- ><><>< ><>< >< x Fort North Denver X Detroit X X Jackson Kansas City Los Angeles X X X X Louisville Minneapolis-S.Paul New Orleans X X X X New York City X X X X ><><><>< ><><><><><><><>< >< ><><><><><><><><><><>< >¢x x><><>< ><>< s i 215 1950 1951 1959 1953 195” 1955 1956 1957 Oakland X X X X X X X X Philadelphia X X X X X X X X Pittsburg X X X X X X Portland X X X X X St. Louis X X X X X X X X San Francisco X X X X X X Y X Seattle X X X X X X X Washington X X X X X X X X For 1958-59, unload totals for 39 cities were used. The corresponding data for 1960-66 covered “1 cities. Seattle and Tacoma were the two cities added. Since the number of cities covered by the data was much larger For 1958 than for 1957, an adjustment ha: to be made for states where calendar year was not used. Individual city unload summaries were not available for W I. the additional cities for the year 1988. Data for these cities could therefore not be subtracted. Instead the data for l957-58 were adjusted so that the ratio between unloads before January 1 and total unloads after January 1 was the same this year as it was in average for the preceding and the succeeding year. A corresponding method of adjustment was also used in a few cases when TSD were used for one part of a year and TUD for the other. In cases when TUD for 1959/60 were used, data Yer Seattle and Tacoma were subtracted from the TUD for 1960 if 1Saylor, letter, October 6, 196". 1 “v‘_ .. ._ 216 they were available.1 Subtractions should have been made, but data were not available for the following states: ? Conn., Ind., Iowa, Mass., Nebr., Nev., N. J., N. Mex., N. Y., N. C., Ohio, Pa., R. 1., S. D., Utah, Vt., and Wis. It is likely that shipments to Seattle and Tacoma from most of these states were zero or very small, however. Description of the Data and Their Treatment I'or Ind1vidual States Space does not allow a complete description of the availability of various categories of data for indi- vidual states. Some deficiencies in the available data for certain states are mentioned below, however. Correc~ tions for such deficiencies are also described if any corrections were made. The main rule followed was to use all available RSD and TSD. TUD were used instead of TSD when the latter were unavailable or inadequate. In general only the cases when the TSD were considered inadequate for some reason are mentioned below. When a state is not mentioned at all this indicates that the available data (RSD together with TSD or with TUD, or any one of the three alone) were used. lMonthly TU data for Seattle and Tacoma by state of origin for certain states were obtained from the U. 8. Dept. of Agriculture, Agricultural Marketing Service, Fruit and Vegetable division, Market News Branch, Fresh Fruit and Vegetable Unloads in Western Cities by Commodities, States, and Months, Calendar Year 1960 (Washington. March, l96l) p. 123. 217 Alabama: For 195“ and 1955 TSD were not available for the period after June 8 and June 11, respectively. Inspection of the data seemed to reveal similar incomp- leteness for 1952-53. Since in addition no TSD were available for 1951, TUD rather than TSD were used for the whole period 1951-55. Arizona: Ariz. TS are interstate only, but the avail- able data were used. California: Two kinds of truck movements are shown in the shipments tables for Calif. "Passings" represent interstate movement, while "unloads" represent receipts of Calif. production at Los Angeles, Oakland, and San Francisco markets. The two were added in the present analysis. Calif. truck passings for 1959 were incomplete for October, November, and December because one station, Truckee was not reported. To adjust for this, average data for Truckee for the corresponding months in 1961-63 were added to the Calif. data.1 Since the preliminary TSD for 1967 gave the impres- sion of being incomplete, TUD were used instead of TSD for the year 1966/67. Connecticut: The TSD for 1951/52 and 1956/57 were dis- regarded. Comparison with TUD seemed to indicate that they had been reported only for a shorter period. 1Data for Truckee were obtained from Saylor, letter, August 13, 1965. 218 Delaware: TUD were used instead of TSD for 1952-55 and 1957-58. TS seemed to have been reported for only part of the season. More TU than TS were reported for all years. Florida: The TSD for Fla. represent interstate movement and shipments from the peninsula area to West Fla. These data were used. 3 Georgia: Production estimates were discontinued for this state in 1965.1 The figure showing quantity sold from farms in 1965 was therefore used also for 1966, since this was about the best that could be done. Illinois: Very few data were available. The available data on sales, shipments, and unloads were added together with corresponding data for N. J. and W. Va. In all these three states only Late Summer potatoes are grown. Kentucky: The monthly data for Tenn. were used for 1951- 55 since no movements data were available for Ky. In both states only Early Summer potatoes are grown. 1 Louisiana: All available data for Miss. were added to the La. data. Maine: All the available data for N. H. and Vt. except the TUD for Vt. for 1957/58 were added to the Maine data. Maryland: Only TUD were used. All shipments data for lQSl—SB pertained to Eastern shore only. This seemed to be the case for shipments data also for later years. TU lSee Potatoes and Sweetpotatoes: Estimates gy States and Seasonal Groups-~Crops of 1965 and 1936, pp. h-B. 219 in several of the later years were spread over a longer season, more in accordance with the listing of the usual marketing period in Agriculture Handbook No. 127. Massachusetts; The TSD for 1951/52 were disregarded since they seemed to have been collected only during part of the season. TUD for Boston, nearby were available for 1950/51 and 1951/52 but they were not included in the TUD for Mass. Data on boat shipments : plus TU were used for 1953/5A-195A/55. Michigan: The TSD for the first half of 1961/62 were disregarded and TUD were used for the whole year. Minnesota: TUD were used instead of TSD for 1953/SA, 19b2/b3, and 196u/65 since TS were reported for only part of the year. Mississippi: Very few data were available. All the available data were added to the data for La. In both states only Late Spring potatoes are grown. Missouri: TUD for St. Louis, nearby were not added to the Mo. data since these TU might also have been shipped from Ill. Montana:‘ The usual adjustment of TU for 1957/58 was not undertaken, since the ratio of TU in the first part of 1958 to TU in the last part of 1957 was smaller than the average of the corresponding ratios for 1956/57 and 1958/59. New Hampshire: Very few data were available. All 220 available data were added to the correspording data for Maine. In both states only Fall potatoes are grown. N§w_ger;ey: All available data for Ill. and W. Va. were added to the correSponding data for this state. NewflMe§icoz Since very few data were available for f 1950/€1-1955/56 and for 1957/58, monthly averages of shipments and uloads data for all other years were E added to the available monthly data for each of these 6 years. N§w_ggrkz The shipments data for N. Y. are reported separately for N. Y., Long Island and for N. Y., Up- state. Since the data were not complete, it was decided to disregard the distinction. When data eXpressed in carlots had been compiled fcr each part of the state, the data were therefore added together for the whole state. Since no TSD were available for 1950/51- 51/52, TUD for N. Y. as a whole were used for those years. For the remaining years, both RED and TSD were available for N. Y., Long Island, but very few shipments data were obtainable for N. Y., Upstate. It was therefore decided to supplement the data for N. Y., Upstate with TUD. For the years 1952/53— 57/58 a special problem was involved in the use of '221 these data, however. The difficulty was that the reporting practice with regard to N. Y. was different for different cities and years during this period and it was not always quite clear whether "N. Y." meant N. Y” Upstate or N. Y” total. The available data for the various cities were .. Q , . - therefore divided into four groups labeled (1) N. Y., (2) N. Y. Long Island and N. Y. Upstate, (3) N. Y., W Upstate, and (A) N. Y., Unspecified. The latter two kinds of data were added together and used to represent a TUD for N. Y., Upstate in the cases when the correspond- ing TSD were not available or were disregarded. Since the data for both parts of N. Y. were later added to- gether, it is clear that by this procedure some TU from Long Island were counted twice. Since the data were used only as weights, this error need not be particularly se- rious, however. The reason for following this procedure was that it was felt more satisfactory to count some Long Island unloads twice than to disregard all unspecified TUD. It is also likely that the unspecified data really were data for N. Y., Upstate in several cases. A few available TSD for N. Y., Upstate for Janu- ary through April 1955 were disregarded. The data labeled N. Y. Unspecified were disregarded for the last part of 1957 since there was no comparable data available 222 for the first part in 1958. Also the TUD used for N. Y., Upstate in 1963/6“ had the handicap that they might in- ? clude some TU from Long Island. I The TUD used to represent TU from N. Y., Upstate for the years l962/63 and 196H/65 had to be adjusted. The reason for this was that unloads from Long Islands seemed to be included in the TUD for 1963-6“ while they were not included in the data used for 1958-62 and 1965. i The lacking TUD for N. Y., Upstate for January-June 1963 and July-December 196% were constructed as an average of E the figures for the corresponding months in 1960/61 and 1961/62. When the time series were updated in 1967, the data for N. Y., Upstate for the period 1962/63-66/07 were treated in a different way. Comparable TUD data for all of N. Y. were then available for the whole peried begin» ning in 1963. These data were used to represent N. Y., Upstate TU for the years 1963/64-66/67.l TUD for N. Y., Upstate were used for the first half of 1963/63, and TUE for N. I. as a whole were used for the second half. The latter data were adjusted so that the ratio between the totals for the two parts of the year was the same for 1962/63 as it was in average for the years 1961/62 and 63/6“. 1Again, this means that some N. Y., Long Island unloads were counted twice. 223 Geographically, N. Y., Upstate belongs to the Great Lakes Region, while N. Y., Long Island belongs to the New England Region. Because the data were added together, all of the state was treated as belonging to the Great Lakes Region, however. The official truck conversation factors estab- lished from January 1, 1966 were different for the two parts of N. Y. The conversion factor for Long Island was used for the whole state, however. The TSD for N. Y., Long Island for July and August, 196h were recorded in one single figure. This figure was split by distributing it on the two months proportionally to the corresponding TUD used for N. Y., Upstate. North Carolina: For the years 1957-66, TSD were avail- able only for June and July. Since this might be due to closing of the shipping point offices during the rest of the season, the TSD were supplemented with all available TUD for the remaining months of the season. North Dakota: TUD were used instead of TSD for the years 1953/54, 1962/63 and l96h/65, since the TSD seemed to be very incomplete. thg: TSD were available only for 1957. and these data were disregarded. The usual adjustment of the TUD for 1957/58 was not undertaken for the same reason as was mentioned for Mont. 1? 22M Wklahoma: Since very few data were available for the years 1956-66, the data were supplemented by adding to these the available data for Ark. In both states only Late Spring potatoes are grown. 922E223 TUD were used instead of TSD for 1956/57 and 196A/65. TSD were used for the first half and TUD for the second half of 1957/58. These data were therefore adjusted in the way described earlier. Pennsylvania: TUD were used instead of TSD for 1957. Rhode Island: The movement data for Mass. were used as distributive weights for R. I. sales in 1952/5? and 1954/55 since no movement data for R. I. were available. South Dakota: The TUD for 1957/58 were disregarded. They would have had to be adjusted for varying number of cities, but this was little to bother with, since so very few data were available. Tegasz TUD were used instead of TSD for all years. The TSD cover only Lower Valley (Rio Grande Valley). }_-1 pattern of these T8 was very different from that of (D ‘h the TU, which seemed to have originated from all over the state. Utah: The usual adjustment of the TUD for 1957/58 was not undertaken for the same reason as was mentioned for Mont. yormont: Very few data were available. All available 225 data, except the TUD for 1957/58, were added to the cor- rCSponding data for Maine and N. N. In all three states only Fall potatoes are grown. Because they were very small, the TU for 1957/58 were disregarded. They would have had to be adjusted in any case. Washington: Since no TSD were available for 1952, the TSD for 1952/53 were disregarded, and TUD were used instead. West Virginia: The very few data that were available were combined with the corresponding data for N. J. and \ I ...- II ...J m. Eififliflii The TUD for 1957/58 were disregariei. “as recorded TU for this year were so small that there was no reason to keep the data and adjust them. 226 o.m ss ss ms ss es as an as as as m s m a m as asss was: m.o as as ss ms ms N am sm om ms ms mm as as ms em .cms .ez m.ms m mm so ow ss we as ms as as we ow ss me we mm .m:< .s> ecm .m .z .mcsmz m.o sm om om m mm mm m as 2 am 0 ms mm as mm as .cma .mmsz ccw 1mg m.o e m s m m s m m m m s s m s w s .cme .se s.o s m as ms am we mm as as mm ms mm ms m as ms .eme .mcmx m.o ms m as 3 ss as as ss as m m s m m mm s ssse macs m.o as am ms ms m ss s: mm mm m m m s m s m sssn .ecs s.ss mm om cm am mm s: mm ms mm me me me ss so as as ssse oemes s.o am ms mm as mm as mm mm ms as m mm om em as sm .ems .ao m.s mm mm mm mm mm mm sm sm mm mm mm om sm mm mm sm .>oz .msa m.o am ms sm mm mm ms ss as mm mm as am am mm mm sm .cme .smo m.o os ss as as as as ms ms m as m m s m m m ssss .scoo s.s aw N am m as mm am mm am am mm mm am mm mm mm ssse .Osoo m.ss me as me me me ss me we we mm me am ms we as am .msa .essmo s.o mm ms as m s m m s o m m m as s ss ss .ess .ssa m.o mm sossos sos sosmos mos sos so am am sm as as mm mm .ema .sssa o.s sm om mm so as mos mos mm mm os mm ms so mm ms we .cme .asa _ assets/m. m m. m m m m m. m. m m m m m a in. assesseewewaensasaamsswas. cofiuozbosm neswH o om .3 (HO 0H u IWQ WQSOWW coauosoosa mcficcswmb smmw swow.so oomph owmucmosmm -_ r . moswm no mam mucmEoHQm mwmhcmosmm B mm mucmEQch poemssoamo mwmusmms asschmop numb» “meadow mo masosw he so mopmom no new neucoe he .mEsmm Eosu moses mo ammusoosoa ”coflmso>so poooxe mmwoosza saw som moODmuomun.ms mqm.3 sea usage—”oz s.siss os oo so oo oo oo so oo so ss oo oo so os os ossL .>mz o.s os oo so so so oo oo oo so os oo oo oo oo os oo ssss .ssmz o.s so os os so ms so os os os ss os ss so oo ss ooi.oss .scoz o.s so so oo oo so oo so so os os o oo oo o os oo .cos .0: s.s so ss os os os oo oo so so oo oo oo so oo so oo ssss .sssz s.o oo so so oo oo oo so oo so oo so oo so oo oo so ssss .sss: .Amwmesmswm .L I T. T. T. I _L T I .L .L .L T. I .T. .L n ,b ,0 r0 Au ,b ,b ,0 r0 Au ,b so Au Re no no ,b mmlmmmav 9 9 9 9 9 9 9 C._ 9 Cs 9 Cs 5 r: S S mash: mmumum so .= so as us asoso noouosbosas muoccswmn new» swam so endow owmucoosom meson mo ohm.mucmeasnm amoucmohmm I UmSCHuCOUII.wH mqm<8 228 .os .Q ..uoo .mm .ecmsnmsom new .zoaoxd .szmx Eosm UoHoQEoon .o assesses as see >s sea sss seasons as sassssmms Lessons ope COHUmQSQEoo mo moonooe was com: mama .osnmaom>m go: who: mucoaaszm sous» cons pom: osos obsess: sense .onso one sows onssoco mocoeaozmm o.s os osi s ss_ss o o os oo oo oo so os ss so so ossm .oss o.o oo. oom so s soo so oo os so so so so oo so oo ss as: .smsz o oo on so so_oo so oo oo os so ss oo os os ss oo mass .nmoz m.s os oo om os_ms so mo mo os mo mo ss mo, mo os mm .coo .m> Ii. 1* I . I .L TL I “T; T. I T. .L TL TL T TL T. I I nAmw-mcsossm 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 I 9 9 9 9 9 OJ 9 C; C4 9 C4 9 C4 9 Cs (3 No wmmav 9 r: .n. cc 8 T. O 6 oo 1. 9 r: W E 8 I cosssssosd _ s o s mass .m .D so ms .raltali. . -Izit!| xillrt?;:: - -- .Lfiis:;e 11 Isom mmowum .coooosnoso. wCHGCHMoo sees no mo QSOAw owoscoosoms l1!- !- .1. {6:11.14 -- . :ei -, I .fli use .so bosom meson mo ohm mucoEQozw mwmocoosom cmSCHDCOOII.mH mqm<9 BIBLIOGRAPHY s.s £5? .1. ‘1 BIBLIOGRAPHY Boyle, John F., Local Representative, Federal State Mar- ket News Service, Consumer and Marketing Service, U. S. Dept. of Agriculture, Letter, Presque Isle, ’“t Maine, November 2“, 1967. g i .‘1 - Cothern, James Harold. "The Importance and Impact of the 1955 and 1956 Government Potato Diversion Program on the Potato Industry." Unpublished Masters dissertation, Dept. of Agricultural Economics, Michigan State University, 1957. T.M-s.l_ne L._,. ‘ Dalrymple, Dana G. Predicting August Potato Prices at Planting Time. Progress Report 29, Storrs, Connecticut: Storrs Agricultural Experiment Sta- tion and the Agricultural Extension Service, Uni~ versity of Connecticut, February, 1959. Reprin- ted 1962 by the Division of Agricultural Economics Programs, U. S. Dept. of Agriculture. Durbin, J. and Watson, G. S. "Testing for Serial Correla- tion in Least Squares Regression. II," Biometri- ca, XXXVIII (1951). pp. 159-78. Foote, Richard J. Analytical Tools for Studying Demand and Price Structures. Agriculture Handbook No. 1A6, Agricultural Marketing Service, U. 8. Dept. of Agriculture. Washington: U. S. Government Printing Office, August, 1958. Fox, Karl A. Econometric Analysis for Public Policy. Ames, Iowa: The Iowa State College Press, 1958. Friedman, Joan and Foote, Richard J. Computational Meth~ ods for Handling Systems of Simultaneous Equa- tions with Applications to Agriculture. Agricul- ture Handbook No. 9H, Economic and Statistical Analysis Division, Economic Research Service, U. S. Dept. of Agriculture. Washington: U. S. Government Printing Office, November, 1962. Goldberger, Arthur S. Econometric Theory. New York: John Wiley & Sons, Inc., 196“. 230 231 Gray, Roger W.; Sorensen, Vernon 8.; and Cochrane, Nillard W. In Economic Analysis 93 the Impact '9: Government Programs g3 the Potato Industry of 222 United States. North Central Regional Pubii- cation No. H2. University of Minnesota Agricul- tural Experiment Station, June, 195A. Hagamann, Ronald A. "An October Prediction of the Michigan March Price for Potatoes at the Farm Level." Unpublished Master's dissertation, Dept. of Agricultural Economics, Michigan State Univer- sity, 1959. Nee, Olman. Demand and Price Analysis for Potatoes. Technical Bulletin No. 1380, Economic and Statis- tical Analysis Division, Economic Research Ser- vice, U. S. Dept. of Agriculture, Washington: U. S. Government Printing Office, July, 1967. Heflebower, R. B. Factors Relating to the Price 9; Idaho Potatoes. Bulletin 1667 Moscow, Idaho: University of Idaho, June, 1929. Hildreth, Clifford and Lu, John Y. Demand Relations with Autocorrelated Disturbances. Technical Bulletin 276} East Lansing: Dept. of Agricul- tural Economics, Agricultural Experiment Station, Michigan State University, November, 1960. Holcomb, E. J., Acting Chief, Vegetable Branch, Fruit and Vegetable Division, Consumer and Marketing Service, U. S. Dept. of Agriculture. Letter, Washington, September 9, 1965. Letter, Washington, November 9, 1967. Interstate Commerce Commission, Bureau of Accounts. Freight Commodigy Statistics 23 Class I Rail- roads :2 the United States, Calendar Year 1965. Washington: U. S. Government Printing Office, 1967. Interstate Commerce Commission, Bureau of Transport Eco- nomics and Statistics. Frei ht Commodit Statis» the tics gI Class I Railroads In n1 e tafes, Tons of Revenue Freight Carried and Freight'Re- venue_§I Large Class I RaiiroadsTi'IQuarterff' issues, 1950-63g \_ _ :3. H‘ ‘ rm worm? .51.: m f” 232 . Summary Tables of the Sixty- -Second Annual Re- port on the Statistics of Railways in the United States for the Year Ended December 31,19A0. 3 Washington: U. S. Government Printing Office, ; 1950. Johnston, J. Econometric Mgthods. New York: McGraw- Hill Book Company, Inc., 1963. Kehr, August E.; Akeley, Robert V.; and Houghland, Geoffrey V. C. Commercial Potato Production. Agriculture Handbook No. 267, Agricultural Research Service, U. S. Dept. of Agriculture. J Washington: U. S. Government Printing 2 Office, July, 196“. a Malinvaud. E. Statistical Methods 9: Econometrics. p Trans. A. Silvey. 2d rev. ed.; Studies in Mathe- i matical and Managerial Economics; Amsterdam: North-Holland Publishing Company, 1970. Marschak, Jacob. 'Economie Measurement for Policy and Prediction," Studies in Econometric Method. Ed. Wm. C. Hood and TjaIl ing C. Koopmans, Cowles Commission for Research in Economics, Monograph No. I“, New York: John Wiley & Sons, Inc., 1953- Meinken, Kenneth W. Factors that Affect Price and Dis- tribu.tion of New Jersey Potatoes. Bulletin 786, New Brunsw1ck: New Jersey Agricultural Experi- ment Station Rutgers in cooperation with Maine Agricultural Experiment Station, June, 1957. National Potato Council. "U. S. Production, Utilization, and Use of Designated Potato Crop." National Potato News, Vol. IV (January, 1957). National Potato Council. U. S. Production, Utilization, and Use of Designated Crops. [Washingtbn.] National—Potato Council, January, 196u Paolo, M. Director, Bureau of Accounts, Interstate Commerce Commission. Letter, Washington, November 16, 1967. Porter, Cedric, Supervising Inspector, Dept. of Agricul- ture, State of Maine, Letter, Caribou, Maine, October 13, 1965. Letter, Carbou, Maine, November 1“, 1967. 233 Pusateri, Francis P. What Factors Influence the Price of Potatoes? Kern County Potato Growers Association, Spec1al Report. Bakersfield, California: Septem- ber 15, 1958. Royston, Reginald; Frost, Oakley M., and Galloway, Frasier T. Potatoes and Sweetpotatoes: Usual Dates for Plantin,, Harvesting, and Marketing by Seasons, In Pr1nc1pal Areas. Agriculture Hand- book No. 127, Crop Reporting Board, Agricultural Marketing Service, U. S. Dept. of Agriculture. Washington: U. S. Government Printing Office, June, 1957. Ruble, William Lewis. Improving the Computation of SI- multaneous Stochastic Linear Equations Esfimates. Agribultural Economics Report Number 116wand Eco- nometrics Special Report Number 1. East Lansing: Dept. of Agricultural Economics, Michigan State University, October, 1968. Saylor, J. M., In Charge, Transportation Reports, Fruit and Vegetable Division, Agricultural Marketing Service, U. S. Dept. of Agriculture, Letter, Washington, August 13, 1965. Letter, Washington, August 13, 1965. Saylor, J. M., In Charge, Transportation Reports, Fruit and Vegetable Division, Agricultural Marketing Service, U. S. Dept. of Agriculture, Letter, Washington, August 25, 1965. Letter, Washington, October 6, 1965. Letter, Washington, November 9, 1967. Scott, Forrest E., Leader, Marketing Resources and Cost Group, Marketing Economics Division, Economic Re- search Service, U. S. Dept. of Agriculture. Letter, Washington, July I, 1965. Letter, Washington, November 9, I967. Shuffett, D. Milton. The Demand and Price Structure for Selected Vegetables. Technical Bulletin No. 1105, Agricultural Marketing Service, U. S. Dept. of Agriculture. Washington: U. S. Government Prin— ting Office, December, 195“. “if! 23A Simmons, Will M. An Economic Study RE the U. S. Potato Industry. Agricultural Economic Report No. b. Economic and Statistical Analysis Division, Eco~ 1 nomic Research Service, U. S. Dept. of Agricul- 1 ture. Washington : March, 1962. Simon, Martin S. "Forecasting Potato Prices," 1961 American Potato Yearbook. Ed., C. Stedman I lac- farland, Jr. New Jersey: By the editor. Talburt, William F. and Smith, Ora. Potato Processing. 2d. ed.; Westport, Connecticut: The Avi Publi- shing Company, Inc., 1967. Thirteenth National Potato Utilization Conference. Proceedings. Riverhead, New York, 1963. TWET? Tim ‘1 Ulrey, Ivon W Fresh Potato Transportation to Large Mar~ kets from Five Major Produ01ng Areas. Marketin. Research Report No. 687, Marketing Economics Division, Economic Research Service, U. S. Dept. of Agriculture. Washington, November, 196A. U. S. Dept. of Agriculture. Potato Preferences Among Household Consumers. Miscellaneous Publication Number 667. Washington: August, 19MB. Potato Charts and Table: Washington 1965. U. S. Dept. of Agriculture, Agricultural Marketing Ser— vice, Crop Reporting Board. Agricultural Prices. [Monthly issues, 1957-59. . Agricultural Prices, January, 1957, Supplement_ No. 2, Potatoes. Monthly and Season Average Pri ces Received by Farmers, by States and United Jtates, 19H9~ 5:5. . Potatoes: Estimates in Hundredweight by States 1866-1953. Statistical Bulletin No. 251. Washington: U. S. Government Printing Office, June, 1959. U. S. Dept. of Agriculture, Agricultural Marketing Ser- vice, Fruit and Vege able Division. Unloads 2i Fresh Fruits and Vegetables. Annual is sues for individual cities for 1950-57. U) 03 C0 235 Dept. of Agriculture, Agricultural Marketing Ser- vice, Fruit and Vegetable Division, Market News g Branch, Fresh Fruit and Vegetable Unloads in ' Western Cities by Commodities, States and Months, Calendar Year 1560. Washington; March, 1961. Dept. of Agriculture, Agricultural Marketing Ser- vice, Marketing Research Division. Farm—Retail Spreads for Food Products. Miscellaneous Publi- cation No. 7&1, Washington: U. S. Government Printing Office, November, 1957. j Dept. of Agriculture, Consumer and Marketing Ser- . vice, Fruit and Vegetable Division, Market News 3 Branch. Fresh Fruit and Vegetable Shipments by Commodities, States, Months. Annual issues for 1950-66.} r' Fresh Fruit and Vegetable Unload Totals for El Cities. Dept. of Agriculture, Economic Research Service, Economic and Statistical Analysis Division, Food Consumption and Utilization Section. U. S. Food Consumption: Sources of Data and Trends, 1909- 63. Washington, June, 1963. Dept. of Agriculture, Statistical Reporting Service. Statistical Reporting Service of the g. S. Depart- ment of Agriculture: Scope, Methods. Miscella- neous Publication No. 957. Washington: U. S. Government Printing Office, December, 1964. . Disposition of 1963 Potato Crop and Acreage for 196H. June, 1955. Dept. of Agriculture, Statistical Reporting Service, Crop Reporting Board. Agricultural Prices. . . . Annual Summary. Washington. TAnnual summaries 1959-6o.) Irish Potatoes: Utilization of . . . Crop with Comparisons. Washington. 'Annual issues for the crops of 1960-66.1 Potatoes, Sweetpotatoes, by States and Season- al Groups Crops of 195U-1959. Statistical Bulle— tin No. 291. Washington, August, 1961. 236 Summary bf Regional Cold Storage Holdings . . . Washington: U. S. Government Printing‘ . Office. LAnnual issues for 1956-66.1 Potatoes and Sweetpotatoes: Estimates by §E2’ tes and Seasonal Groups--Crbb§ b: . . . . Washington. {Annual issues for crops of the current and preceeding year, 1961-66.] Potatoes, Sweetpotatoes: by States and Season- a1 Groups--Crops of 1959-6“. Statistical Bulle- tin No. A09. Washington: U. S. Government Prin- ting Office, July, 1967. . Crop Production, United States Crop Summar .as .23 OCtober l, 1965. Washington, October TI, i9bi. . Crop Production, 1967 Annual Summary, Acreage, Yield, Production, by States. Washington: . . Government Printing Office, December 19, 1967. U. S. Dept. of Commerce, Bureau of Census. United States Imports 93 Merchandise for Consumption: Commod- bby by Countr bf Ori in. Washington: U. S. overnment rinting fice. ‘Monthly issues, January, 1951-{January,l96h.] . b. S. Imports 23 Merchandise for Consumption. Washington: U. S. Government Printing Office. [Monthly issues for February, 196A, through December, 1966.1 . U. S. Exports: Commodiby by Country. Washing- ton: U. S. GOvernment Printing ice. {Monthly issues for the period 1951-66.] Current Population Reports: ‘Population‘Esti- mates. Washington: U. S. Government Printing Off1ce, July 16, 1965. Current Population Reports:‘ Population Esti- mates. Washington: U. S. Government Printing Office, August 21, 1967. U. S. Dept. of Commerce, Office of Business Economics, Business Statistics, . . . Biennal Edition, A Supplement bb the Survey bS Current Business. Washington: U. 8. Government Printing Office. {Egitions for 1955. 1957. 1959, 1961, 1963 and 19 5-3 237 Survey bf Current Business. [Monthly issues, July, 1961; July, 196“; September, 1965; and July, 1967. ] U. S. Dept. of Labor, Bureau of Labor Statistics. Month- ly Labor Review. [Monthly issues, 1951-67. J Waugh, Frederick V. Forecasting Prices b: New Jersey White Potatoes and Sweet Potatoes, Circular No. 78, Trenton, New Jersey: State of New Jersey Dept. of Agriculture, July, 192“. White, Clarence E., Agricultural Statistician In Charge, Statistical Reporting Service, U. S. Dept. of Agriculture, Letter, Boise, Idaho, July 20, 1965. Letter, Boise, Idaho, September 28, 1965. Letter, Boise, Idaho, November 15, 1967. Working, Holbrook. Factors Affecting the Price bf Minnesota Potatoes. Technical Bulletin 29. St. Paul: University of Minnesota Agricultural Experiment Station, October, 1925. Zusman, Pinhas. "Econometric Analysis of the Market for California Early Potatoes," Hilgardia, Vol. XXXIII [December, 1962 3", G ‘ _ ‘ A ‘.' “NY." S‘R .‘3‘3 ‘u'l