|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 INDUSTRY WIDE ADOPTION OF MECHANICAL HARVESTERS BY THE FLORIDA CITRUS INDUSTRY: COORDINATION ISSUES AND ECONOMIC TRADE OFFS By JACOB W. SEARCY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA I N PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011
2 2011 Jacob W. Searcy
3 To my grandparents, C.C. and Marg a ret Searcy and Herb and Muncy Chapman
4 ACKNOWLEDGMENTS I would first li ke to thank my dissertation c ommittee chair Dr. Fritz Roka for his total support and encouragement throughout this project It has been an honor to work with such a skilled economist and an even better person. Special thanks also go out to my committee co chair, Dr Tom Spreen, for sharing his wealth of knowledge of the citrus industry, along with some pretty entertaining stories. Additionally, I o we great thanks to committee members Dr. Gary Fairchild, Dr. Tom Bur ks, and Dr. Mark Brown It was extrem ely encouraging to see the effort and care put forth by all of my committee members to advance this research endeavor. I would like to express my thanks to all the fine folks involved in the Florida citrus industry who gave freely of their time. Special thanks to Dr. John VanSickle, Dr. Rene Goodrich Schneider Dr. Fred Davies Dr. Gene Albrigo Alan Morris, Bob Terry, Fran Becker, Jim Zellner, Mitch Willis, Bob Behr, Dave Crumbly, Jim Shuford, Will Elliott, Mike Murphy, Tom Visser, Mike Jones, Paul Mead or, Charlie Lucus, and Wade Timpner Finally, I would like to thank my family and friends for their support, advice, prayers, and the occasional loan The most sincere thanks to m y wife Jennison and son Isaac who were my motivation for completing this di ssertation. I am truly grateful for all the love and support you have shown me. Special thanks go out to my Mom and Dad Jill and Charles Searcy, who ever so gently encouraged the completion of this project. I a m greatly appreciative of my other parents Richard and Phyllis Kipp for their moral support as well as editing assistance I would also like to thank my grandparents, C.C. and Marg a ret Searcy and Herb and Muncy Chapman, for their unwavering support of my academic pursui ts. It is quite a luxury to have grandparents who are willing and able to provide professional quality editing services and a heavenly pound cake.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 8 LIST OF FIGURES ................................ ................................ ................................ ....................... 10 ABSTRACT ................................ ................................ ................................ ................................ ... 17 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 19 Problem Statement ................................ ................................ ................................ .................. 21 Research Question ................................ ................................ ................................ .................. 24 B ehavioral Nature of the Problem ................................ ................................ .......................... 25 Research Goals ................................ ................................ ................................ ....................... 27 Research Objectives ................................ ................................ ................................ ................ 30 Importance of the Study ................................ ................................ ................................ .......... 30 Research Scope ................................ ................................ ................................ ....................... 32 Dissertation Format ................................ ................................ ................................ ................ 32 2 THE FLORIDA ORANGE JUICE PRODUCTION INDUSTRY ................................ ......... 35 Industry Operations ................................ ................................ ................................ ................ 36 Fruit Production ................................ ................................ ................................ ............... 36 Harvesting ................................ ................................ ................................ ........................ 37 Hand harvest ................................ ................................ ................................ ............. 37 Mechanical harvesting systems ................................ ................................ ................ 39 Processing Operations ................................ ................................ ................................ ..... 42 Fruit procurement ................................ ................................ ................................ ..... 43 Procurement strategies ................................ ................................ ............................. 44 Payment ................................ ................................ ................................ .................... 46 Scheduling and load allocations ................................ ................................ ............... 48 Juice extraction ................................ ................................ ................................ ......... 51 Storage ................................ ................................ ................................ ...................... 55 Potential Changes to Industry Coordination and Operation ................................ ................... 58 Cha nges in Fruit Harvest Volume ................................ ................................ ................... 58 Harvest Season Length ................................ ................................ ................................ .... 60 Labor Supply and Demand ................................ ................................ .............................. 63 Grower Costs and Operations ................................ ................................ .......................... 64 Load Allocations ................................ ................................ ................................ ............. 66 Processing Costs and Operations ................................ ................................ ..................... 67
6 Capacity Constraints ................................ ................................ ................................ ........ 69 Late Harvest Limitations ................................ ................................ ................................ 72 Summary ................................ ................................ ................................ ................................ 73 3 CONCEPTUAL AND THEORETICAL FRAMEWORK: AN APPROACH TO ANALYZING THE ECONOMIC IMPACTS OF ADOPTING MECHANICAL CITRUS HARVESTERS ................................ ................................ ................................ ....... 87 Introduction ................................ ................................ ................................ ............................. 87 New Technology Adoption ................................ ................................ ................................ ..... 88 Operations Research ................................ ................................ ................................ ............... 89 Mathematical Progr amming Model ................................ ................................ ........................ 92 Explanatory Variables ................................ ................................ ................................ ............ 97 Optimality and Efficiency ................................ ................................ ................................ ..... 105 Summary ................................ ................................ ................................ ............................... 107 4 BIOLOGICAL CHARACTERISTICS OF FLORIDA ORANGES ................................ .... 110 Biological Characteristics ................................ ................................ ................................ ..... 111 Biological Modeling ................................ ................................ ................................ ............. 119 Fruit Maturity Data ................................ ................................ ................................ ............... 120 Summary ................................ ................................ ................................ ............................... 124 5 THEORETICAL MODELING: SCHEDULING OF THE ORANGE HARVEST TO OPTIMIZE RETURNS ................................ ................................ ................................ ........ 135 Intensity and Duration ................................ ................................ ................................ .......... 136 Cost Impacts ................................ ................................ ................................ .................. 138 Total Pound Solids Captured ................................ ................................ ......................... 139 Total Industry Returns ................................ ................................ ................................ .......... 142 Grower Returns ................................ ................................ ................................ ............. 142 Processor Returns ................................ ................................ ................................ .......... 146 Industry Returns ................................ ................................ ................................ ............ 151 Summary ................................ ................................ ................................ ............................... 153 6 EMPIRICAL MODELING AND RESULTS ................................ ................................ ...... 157 Grower Decision Model ................................ ................................ ................................ ....... 157 Processor Decision Model ................................ ................................ ................................ .... 160 Industry Decision Model ................................ ................................ ................................ ...... 166 Description of Data ................................ ................................ ................................ ............... 170 Processing Data ................................ ................................ ................................ ............. 170 Harvesting Data ................................ ................................ ................................ ............. 174 Bearing Acreage ................................ ................................ ................................ ............ 176 GAMS Modeling ................................ ................................ ................................ .................. 176 Constrained Industry Production Models Assuming 2007 08 Processing Volumes ..... 177 CIPM1 Results ................................ ................................ ................................ ............... 180 CIPM2 Results ................................ ................................ ................................ ............... 182
7 Constrained Industry Production Models Assuming a Single Fixed Processing M aximum ................................ ................................ ................................ ................... 183 CIPM3 Results ................................ ................................ ................................ ............... 185 CIPM4 Results ................................ ................................ ................................ ............... 187 Summary ................................ ................................ ................................ ............................... 191 7 CONCLUSIONS AND FUTURE WORK ................................ ................................ ........... 221 Implications of Mechanical Harvesting ................................ ................................ ................ 222 Possible Industry Changes to Expedite Mechanical Harvester Adoption ............................ 224 Future Research ................................ ................................ ................................ .................... 228 APPENDIX A IN GROVE FRUIT MAT URITY MODELING ................................ ................................ .. 230 B SAS PROGRAMMING CODE ................................ ................................ ............................ 239 C BIOLOGICAL FRUIT AND DROP DATA ................................ ................................ ........ 247 D GAMS PROGRAMMING CODE ................................ ................................ ....................... 300 LIST OF REFERENCES ................................ ................................ ................................ ............. 307 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 315
8 LIST OF TABLES Table page 2 1 Estimated average picking, roadsiding, and hauling charges for Florida citrus 2003 04 through 2005 06. ................................ ................................ ................................ ........... 83 2 2 Largest Florida orange juice processors operating during the 2003 04 season. ................ 84 2 3 Minimum maturity standards for Florida oranges for processing. ................................ .... 84 2 4 Minimum Brix/Acid Ratio standards, as related to total soluble solids; minimum ratios of solids to acid. ................................ ................................ ................................ ....... 85 2 5 Minimum maturity standards for Florida processed juice and concentrate. ...................... 86 2 6 Certain orange juice: reported U.S. production capacity, production, and capacity utilization, crop years 2001/02 2004/05. ................................ ................................ ........... 86 4 1 Mean and standard variation for each biological characteristic by fruit variety and region, including the number of samples (N). ................................ ................................ 132 4 2 Esti mated parameter results for pound solids production per acre. ................................ 133 4 3 Optimal week of harvest to maximize pound solid production, weeks from Aug. 1. ..... 134 4 4 Designated week numbers with corresponding calendar date. ................................ ........ 134 6 1 Definition of variables for grower, processor, and industry models. ............................... 198 6 2 Price and cost estimates for fruit inputs, processing, and storages. ................................ 200 6 3 Bearing orange tree acreage in Florida for 2004 and estimated for 2008. ....................... 201 6 4 Total boxes of oranges processed in Florida during the 2007 08 season. ....................... 203 6 5 Summary results for all constrained indus try production models. ................................ .. 204 6 6 CIPM1, Processing results for constrained industry production model with 95% hand harvesting. ................................ ................................ ................................ ........................ 205 6 7 CIPM2, Processing results for constrained industry production model with 95% mechanical harvesting. ................................ ................................ ................................ ..... 207 6 8 Harvesting results for CIPM1 and CIPM2, constrained industry production models with 95% hand harvesting and 95% mechanical harvesting, with harvest date in weeks from August 1 and portion harvested weekly in parenthesis. ............................... 209
9 6 9 Estimated weekly volume in storage for CIP M1 and CIPM2, NFC reported as pound solids and FCOJ reported as pound solids and gallons at 42 degree Brix. ...................... 211 6 10 CIPM3, processing results for constrained industry production model with 95 % hand harvesting. ................................ ................................ ................................ ........................ 213 6 11 CIPM4, processing results for constrained industry production model with 95% mechanical harvesting. ................................ ................................ ................................ ..... 215 6 12 Harvesting results for CIPM3 and CIPM4, constrained industry production models with 95% hand harvesting and 95% mechanical harvesting, with harvest date in weeks from August 1 and portion harvested weekly in parenthesis. ............................... 2 17 6 13 Estimated weekly volume in storage for CIPM3 and CIPM4, NFC reported as pound solids and FCOJ reported as pound solids and gallons at 42 degree Brix. ...................... 219 A 1 Drop, regression parameter results. ................................ ................................ ................. 230 A 2 Fruit weight, regression parameter results. ................................ ................................ ...... 231 A 3 Pound solids per bo x, regression parameter results. ................................ ........................ 232 A 4 Brix to acid ratio, regression parameter results. ................................ .............................. 233 A 5 Total soluble solids, regress ion parameter results. ................................ .......................... 234 A 6 Average pieces per tree (PPT), average tree density (DEN), and bearing acreage (A) point estimate parameter results. ................................ ................................ ...................... 235 A 7 Boxes per acre, regression parameter results. ................................ ................................ .. 237
10 LIST OF FIGURES Figure page 1 1 Hypothetical annual harvest, process ing, and utilization schedules. ................................ 34 2 1 Fresh and processed orange utilization in Florida from the 1988 89 season to the 2007 08 s eason, boxes ................................ ................................ ................................ ...... 74 2 2 Florida citrus industry processing intensity for 2005 06; monthly processing as a percentage of total season p roduction ................................ ................................ ............... 74 2 3 Flow of fruit and juice movement. ................................ ................................ ..................... 75 2 4 Pick and roadside harvest rates for Valencias from a sample harvesting company for 2 003 04 season ................................ ................................ ................................ .................. 76 2 5 Process flow diagram of ora nge juice production ................................ ............................. 77 2 6 Steps in the production of canned frozen concentrat ed orange juice ................................ 78 2 7 Chilled orange juice expressed in 'standard gallons' (11.8 degree Brix ) for 2007 ............ 79 2 8 Chilled juice, net annual production from fruit, 2000 01 to 2005 06 ............................... 79 2 9 FCOJ, Net annual production from fruit, 2000 01 to 200 5 06, gallons at 48 degree Brix ................................ ................................ ................................ ................................ ... 80 2 10 FCOJ production and storage, 2000 01 through 200 5 06, gallons at 48 degree Brix ...... 80 2 11 Chilled juice, gallons on ha nd from 2000 01 through 2005 06 ................................ ........ 81 2 12 FCOJ, gallons on hand from 2000 01 through 200 5 06 gallons at 48 degree Brix ......... 81 2 13 Chilled juice production and s torage, 2000 01 through 2005 06 ................................ ..... 82 3 1 Four cases f or storag e and inventory ................................ ................................ .............. 109 3 2 Kaldor compens ation criteria. ................................ ................................ .......................... 109 4 1 Total soluble solid percentage concentration for Valencia orange s sampled from 1935 1938 ................................ ................................ ................................ ..................... 126 4 2 Acid percentage concentration for Valencia orange s, sampled from 1935 1938 ........ 126 4 3 Juice weight per box for Valencia oranges, sampled fr om 1935 1938 ........................ 127 4 4 Brix to acid ratio for Valencia orange s, sampled from 1935 1938 .............................. 127
11 4 5 Pound solids change for all Valencia oranges averages from 1935 1938 ................... 128 4 6 Percent fruit droppage of tagged fruit in 1937 1938, assum ed to be monthly drop rates ................................ ................................ ................................ ................................ 128 4 7 Florida juice orange harvesting seasons by month, fro m August through July .............. 129 4 8 Florida commercial citrus production areas as defined by the Florida Agricultural Statis tics Service ................................ ................................ ................................ ............. 130 4 9 Pound solids per acre during the harvest season, late fruit by age group in the central region, week s from Aug 1 ................................ ................................ ............................... 131 4 10 Pound solids per acre during the harvest season, early, mid, and late varieties, 14 to 23 year old trees in central region, week s from Aug 1 ................................ ................... 131 5 1 Expected marginal and average unit cost curves for processing in a typical orange juice processing plant as a function of quantity. ................................ .............................. 154 5 2 Average cost curve for juice or concentrate storage in a typical orange juice processing plant as a function of quantity. ................................ ................................ ....... 154 5 3 Grower returns as a function of total pound solids. ................................ ......................... 155 5 4 Processor returns as a function of total pound solids. ................................ ...................... 155 5 5 Total industry returns as a function of total pound solids. ................................ ............... 156 6 1 CIPM1 weekly pound solid production for NFC and FCOJ. ................................ ........... 192 6 2 CIPM2 weekly pound solid production for NFC and FCOJ. ................................ ........... 192 6 3 NFC weekly storage volume estimates and weekly average from 2000 2006, reported in pound solids. ................................ ................................ ................................ .. 193 6 4 FCOJ weekly storage volume estimates and weekly average from 2000 2006, reported in gallons at 42.0 degree Brix. ................................ ................................ ........... 193 6 5 CIPM3 weekly pound solid production for NFC and FCOJ. ................................ ........... 194 6 6 CIPM4 weekly pound solid production for NFC and FCOJ. ................................ ........... 194 6 7 CIPM1 and CIPM3 weekly total box production. ................................ ........................... 195 6 8 CIPM1 and C IPM3 weekly total pound solid production. ................................ .............. 195 6 9 CIPM2 and CIPM4 weekly total box production. ................................ ........................... 196 6 10 CIPM2 and CIPM4 weekly t otal pound solid production. ................................ .............. 196
12 6 11 CIPM1 and CIPM4 weekly total box production. ................................ ........................... 197 6 12 CIPM1 and CIPM4 weekly total pound s olid production. ................................ .............. 197 C l Pound solids per box for early fruit sampled in the Indian River region. ........................ 247 C 2 Acid percentage for ea rly fruit sampled in the Indian River region. ............................... 247 C 3 Brix to acid ratio for early fruit sampled in the Indian River region. .............................. 248 C 4 Total soluble solid percentage for early fruit sampled in the Indian River region. ......... 248 C 5 Fruit weight for 15 piece sample of early fruit in the Indian River region. ..................... 249 C 6 Juice weight for 15 piece sample of early fruit in the Indian River region. .................... 249 C 7 Pound solids per box for early fruit sampled in the N ort h region. ................................ .. 250 C 8 Acid percentage for early fruit sampled in the N orth region. ................................ .......... 250 C 9 Brix to acid ratio for early fruit sample d in the N orth region. ................................ ......... 251 C 10 Total soluble solid percentage for early fruit sampled in the N orth region. .................... 251 C 11 Fruit weigh t for 15 piece sample of early fruit in t he N orth region. ................................ 252 C 12 Juice weight for 15 piec e sample of early fruit in the N orth region. ............................... 252 C 13 Pound solids per box f or early fruit sampled in the C entral region. ................................ 253 C 14 Acid percentage for early fruit sampled in the C entral region. ................................ ........ 253 C 15 Brix to acid ratio for early fruit sampled in the C entral region. ................................ ...... 254 C 16 Total soluble solid percentage for early fruit sampled in the C entral region. .................. 254 C 17 Fruit weight for 15 piec e sample of early fruit in the C entral region. ............................. 255 C 18 Juice weight for 15 piec e sample of earl y fruit in the C entral region. ............................. 255 C 19 Pound solids per box for early fruit sampled in the W est region. ................................ .... 256 C 20 Acid percentag e for early fruit sampled in the W est region. ................................ ........... 256 C 21 Brix to acid ratio for early fruit sampled in the W est region. ................................ .......... 257 C 22 T otal soluble solid percentage for early fruit sampled in the W est region. ..................... 257 C 23 Fruit weight for 15 piec e sample of early fruit in the W est region. ................................ 258
13 C 24 Juice weight for 15 piec e sample of early fruit in the W est region. ................................ 258 C 25 Pound solids per box for early fruit sampled in the S outh region. ................................ .. 259 C 26 Acid percentage for early fruit sampled in the S outh region. ................................ .......... 259 C 27 Brix to acid ratio for early fruit sampled in the S outh region. ................................ ......... 260 C 28 Total soluble solid percentage for early fruit sampled in the S outh region. .................... 260 C 29 Fruit weight for 15 piec e sample of early fruit in the S outh region. ................................ 261 C 30 Juice weight for 15 piec e sample of early fruit in the S outh region. ............................... 261 C 3l Pound so lids per box for mid season fruit sampled in the Indian River region. ............. 262 C 32 Acid percentage for mid season fruit sampled in the Indian River region. ..................... 262 C 33 Brix to acid ratio for mid season fruit sampled in the Indian River region. .................... 263 C 34 Total soluble solid percentage for mid season fruit sampled in the Indian River region. ................................ ................................ ................................ .............................. 263 C 35 Fruit weight for 15 piece sample of mid season fruit in the Indian River region. ........... 264 C 36 Juice weight fo r 15 piece sample of mid season fruit in the Indian River region. .......... 264 C 37 Pound solids per box for m id season fruit sampled in the N orth region. ........................ 265 C 38 Acid percentage for m id season fruit sampled in the N orth region. ................................ 265 C 39 Brix to acid ratio for m id season fruit sampled in the N orth region. ............................... 2 66 C 40 Total soluble solid percentage for m id season fruit sampled in the N orth region. .......... 266 C 41 Fruit weight for 15 piece sam ple of mid season fru it in the N orth region. ..................... 267 C 42 Juice weight for 15 piece sam ple of mid season fruit in the N orth region. ..................... 267 C 43 Pound solids per box for m id season fruit sampled in the C entral region. ...................... 268 C 44 Acid percentage for m id season fruit sampled in the C entral region. ............................. 268 C 45 Brix to acid ratio for m id season fruit sampled in the C entral region. ............................ 269 C 46 Total soluble solid percentage for m id season fruit sampled in the C entral region. ....... 269 C 47 Fruit weight for 15 piece sam ple of mid season fruit in the C entral region. ................... 270 C 48 Juice weight for 15 piece sample of mid seas on f ruit in the C entral region. ................... 270
14 C 49 Pound solids per box for m id season fruit sampled in the W est region. ......................... 271 C 50 Acid per centage for m id season fruit sampled in the W est region. ................................ 271 C 51 Brix to acid ratio for m id season fruit sampled in the W est region. ................................ 272 C 52 Total soluble solid percentage for m id season fruit sampled in the W est region. ........... 272 C 53 Fruit weight for 15 piece sam ple of mid season fruit in the W est region. ....................... 273 C 54 Juice weight for 15 piece sam ple of mid season fruit in the W est region. ...................... 273 C 55 Pound solids per box for mid season fruit sampled in the S outh region. ........................ 274 C 56 Acid percentage for m id season fruit sampled in the S outh region. ................................ 274 C 57 Brix to acid ratio for m i d season fruit sampled in the S outh region. ............................... 275 C 58 Total soluble solid percentage for m id season fruit sampled in the S outh region. .......... 275 C 59 Fruit weight for 15 piece sam ple of mid season fruit in the S outh region. ..................... 276 C 60 Juice weight for 15 piece sam ple of mid season fruit in the S outh region. ..................... 276 C 6l Pound solids per box for late season fruit sampled in the Indian River region. .............. 277 C 62 Acid percentage for late season fruit sampled in th e Indian River region. ...................... 277 C 63 Brix to acid ratio for late season fruit sampled in the Indian River region. ..................... 278 C 64 Total solu ble solid percentage for late season fruit sampled in the Indian River region. ................................ ................................ ................................ .............................. 278 C 65 Fruit weight for 15 piece sample of late season fruit in the Indian River region. ........... 279 C 66 Juice weight for 15 piece sample of late season fruit in the Indian River region. ........... 279 C 67 Pound solids per box for early frui t sampled in th e N orth region. ................................ .. 280 C 68 Acid percentage for early fruit sampled in the N orth region. ................................ .......... 280 C 69 Brix to acid ratio for early frui t sampled in the N orth region. ................................ ......... 281 C 70 Total soluble solid percentage for early fruit sampled in the N orth region .................... 281 C 71 Fru it weight for 15 piece samp le of late season fruit in the N orth region. ...................... 282 C 72 Juice weight for 15 piece samp le of late season fruit in the N orth region. ...................... 282 C 73 Pound solids per box for la te season fruit sampled in the C entral region. ....................... 283
15 C 74 Acid percentage for late season fruit sampled in th e C entral region. .............................. 283 C 75 Brix to acid ratio for la te season fruit sampled in the C entral region. ............................. 284 C 76 Total soluble solid percentage for la te s eason fruit sampled in the C entral region. ........ 284 C 77 Fruit weight for 15 piece samp le of late season fruit in the C entral region. .................... 285 C 78 Juice weight for 15 piece samp le of late season fruit in the C entral region. ................... 285 C 79 Pound solids per box for la te season fruit sampled in the W est region. .......................... 286 C 80 Acid percentage for la te season fruit sampled in the W est region. ................................ .. 286 C 81 Brix to acid ratio for late season fruit sampled in the W est region. ................................ 287 C 82 Total soluble solid percentage for la te season fruit sampled in the W est region. ............ 287 C 83 Fruit weight for 15 piece sam p le of late season fruit in the W est region. ....................... 288 C 84 Juice weight for 15 piece samp le of late season fruit in the W est region. ....................... 288 C 85 Pound solids per box for la te season fruit sampled in the S outh region. ......................... 289 C 86 Acid percentage for la te season fruit sampled in the S outh region. ................................ 289 C 87 Brix to acid ratio for la te season fruit sampled in the S outh region. ............................... 290 C 88 Total soluble solid percentage for late season fruit sampled in the S outh r egion. ........... 290 C 89 Fruit weight for 15 piece samp le of late season fruit in the S outh region. ...................... 291 C 90 Juice weight for 15 piece sa mp le of late season fruit in the S outh region. ...................... 291 C 91 Percentage fruit drop for early fruit in the Indian River region. ................................ ...... 292 C 92 Percentage fru it drop for early fruit in the N orth region. ................................ ................. 292 C 93 Percentage fru it drop for early fruit in the C entral region. ................................ .............. 293 C 94 Percentage fru it drop for early fruit in the W est region. ................................ .................. 293 C 95 Percentage fru it drop for early fruit in the S outh region. ................................ ................. 294 C 96 Percentage fruit drop for mid season fruit in the Indian River region. ............................ 294 C 97 Percentage fruit drop for mid season fruit in th e N orth region. ................................ ...... 295 C 98 Percentage fruit dr op for mid season fruit in the C entral region. ................................ .... 295
16 C 99 Percentage fruit drop for mid season fruit in the W est region. ................................ ........ 296 C 100 Percentage fruit dr op for mid season fruit in the S outh region. ................................ ...... 296 C 101 Percentage fruit drop for late fruit in the Indian Ri ver region. ................................ ........ 297 C 102 Percentage fr uit drop for late fruit in the N orth region. ................................ ................... 297 C 103 Percentage fr uit drop for late fruit i n the C entral region. ................................ ................ 298 C 104 Percentage fr uit drop for late fruit in the W est region. ................................ .................... 298 C 105 Percentage fr uit drop for lat e fruit in the S outh region. ................................ ................... 299
17 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy IND USTRY WIDE ADOPTION OF MECHANICAL HARVESTERS BY THE FLORIDA CITRUS INDUSTRY: COORDINATION ISSUES AND ECONOMIC TRADE OFFS By Jacob W. Searcy December 2011 Chair: Fritz M. Roka Cochair: Thomas H. Spreen Major: Food and Resource Economics The rising cost of labor facing the Florida citrus industry has been a major force limiting its global competitiveness. One potential strategy for Florida citrus to increase its competitiveness is to implement measures that lower production costs. Replacing the current hand harvest system with mechanical harvesters may significantly change the structure of the processed citrus industry. While potentially eliminating labor shortages and variability, the adoption of mechanical harvesting will present new hurdles for the i ndustry. Coordination issues between orange growers and processors arise due to changes in the traditional harvesting and processin g procedures that have taken place since the invention of frozen concentrate orange juice. For current mechanical harvestin g technology to be economically viable, the timing of harvest and processing may need to be altered. Such changes in the way business is conducted alter the incentive structure and motivational forces driving industry players. This research analyzes neces sary conditions for growers and processors to reach a binding, sustainable agreement on the adoption of mechanically harvesting citrus. Such a solution must allow both the growers and the processors to be at least as well off, with one party made better o ff, as compared to operating under the current hand harvesting system.
18 To develop a market mechanism that brings all parties into agreement on the adoption of mechanical harvesting, each party is analyzed individually. Growers of processed oranges attemp t to maximize on tree returns by harvesting fruit when pounds solids per acre approach their maximum and harvest costs remain low. The actual harvest date for a specific grove however, is set in conjunction with processor objectives. Processors attempt to maximize returns by scheduling fruit inputs to fill processing plant production capacity, minimizing processing and storage costs, and maintaining quality parameters. Harvest and processing schedules within each processing firm seek to align o perationa l capacity and inventory management decisions with optimal plant operation goals Mechanical harvest systems have the potential to improve harvest labor productivity, allow increased flexibility dur ing harvest, and increase daily harvest capacities. All model simulations predict total industry profits will increas e with the addition of mechanical harvesting. These r esults suggest that the Kaldor Hicks compensation test holds and it is theoretically possible to make both parties better off through mechan ical harvester adoption, satisfying a necessary but not sufficient condition of adoption. Under modeling conditions, g rower profits were al ways shown to increase, while pro cessor profits were shown to decrease. So while it is theoretically possible to ma ke all parties better off, under current market conditions processors are shown to be made worse from the adoption of mechanical harvesters. Re sults suggest the need for development of a new marketing mechanism that will allow both parties to capture a sh are of the financial gains from industry wide adoption.
19 CHAPTER 1 INTRODUCTION The Florida orange juice industry processed 7,659,000 tons of juice during the 2007/08 season ( USDA, FASS, 2009 b ), accounting for about 35% of the world supply of orange jui ce (USDA, FAS, 2008 b typically amounts to over 10 billion pieces of fruit, harvested almost exclusively by hand ( USDA, FASS, 2009 b ). While the techniques and technologies of producing and processing oranges in Florida have changed drastically in the past 150 years, hand pick ing of fruit has remained a constant. Diesel driven fruit loaders and transporters have replaced the mule and buggy, tractor drawn mowers and chemical sprayers have replaced the hoe, and highly automated juice processing facilities have replaced hand ream ers, yet each piece of fruit continues to be harvested by hand. The current system of harvest consists of fifteen to thirty person hand labor crews removing fruit from the tree and placing it in field bins. A typical twenty five person crew is capable o f harvesting about three citrus trailer loads of oranges used in juice processing per day (Polopolus et al., 1996). The logistics of harvesting and processing have developed around these hand harvesting crews, used since the development of the first comme rcial groves. Finely tuned fruit transport, handling, and processing schedules operate efficiently in tandem with the hand harvest model. The rising cost of labor, however, has been a major force affecting the ability of the Florida citrus industry to co mpete with the Brazilian citrus industry, whose processors and growers possess a significant cost advantage in harvest labor (Muraro, Spreen, and Pozzan, 2003). This labor cost disparity has created an incentive for the Florida citrus industry to research develop, and implement measures to lower the cost of production, increasing the
20 competitiveness of Florida citrus. Mechanical citrus harvesters have the potential to lower harvest costs and allow the Florida citrus industry to maintain a competitive eco nomic position in world markets. Two basic types of mechanical harvesting systems have emerged for commercial use: the continuous canopy shake and catch system (Futch and Roka, 2005a) and the trunk shake system (Futch and Roka, 2005b) The canopy shake a nd catch system uses shaker heads with many long tines to shake the tree branches, causing fruit to detach. The fruit falls to c onveyers moving below the tree that load the fruit into field trucks. Fruit can also be shaken to the ground and then picked u p by hand or by machine and loaded into citrus trailers for delivery to the processing plant. Canopy shake and catch machines are capable of an average harvest of 368 trees per hour for early and mid season fruit and 462 trees per hour for late season fru it ( Roka and Hyman, 2004). During a full day of harvest in an average Florida grove, this is equivalent to an output of 10 to 18 citrus trailer loads per day per set of harvesters (Brown, 2005). Trunk shake harvesters operate by attaching a shaker boom to the truck and shaking the tree for five to 10 seconds. Fruit shakes loose from the tree and is deflected to the conveyer system that loads the fruit into field trucks following the harvester. Trunk shake harvesters are capable of harvesting 190 to 229 trees per hour of operation under normal Florida grove conditions, yielding about three to five citrus trailer loads per day per set of harvesters (Hyman, Roka, and Burns, 2005). Mechanical harvesting systems have seen much improvement in recent years and operate at much higher efficiencies than many previously developed models (Whitney, 1995). To make the machines economically efficient, however, the harvesters must capture large quantities of
21 fruit each season. This requires high daily harvest volumes, operation without interruption for extended periods, and operation in a geographically concentrated area. Ideally, mechanical harvesters are used in large groves with long, straight tree rows and short travel distances between harvesting locations. Such conditions allow for continuous, uninterrupted harvester operation, increasing operational efficiency. The switch from hand labor to mechani cal harvesters appears to be a natural transition for Florida citrus ; however, many barriers stand in the way of in dustry wide mechanical harvester adoption. Efficient operation within the current hand harvesting system does not necessarily ensure efficient operation with mechanical harvesting. Simply adding mechanical harvesters to present industry operation without addressing logistical barriers may disrupt harvesting and processing, leading to less efficient operation and potential financial losses. Identifying necessary operational changes and assessing their potential economic impact on industry returns could fa cilitate a successful transition to mechanical harvesting. These operational methods include currently used methods of fruit production and grove management, coordination between industry players, harvesting logistics, and processing operations. Changes to these systems may be needed to achieve optimal production under mechanical harvesting. The flexibility available to make these changes and the cost to change may ultimately determine the fate of mechanical citrus harvesting in Florida. Problem Statemen t In the early days of citrus growing in Florida, t he relatively low cost of labor allowed the hand harvest system to operate efficiently and effectively. Changing market conditions, specifically the rising costs of labor and the increasingly competitive world orange juice markets, are currently leading a push toward more cost effective harvesting methods. Coordination issues arise between orange growers and processors, however, when mechanical harvesters are used in
22 place of traditional hand harvest crew s. For mechanical harvesting technology to be economically viable, the traditional logistics of coordination between harvesting and processing may need to be altered. Changing the way business is conducted will alter the incentive structure and motivatio nal forces driving industry players. This research aims to identify coordination issues and logistical factors hindering the adoption of mechanical harvesters, identify possible industry changes required to allow for the adoption of mechanical harvesters, estimate the economic consequences of industry wide adoption, and suggest operational changes that could allow for economic gains to the industry. Before identifying coordination issues between industry players, the incentive structure and motivational f orces driving each player are maximize profit have historically been to maximize pound solids and reduce harvest costs. During the years of increasing concentrate production capacity in the 1960s, 1970s and 19 80s (F DOC, 2008b ), pound solid s were used as the standard quantity measurement. A pound solid based system is ideally suited for trading concentrate because processors can buy fruit and sell processed product based on the same unit of measure. Pound soli d s continue to serve as the unit of measure for the sale of processed fruit inputs, despite the growing importance of not from concentrate (NFC) production. Purchasing fruit inputs in terms of pound solid s can be problematic for NFC production however b ecause processors are buying pound solid s and selling gallons. This discrepancy may result in misaligned incentives, causing growers to maximize the sugar content of the juice and disregard other quality characteristics. While es focus on increasing pound solids they also aim to decrease total harvest costs. The piece rate cost structure of the hand harvest system historically has served growers well by providing a stable unit harvesting cost. Increasing minimum wage rates an d
23 expansion of workplace regulations are adding to the cost of labor, driving harvest costs higher. Further, changes in U.S. immigration policy could create greater variability in the farm labor market, increasing the risk of cost fluctuations. The poten tial for mechanical systems to deliver lower and less variable unit harvesting costs than those of hand harvesting systems is seen by many growers as a significant advantage of mechanical systems. nderstood without also analyzing produce a high quality juice product while minimizing operating costs. Processing managers control fruit inputs and juice outpu ts on a daily basis while simultaneously addressing a broad harvesting include potential changes in the quality of fruit inputs, increases in woody debris and o ther trash (Spann and Danyluk, 2010) and changes in daily fruit input volumes, each of which could impact plant operational efficiencies by changing the logistical coordination among harvest, extraction, and storage operations. These are very real concer ns for processors as they foresee the possibility of significant structural changes in industry operations due to changes in the harvest system. 1 2 Despite such concerns, these same processors recognize the possible benefit s allowed with further adoption of mechanical harvesters, such as more preferential fruit delivery schedules, larger peak season input volumes, the ability to harvest intermittently, and additional late season harvest capacity. Replacing the current hand harvest system with mechanical harvesters will change the operational structure of the processed citrus industry in significant ways. While potentially 1 Personal communication with anonymous manager (1) of a Florida orange processing facility, May 2006. 2 Personal communication with anonymous manager (11) of a Florida ora nge processing facility, April 2006.
24 reducing the risk of labor shortages and harvest cost variability, lowering the cost of harvest, and changing harvest capacity, the ad option of mechanical harvesters brings new coordination challenges to the industry. These barriers have impeded the transition to industry wide mechanical harvesting of the Florida juice orange crop, despite the potential gains to the industry. This rese arch will address these coordination issues by modeling the economic savings from industry wide adoption of mechanical harvesters, and finally, suggesting possible op erational scenarios in which both growers and processors could be made better off through the adoption of mechanical harvesters. Research Question What are the economic impacts on growers and processors from the in troduction of mechanical harvesters to th e Florida citrus industry ? More specifically, what economic trade offs do growers and processors face with a switch from a hand harvest to a mechanical harvest system? To answer these questions, a the oretical model is developed to explain the roles and re sponsibilities that growers and processors hold within the industry, and which market factors account for gains and losses to each industry group. From this theoretical model, an empirical model is developed and used to estimate current market conditions under hand harvesting. Current industry operations must be described to provide a starting point or reference point against which to measure the effect of future changes. This industry model is intended to simulate static, single season industry operatio ns for a representative season, not to function as a predictive forecasting model. The estimated parameters are based on previous production values and set the baseline for comparing how future market changes could affect the industry as a whole. This in dustry model is then altered to project the financial gains and losses to growers
25 and processors when using mechanical harvesters. To understand the impact of industry changes, the empirical model is run under various market conditions These conditions simulate industry changes from the adoption of mechanical harvesters, varying the intensity and duration of both harvest and processing. Comparisons across resulting estimates give insight to the expected economic impacts that changes in harvesting and in dustry operation will have on industry players. Results estimating little or no change across models suggest few ben efits f r o m mechanical harvester adoption, while larger differences in profit estimates across models suggest that industry players could be nefit from adoption. Addressing thi s research question provide s estimates of the economic and financial impacts mechanical harvesters will have on the citrus industry. Using analysis results, growers and processors should better understand potential fin an cial outcomes due to a change from the hand harvest system. The empirical modeling results also allow for optimization of the total returns to the industry under varying constraints. Demonstrating the possibility of increasing industry returns by adopt ing new harvesting technology provide s evidence for the existence of unrealized returns, suggesting that the industry could see Pareto improvements from adoption. Behavioral Nature of the Problem The development and implementation of the mechanical citrus harvester draws many parallels with the tomato harvesters introduced in California in the 1950s. In an attempt to decrease harvest costs and reduce exposure to the variability of labor market conditions, harvest labor is replaced by capital. The harveste d product is a highly perishable agricultural commodity headed for processing. While the products in question are quite similar, the closest parallel may be in the approach t o implementing mechanization. Players in th e California processed tomato industr y realized that a successful transition to mechanical harvesting would require analyzing and rethinking industry operation as a whole. The y brought growers, processors, engineers, and
26 horticulturists together to identify and overcome the major barriers re stricting the successful adoption of mechanical harvesters in their industry (Rasmussen, 1968). The systems approach used in California is being emulated by the Florida citrus industry and is a key component to the potential success of adoption. The citr us industry is currently undertaking research and development efforts that involve a wide range of experts from various fields of study to collectively advance the prospect of mechanical citrus harvesters. This cohort of expertise includes growers, harves ters, and processors, and well as horticulturists, plant physiologists, soil scientists, engineers, food scientists, and economists. It is this type of systems approach to development and implementation that will increase the feasibility of adoption. Ind ependent tomato growers and processors were able to agree on system wide changes necessary to develop a new tomato, a new harvesting system, and a new processing system that not only saved their industry, but also enabled California to be come argest supplier of processed tomatoes (USDA, ERS, 2008 a ). Similarly, independent orange growers and processors may advance the Florida orange juice industry by reaching a consensus on system wide changes enabling the adoption of mechanical harvesting. To reach a binding, sustainable agreement between Florida citrus growers and processors on mechanically harvesting citrus, neither party can be made worse off than their current situation. Such a change, in which at least one party is made better off and no party is made worse off, is termed a Pareto improvement. For a possible Pareto solution to exist, the gains from harvester adoption must be larger than any accompanying losses due to increased costs. If there are outweighing gains to be had, then a Paret o improvement should exist, making it theoretically possible to make at least one industry player better off without making the other
27 situation must be assessed befor e and after changes are made to industry operation with the addition of mechanical harvesting. Economically relevant aspects of operation are identified Whi le the existence of a Pareto optimal solution may be shown to exist, solving the coordination problem may require an actual transfer of money. The opportunity to make Pareto improvements is a necessary but not sufficient condition to change industry opera tion. Some type of contracting agreement, payment system, or transfer scheme would still have to be developed to allow for this transfer. Suggestions as to the potential working arrangements that could facilitate this exchange and bring all parties into agreement on the adoption of mechanical harvesting are included as part of this research. Under the current cost and payment structure, gains and losses from adoption are not e xpected to be uniform across industry players. Growers may be positioned to see the greatest short term financial gains from the adoption of mechanical harvesters. Current industry buying and selling practices dictate that growers bear the cost of harvest. Growers typically cover the cost of fruit production, harvest, and delivery to the plant and are then paid for the volume of solids they deliver to the processing plant. Processors commonly pay a contracted or cash price for each pound solid delivered to their processing plant, as well as processing and storage costs. Processors receive revenues from the juice when the y sell to retail or wholesale buyers. Processor s therefore may see fewer short term gains from increased us e of mechanical harvesters and could face increased processing costs and juice quality changes imposed on t hem by the increased use of mass harvesters. Research Goals The main goals of this research are to develop conceptual and empirical models that integrate grower and processor economic objectives while imposing a technology c hange from
28 hand harvesting to mechanical system s These models should allow for analysis of trade offs between grower and processor objectives and provide estimates of their financial magnitude. Identifying and investigating these trade offs is intended to inform decision makers a bou t expected results from the widespread adoption of mechanical harvesting. Overcoming industry coordination issues will be dependent on the presumption that industry wide economic improvements are achievable due to the proposed operational changes. If the industry as a whole can be shown to be made better off through the addition of mechanical harvesters, then it may be possible for some players to be made better off while no player is made worse off. To address this research goal, the focus will first be current state of operation and potential coordination problems. The objective is to quantify and optimal harvest window is de fined as the period during which harvest will maximize profits to the grower. Once the optimal harvest time is determined, the economic consequences of harvesting outside the optimal window can be assessed. Harvesting only at the peak of production, depi cted graphically as the two month harvest schedule in Figure 1 1, shows the hypothetical preferred annual harvest schedule for growers with early and late season fruit. The optimal windows for harvest are expected to be the brief periods during which each variety will yield the maximum profits for the grower. Deviating from this optimal harvest window should decrease profits due to decreased fruit quality, decreased fruit or juice quantity, and increased harvesting costs. Industry operation and coordina reference. The main issues facing the processor during daily operations are operational costs and juice quality, both potentially affected by the use of mechanical harvesters. The processor
29 controls the m aximum daily volume and flow rate of fruit inputs to the plant, but has less control over fruit quality and the demand of the end consumer. The processor, therefore, attempts to maximize returns by managing the flow of fruit to maximize plant production c apabilities and minimize unit processing costs while meeting juice quality constraints. Similar to the grower, the processor has an optimal harvest and processing timing : that which will yield the highest profits and is production schedule. This schedule varies given the characteristics of the processor, production goals, seasonal fruit characteristics, and market conditions. Figure 1 1 shows several hypothetical harvesting schedules for facilities operating under ten, six, four, or two month processing schedules. Assuming that plants have an optimal monthly production capacity, the four scenarios depict potentially optimal operation over varying seasonal lengths. The choice of operating schedule depends on individual operational goals, available capital, and seasonal variations in fruit supply. Important to the processor model is the assessment of the economic consequences of operating outside of this optimal production hedule is expected to maximize profits by maximizing the operational efficiency of the processing plant and minimizing costs subject to processing goals. Deviating from this schedule should decrease processor profits due to decreases in juice quality and quantity, and increases in processing and storage costs. Combining grower and processor objectives into a single economic model allows for an integrated analysis of the entire Florida orange juice production industry under differing harvest systems. Optim izing the complete industry model under different sets of constraints and market assumptions simulate s industry performance under the adopti on of mechanical harvesting technolog y The results of this model are intended to provide evidence of the potential financial impact that mechanical harvesting could have on industry players and allow for the
30 recommendations on the necessary operational changes required to realize any projected gains or minimize any projected losses. Research Objectives Identify the im portant grower and processor decision variables and how changes to these decisions affect industry coordination and the adoption of mechanical harvesting. Construct a model to estimate the economic and financial consequences to industry players from mechan ically harvesting and processing fruit. Quantify the expected returns from the decision to harvest and process juice oranges, focusing on the intensity and duration of harvest and processing operations, and comparing estimates of net changes to individual and overall industry returns under multiple harvesting scenarios and market assumptions. Predict the required number of mechanical harvesters, weekly extraction and processing capacity, and juice storage capacity for the industry under differing market co nditions. Estimate any potential financial gains available to the industry through adoption of mechanical harvesters, and present possible market changes that could allow for Pareto improvements as the industry transitions towards adoption. Importance of the Study Findings from this research are important to the future of the Florida citrus industry, which is currently facing increasingly competitive world market s, destructive citrus diseases, and a constantly changing labor market Brazil ian producers co leader of orange juice and operate at a lower cost than Florida producers (Muraro, Spreen, and Pozzan, 2003). Citrus d iseases, such as Huanglongbing (HLB), or greening, are increasing grove management costs, decreasing fruit yields, and destroying trees (Morris and Muraro, 2008 ; Morris, Erick, and Estes, 2009 ) Labor market fluctuations and legislati ve changes to immigration policy add uncertainty to harvest labor costs (Waters Emerson, and Iwai 2008). Successful ind ustry wide adoption of mechanical citrus harvesters by Florida growers could s by decreasing harvest costs and reducing the risk of harvest cost fluctuations. U.S. orange juice consumers also stand to gain from the
31 adoption of mechanical harvesters. Decreases in production costs and increases in industry profits should encourage increase s in orange juice production. Ultimately, s upply increases have the potential to decrease the unit price of orange juice at retail and increase consumer welfare. Research advancing the knowledge of economic conditions resulting from the adoption of mechanical harvesters should aid the transition to mechanization of the harvest. While this research alone may not alter the tr ansition, findings from this analysis comprise an important part of the research being conducted on mechanical harvesters, an integral part of the systems approach required for successful adoption. Similar to the development and adoption of the tomato har vester by the California tomato industry in the 1960s (Rasmussen, 1968), this study is just one part of the coordinated research effort being conducted by many individuals within and around the industry. Findings from this analysis are intended to advance the economic component of the systems approach to mechanical harvester adoption. There is currently no complete, published study of the coordination issues delaying the industry wide adoption of mechanical citrus harvesters. Extensive research has been c onducted over the past 40 years on improving and implementing mechanical citrus harvesters in Florida (Whitney, 1995). The overwhelming thrust of this work, however, has been on the engineering, labor, and horticultural aspects of mechanical harvesting. Some more recent papers have discussed the economic implications of mechanical harvester adoption in other industries and the potential costs and benefits of switching to mechanical harvest (Cembali et al. 2007; Durham, Sexton, and Song, 1996). A complet e analysis of the expected economic impacts of mechanical harvesters on the Florida orange juice industry has not yet been published. This research aims to fill this void in the literature.
32 Research Scope The broader scope of this research is to identify coordination problems hindering industry wide adoption of mechanical harvesters, estimate the economic impact of potential changes, and suggest strategies to capture any additional rents from mechanical harvester use. Mechanical harvesting has the potenti al to alter the structure and the future economic outlook of the Florida citrus industry, but industry wide adoption will not be possible without a concerted effort from both growers and processors. A goal of this research is to suggest possible changes t hat will allow for Pareto improvements within the industry, while avoiding normative judgments as to which party should receive any additional rents. This research does not attempt to address issues of the social equity of potential solutions. It also def ers the welfare analysis of displaced workers due to mechanical harvester adoption to previously published literature (Fairchild, 1974; Schmitz and Seckler, 1970). While the interests of farm labor are important and deserve much attention, they are outsid e this research scope Additionally, this research does not focus on specific aspects of mechanical harvester operation, engineering, or effects on tree health. While these topics are central to the adoption decision, in depth investigation and discussio ns will be left to the experts in the ir respective fields. The current outlook on mechanical harvester operation is positive, as these systems have been shown to be feasible and favorable (Buker et al. 2004; Brown, 2005). The opportunity for successful adoption of mechanical citrus harvesters and the potential financial gains that could come from industry wide adoption warrant continued research into its economic feasibility (Neff, 2006). Dissertation Format Chapter 1, Introduction, is an overview of t he research project, problem statement, behavioral nature of the problem, research goals and objectives, importance of the study, and the
33 scope of the research. Chapter 2, The Florida Orange Juice Production Industry, outlines the current state of the Flo rida citrus industry, including industry operations, grower and processor incentives, payment and contracting systems, an overview of the current systems of harvest, available mechanical harvesting systems, and possible industry changes to coordination and operation from adoption of mechanical harvesters. Chapter 3, Conceptual and Theoretical Framework, discusses the adoption of mechanical harvesters by other industries, reviews past studies of adoption of new harvesting technology, and explains how previo us economic research sets the theoretical groundwork for this research. Chapter 4, Biological Characteristics of Florida Oranges, outlines the specific biological characteristics of oranges used in juice processing, develops a theoretical model explaining biochemical changes in Florida oranges, and describes the biological data used to estimate the juice production model. Chapter 5, Theoretical Modeling: Scheduling of the Orange Harvest to Optimize Returns, defines processing intensity and duration and ou tlines the theoretical economic model explaining the Florida orange juice production industry operations. Chapter 6, Empirical Modeling and Results, develops and estimates a mathematical programming model simulat ing current industry conditions and account ing for changes from adoption of mechanical harvesters. This chapter also includes a complete description of the data and procedures used to estimate model parameters. Chapter 7, Conclusions and Future Work, analyzes results from the empirical model and the economic gains and losses within the industry as adoption proceeds, describes how projected economic outcomes vary under different harvesting and processing scenarios, summariz es the main findings of the research, suggests possible changes to the curre nt system of operation that could allow for improvements to the industry as a whole, and presents ideas for future research.
34 Figure 1 1. Hypothetical annual harvest, processing, and utilization schedules.
35 CHAPTER 2 THE FLORIDA ORANGE J UICE PRODUCTI ON INDUSTRY The Florida orange juice industry produces one accounting for 7,659,000 tons of juice produced worldwide during the 2007 08 season (USDA FAS, 2008 b to juice annually ( Figure 2 1). resulting in decreased juice production, the state remains second only to Brazil as a leading global orange juice producer (U SDA FAS, 2008 b ). Seasonal orange juice production usually starts on a limited basis in November, runs heavily from January to May, and finishes by early July. An example of a typical production schedule is shown in Figure 2 2, which details the percent age of the annual crop that was processed each month during the 2005 06 s eason (FDOC, 2007 ). Fruit growth, harvesting, and processing during the 2005 06 season followed a typical November to July seasonal production cycle. The annual production schedule is the end result of market forces, biological characteristics of the trees and fruit, and physical limitations of industry operations culminating into a single production outcome. This research examines how changes in the method of fruit harvest will fu ndamentally change industry operation and the production outcome, focusing on the resulting economic consequences of these changes. To fully investigate the factors shaping this production schedule and outcome, each production determining factor is analyz ed. The purpose of this chapter is to describe industry operation; including details of how fruit moves from the grower to the processor currently used h arvest ing m ethods the payment system used to compensate growers and harvesters harvesting and proce ssing schedules, and processed product
36 storage. The final section of this chapter describes some potential changes to industry coordination and operation brought on by the adoption of mechanical harvesters. Industry Operations Operations of the Florida or ange juice processing industry can be separated into three distinct steps: fruit production, harvesting, and processing. Growers handle all operations and decisions involving fruit production and fruit harvesting, including both hand and mechanical harves ting. The harvesting process is defined for this analysis as the steps required to move fruit from the tree to the juice plant, including fruit picking, roadsiding, and hauling to the processor. While many of these grower operations are often performed b y independent contractors, all takes place when bulk trailers are e processing line. Several mechanisms are used for purchasing fruit and structuring payments, including different contracting options and cash market transactions. Processing operations include fruit procurement, scheduling of fruit to the processing plant, fruit extraction and juice processing, storage of juice or concentrate, and shipment of product from the plant. Figure 2 3 gives an overview of the steps required to convert on tree fruit into processed orange juice ready for distribution. Fruit Production Growers begin each production season with a fixed acre age of grove from which they will produce their annual orange crop. Given this acreage, the typical operational goal for juice orange growers is to maximize the pound solid production per acre of grove, subject to cost constraints. Fruit production, as d efined for this analysis, involves all grower operations, from scheduling of chemical applications, irrigation, and freeze protection. Growers form operatio nal
37 decisions in an attempt to maximize profits by maximizing total pound solid production, reducing exposure to risk of crop damage from freeze or disease, and minimizing the costs of required capital and labor inputs Harvesting In addition to grove prep aration, management, and fruit production, g rowers are responsible for harvest and delivery of their fruit to a processing facility. Harvest and delivery are t ypically carried out by a third party harvesting specialist and are viewed by the i ndustry as th ree distinct steps: picking, roadsiding, and hauling. Picking fruit is defined as removing fruit from the tree and placing it into field tubs, fruit containers that hold about ten 90 pound boxes. Roadsiding is the movement of picked fruit from the field to open fruit trailers. Roadsiding also includes the costs to supply field equipment, worker transport, and additional worker expenses, including payroll services, taxes, and insurance. Hauling is the actual transport of the loaded fruit trailers from th e grove to the processing facility. The grower decides whether to conduct the harvest using traditional hand labor or mechanical harvesting equipment but rarely engages in the actual harvest The harvest and ownership of mechanical harvesting equipment will, therefore, be assumed to be carried out by third party harvesting companies. While growers and processors could purchase their own harvesting equipment, such investment is typically beyond their scope of operation and would involve additional risk e xposure. Hand h arvest Hand harvesting is used to harvest almost every piece of citrus fruit produced in Florida. The system of hand harvest most commonly used consists of a group of 15 to 30 hand pick laborers and a field truck whose driver usually serv ers as the crew leader. Pickers are paid a piece rate for every 90 pound box of fruit they harvest and place into tubs positioned along tree
38 transports the fr uit to bulk fruit semitrailers. Crew leaders manage work crews and orchestrate all in grove harvesting operations. This type of hand labor system has been used since the first commercial groves were planted in the state, and the citrus industry has subse quently developed around this harvesting system. Fruit transport, off loading, and processing schedules operate in concert with the hand harvest model. A 25 person crew is typically capable of harvesting about three bulk fruit trailer loads, or 1400 to 1 600 boxes of oranges per day. 1 This harvest rate example assumes 20 workers are each able to pick about 80 boxes per day (Roka and Cook, 1998), with four workers operating goats and one crew leader. Harvest worker productivity will vary by worker experie nce, motivation, grove conditions, and fruit characteristics. Using hand crews, the average cost to pick and roadside fruit for processing in Florida during the 2005 06 season was about $1.95 per box for early and mid season fruit and $2.07 per box for la te season fruit (Muraro, 2006). Additional costs for hauling fruit to the processing facility vary based on distance traveled, averaging $0.50 per box for transporting fruit 30 to 50 miles. Additional information on average picking, roadsiding, and hauli ng costs is shown in Table 2 1. Figure 2 4 gives an additional example of hand harvest cost, showing how costs changed for one harvesting company during the 2003 04 season (Unpublished data from personal interviews with selected citrus harvesters during t he 2003 04 season conducted by B. Hyman and F.M. Roka, 2004). In this example, costs remained relatively steady early in the season and then increased after May 15, possibly due to decreases in labor availability, worsening picking conditions, or the harv esting of more low yielding groves. Labor supply typically decreases late in the harvest season as other agricultural crops begin to ripen across the country and workers migrate out of the state (Burns et al., 2006). Cost increases are consistent with th e findings of Muraro (2006) 1 Personal communication with anonymous manager (9) of a Florida orange harvesting company, July 2006.
39 showing a 19% increase in cost over the season average to harvest Valencia oranges after May 15. The higher cost of harvest labor in the U.S. has been a major force affecting the ability of the Florida citrus industry to compete with the Brazilian citrus industry. Citrus producers in Brazil possess a significant cost advantage with respect to harvest labor which allows them to operate at a much lower cost than the U.S. industry In addition to higher wage rates in the U.S., st ricter labor laws increase employment costs and expose U.S. industry to the risk of future cost increases due to increased re gulatory requirements (Fernandes 2003). One potential solution to the high, variable labor costs facing Florida citrus is to impl ement measures that lessen the need for labor, such as increasing mechanized harvesting and processing operations. Trading labor for of Florida citrus (Muraro S preen, and Pozzan, 2003). Mechanical harvesting systems First researched and deve loped in the 1940s, mechanical harvest of the juice orange crop is one possible solution to lowering harvest costs, making the Florida citrus industry more competitive in the world citrus market. The highly competitive world market forces the Florida industry to search constantly for new opportunities to improve their competitive position. Adding mechanical harvesters to the current production system will substitute capital i mprovements for labor inputs, a high cost input for U.S. growers. This capital for labor substitution has long been the trend in the advancement of production agriculture; however, capitalization of the citrus harvest has experienced varying success durin g the past five decades. Despite extensive research and several different mechanical harvester designs in commercial production, mechanical harvesters still account for very little of the total annual harvest. Since 1995 however, the Florida citrus indus try has increasingly pursued widespread use of mechanical
40 harvesting. During the 2007 2008 season, approximately 32,500 acres were mechanically harvested yielding approximately 9.6 million boxes of fruit (Citrus Mechanical Harvesting, 2009). The motivat ion for pursuing mechanical harvesting is to significantly reduce harvesting costs and close the competitive cost gap between growers in Florida and Sao Paulo, Brazil. Two basic types of mechanical harvesting systems are availabl e for commercial use in F lorida: the continuous canopy shake and catch system and the trunk shake system. The continuous canopy shake and catch system uses a pair of large harvesters, each carrying several shaker heads with many long tines. As the harvesters move together on opp osite sides of a single row of trees, the tines shake the tree branches, causing fruit to detach. The fruit falls to conveyers moving below the tree that load the fruit into field trucks. Additionally, a tractor pulled canopy shake harvester can operate independently, allowing fruit to fall to the ground. Fruit shaken to the ground is then picked up by hand or machine and loaded into citrus trailers for delivery to the processing plant. Roka and Rouse (2004b) report that the continuous canopy shake and catch systems have the capacity to enhance worker productivity tenfold over th at of hand harvesting crews. Canopy shake harvester crews have been reported to harvest an average of 368 trees per hour for Hamlin oranges and 462 trees per hour for Valencia o ranges (Hyman, Roka, and Burns, 2005). During a full day of harvest in an average Florida grove, these harvest rates are equivalent to about 10 to 18 citrus trailer loads per day, depending on the number of boxes on each tree. Brown estimated potential c osts savings of 20% to 75% from switching to mechanical harvest (Brown, 2005). Trunk shake mechanical harvesters have been used successfully to harvest Florida juice d shaking the tree for five to 10 seconds. The fruit shakes loose from the tree and is deflected to
41 the conveyer system that loads the fruit into field trucks following the harvester. A typical trunk shake harvester crew can average about 190 trees per h our for Hamlin oranges and 229 trees per hour fo r Valencia oranges (Hyman, Roka and Burns, 2005). Under normal Florida grove conditions, this will yield about three to five citrus trailer loads per day, depending on the boxes available per tree. Several operational changes may be necessary to increase the efficiencies of harvesting operations, allowing mechanical harvesting technologies to achieve their inherent economies of scale and significantly lower harvest costs. While mechanical harvesting systems have seen many improvements in recent years to operate with much greater efficiency than previously developed models, optimizing returns requires harvesters to operate without interruption for extended periods in a geographically concentrated area, yieldi ng high daily harvest volumes. To fully capture scale efficiencies, mechanical harvesters need to operate near full capacity for extended periods of time and yield high harvest volumes per machine. A typical trunk shake harvester crew requires a larger n umber of daily trailer allocations to operate without interruption, as compared to a hand crew. Such intense operati on in fewer locations is depende nt on the availability of a high volume of timely fruit delivery opportunities to avoid costly downtime due to transport delays. Additionally, adjacent groves may be required to be harvested in succession due to the high cost of relocating the harvesters. Ideally, mechanical harvesters are best suited for large groves with long straight tree rows that allow f or continuous operation. Harvesting groves in a manner consistent with these operational goals will alter harvesting schedules with respect to the date on which a grove is harvested. Altering the date of harvest can change the quality and quantity of fru it yields, changing industry profits. Changing the harvest schedule adversely affects industry when harvest is forced at a less than the optimal time in the
42 maturation cycle and positively affects returns if harvest is conducted during a more optimal time period. Concentrating daily harvest locations to fewer geographic locations may also limit a High volume harvesting systems may induce changes to juice stora ge and juice inventory management strategies due to increasing the rate and volume of product entering storage. Understanding and addressing these potential changes to harvest will aid in smoothing the transition from hand harvest to mechanical harvest, i mproving the likelihood of success. Processing Operations Processor operations are defined for this research to include fruit procurement, payment, scheduling of fruit to the processing plant, juice extraction from oranges into NFC or frozen concentrate or ange juice ( FCOJ ) storage of juice, and movement of a blended juice product to a buyer. Florida orange juice processing plants depend entirely on oranges produced within the state and are continuously adjusting processing capacity to match expected fruit volumes. Processing facilities adapt available capital and labor to best fit the annual output of oranges, attempting to process variable quantities of fruit with relatively fixed processing capacity For example, during the 2006 2007 season, Florida pro cessing plants used over 123 million boxes of orange inputs with 75 million of these boxes (61%) going to NFC production. The processing season began in early October and ran through the end of June (FDOC, FCPA, 2008b ), with the smallest weekly processin g volume, over 19,000 boxes, run during the first week of processing, and the largest weekly total, over 5.5 million boxes, run during the third week of January. These production levels were vastly reduced from the 2003 04 season annual production, down 4 7% from the nearly 234 million boxes of oranges processed during that season, of which 96 million (40%) went to NFC production. Despite the large decrease in fruit production and an 88 million
43 gallon decrease in FCOJ production (84%), there was only a 21 million gallon decrease in NFC production (22%). Processing during 2003 04 season began similarly in early October and continued through the second week of July. Weekly input usage ranged from 34,000 boxes during the first production week to over 9.5 mil lion boxes in the first w eek of January (FDOC, FCPA, 2005 ), 4.0 million more boxes than the 2006 07 peak. Such large swings in annual fruit availability and peak production levels exemplify the highly variable operational environment in which citrus proce ssors are attempting to conduct business. Optimal production levels for each individual processing facility in the state are privately held and therefore not readily available. Total industry production and estimates of individual processing are more read ily available. One such estimate presented in Table 2 2, shows the estimated percentage of the 234 million boxes of fruit inputs used by the twelve largest Florida orange juice processors operating during the 2003 04 season ( based on u npublished data of a nnual p roduction e stimates for Florida c itrus p rocessors compiled by R. Hunter, 2006 ). Since plant have ceased operations. These three plants combined acc ounted for less than one quarter of 2003 04 juice processing. Specific information as to the type of end product being produced at each plant, FCOJ or NFC, or total finished product storage capacities is not available Fruit procurement Fruit procurement is defined as the active searching and purchasing of fruit supplies by a processor to be used as production inputs. The operational objective of the processing plant s f ruit p rocurement m anager is to establish a long term fruit procuremen t strategy that w ill fill fruit input needs on both a daily and a seasonal basis. To fill these supply needs, the company must first determine the quantity of inputs required. The plant s base needs can be estimated based on plant size, available processing equipment, an d historical processing volume data. The
44 p rocurement m anager then work s before and during the season to secure this base supply of fruit and makes daily adjustments to maintain an orderly and sufficient input volume To avoid paying high fruit prices, pr ocessors try to avoid purchasing large quantities of fruit in the cash market to fill plant capacity. Such buying conditions may signal a strong demand from inputs and can trigger a quick increase in fruit prices. To avoid this potential increase in cost s processor s operate within their long term fruit procurement strategy while constantly adjusting fruit needs throughout the season. Procurement s trategies There are two fruit procurement strategies used by processors to secure fruit to meet the plant s base input needs. Processing plant manager s can either purchase fruit or they can purchase citrus groves to meet input requirements. While some Florida proce ssors still vertically integrate and own citrus groves, most processing plants fill base supply needs by contracting with local growers Fruit purchases are conducted either by directly contracting with grove owners or through the cash market An average size processing plant may deal with 40 or more different suppliers to fill their annual load re quirements (Becker, 2006), u sing several different types of contracts to secure and purchas e fruit inputs well prior to the time of processing. Long term contracts often extend for five to 20 years, while short term contracts are used for purchasing fruit less than five years in advance. Additionally, spot market purchases are used to increase fruit input levels within a few weeks of purchase. These cash deals are often arranged by fruit buyers working for the processor. Processors use a combination of contracting and spot market purchases to manage input flows, attempting to optimize processing operations. About 65 % of annual fruit inputs will typically come from long term contracting, 20% from shorter multiyear contracts of less than five years, and the remaining 15% of inputs from year to year contracting and cash market
45 purchases. 2 This system allows the processor to secure about 85% of the needed fruit supply before the processing season begins. Capital and labor requirements for processing the e xpected input levels can then be set at the beginning of the season, with additional inputs acquired later to adjust for actual crop yields and fill any remaining processing capacity Due to uncertainties associated with annual yields, leaving a portion o f fruit supply needs to be purchased through the cash market avoids over purchasing of inputs Scheduling of plant operations allow s for excess capacity to adjust processing due to weather, plant breakdowns, and seasonal fruit supply variations Purchase of fruit in the cash market is often coordinated during the harvest season by fruit b uyers employed by the processor A typical processor employs three to six buyers throughout the state, with each buyer managing several field workers 3 The fruit buyers search out and purchase the needed fruit in the cash market, while field workers arrange for harvest and delivery of this frui t t o the plant. Supplementing shortages in contracted fruit with cash market purchases allows processing plant managers to adjus t the flow of inputs to maximize plant operation. Securing fruit to fill plant capacity can also be achieved by purchasing fruit through a third party broker, know n in the industry as a b ird dog. Bird dogs are essentially fruit brokers who are also involv ed with harvest and delivery They deal mainly with small growers, typically less than 500 acres, and are responsible for marketing, pi cking, and hauling fruit. Rarely using long term contracting, they instead purchase fruit directly from the grower or i n spot markets. A bird dog agree s to purchase a grower at a cash price that includes the price of the fruit plus a handling fee. Processors then buy directly from bird dogs and do not have to venture onto the 2 Personal communication with anonymous manager (6) of a Florida orange processing facility, April 2006. 3 Personal c ommunication with anonymous manager (6) of a Florida orange processing facility, April 2006.
46 cash market. B ird dog s handle all th e transactions with the grower and allow the processor to easily and anon ymously fill plant capacity, reduce transactions cost and lessen the chance of driving up prices. In addition to securing fruit inputs, a central reason for long and short term contr acting is to decrease uncertainties associated with fruit inp ut costs. While contract terms vary widely, they are usually written to include a variable selling price, designed to lessen risk exposure to both parties. A price floor guarantee is often incl uded that requires the processor to pay a minimum price per pound solid, independent of current market prices This floor price is designed to cover the cost of fruit production decreasing the risk faced by growers. When fruit is exchanged with the proc essor, the grower receives the higher of the floor price or the market price. I f the market price is above the floor price, the grower receives the market price less a small percentage discount for the processor. Paying less than market price in this sit uation lessens the processors exposure to high fruit input prices. This contracting system allows both parties to hedge their positions in the market, while growers are guaranteed a buyer for their fruit and are protect ed against low market prices and pro cessors secure a fruit supply that is prote cted against high market prices. Payment Growers are paid for their fruit based on the total quantity of soluble solids extracted at the processing facility. Pounds of soluble solids, or pound solids, are calcula ted by state inspectors at the processing plant by extracting juice from a sample of fruit entering the plant (Tetra Pak, 1998). Pound solids per box are calculated as the temperature adjusted degree Brix divided by 100, times the pounds of juice per 90 p ounds of fruit (Wardowski et al., 1995). The total pound solids delivered to the processor times an agreed upon delivered in price per pound solid in prices per pound solid fluctuate during the season,
47 between sea sons, and given the exchange mechanism used for the sale. The delivered in price received varies for fruit sold in the cash market, through a participation plan or by contracting directly with the processor (Ward and Kilmer, 1989). Growers strive to maxi mize pounds solid s b ecause they are paid based on the quantity of soluble solids delivered. A grower is therefore not as concerned w ith other fruit and juice quality attributes that processors value highly G rower s the n attempt to capture the greatest re turns by harvesting fruit at the time coinciding with peak pound solid s per acre yields, understanding that they must also be wi lling to have fruit processed at less than optimal times due to the physical constraints of harvesting and processing. Payment f or fruit typically occurs a fter a grower fruit has been processed This first payment is usually received within the week but is only a partial payment, normally about 80% of the total value When the juice is sold and a final sales price is determined the grower will then receive the remaining payment due. 4 A grower can face additional fees for delivering fruit below minimum quality standards, with excessive decay, or that is improperly delivered. 5 When fruit is held on the tree until late in the ha rvest season, growers are often compensated through an increase in price. Processors have paid an additional 5% price premium to growers for holding fruit on the tree until June. 6 Processors are willing to pay this additional cost because the decreased c ost of storing processed juice offsets the increased price. Processors commonly refer agreements because of the increased price paid per box. The actual savings or cost to either 4 Personal communication with anonymous manager (6) of a Florida orange processing facility, April 2006. 5 Personal communication with anonymous manager (1) of a F lorida orange processing facility, July 2008. 6 Personal communication with anonymous manager (7) of a Florida orange grove management company, July 2006.
48 party depends on several factors, including any pound solids change due to the delayed harvest and the realized processing or storage cost differences. Some NFC processors have attempted to move away from the pound solids based payment system and adopt a payment system compensating growers for the gallons of juice delivered to the processing plant 7 Such a payment change would incentivize growers to manage fruit for gallon production. Additionally, growers and producers would both be s elling their product based on a gallon system, allow ing both parties to be driven by a common incentive While this change to the payment system could align juice production goals, fruit is not typically being traded on a gallon system Scheduling and loa d allocations Once procured, fruit must be scheduled for delivery to the processing facility. Most processing plants employ a f ruit s cheduler who is responsible for arranging loads to be supplied to the plant for the coming week s During a typical work w eek, the f ruit scheduler designates which growers or suppliers will be bringing fruit to the plant each day and how many loads each will deliver. To set delivery requirements the scheduler must first determine how many loads will be required during the s cheduling period. The target input volume is based on fruit availability, harvesting and processing capabilities, and weekly processing goals. Once the total number of loads needed for the week is determined the f ruit s cheduler tallies all outstanding f ruit under contract. The number of loads each grower is allocated to deliver is prorated based on the estimated number of boxes of fruit the grower has contracted to be delivered. Loads are allocated to each grower based on their remaining balance share G rowers with more on tree 7 Personal communication with anonymous manager (11) of a Florida orange processing facility, April 2006.
49 fruit under contract are permitted to bring more loads to the plant and growers with less fruit under contract are allowed fewer loads Consider, for example, a processor attempting to schedule the operation for early and mid s eason fruit from November to March. T he f ruit p rocurement m anager and the f ruit s cheduler work to ens ure a steady stream of fruit inputs by first calculating the number of loads need ed each processing day. Accounting for holidays, rain days, and mechan ical breakdowns, there are typically about 120 day s of plant operation available to process early and mid s eason fruit Dividing the total number of boxes under contract by 120 days will give the number of boxes per day that ne ed to enter the plant. If t he processor has 10 million boxes under contract, then about 83,000 boxes of fruit need to be processed daily Assuming 50 0 boxes of fruit per trailer the processor needs to run 166 trailers per day. Estimates of the daily supply can then b e adjusted an d recalculated throughout the season as condi tions change due to storage space availability, throughput limit ations equipment breakdowns, or s easonal changes in fruit characteristics Additional quantities needed to fill plant production capacity can be purchased in the cash market. Processors operating multiple plants will often have only one f ruit s cheduler responsible for coordinating loads to all processing facilities allowing for the most efficient routing of loads. Available load allocations are assigned to growers to fill the processing schedule determined by plant management. The number of loads a grower is allocated is prorated based contracted f ruit remaining in the field. Continuing with the previous example, the fruit scheduler distributes the 166 trailer loads of fruit needed per day, or 1,162 loads per week among growers. Now assume t he processor estimates that 10 million boxes of fruit ar e under contr act in the field from which the loads can be drawn. Of those 10 million available boxes, suppose a single
50 grower controls about 250,000 boxes, or 2.5% of the contracted supply The scheduler then sets weekly load allocations for that grower at 2.5% of the weeks required loads, or about 29 loads, and informs the grower to deliver the loads the following week, approximately four loads per day The grower makes arrange ments to have picking crews harvest the required fruit volume da il y and to ma ke delivery to the processor. These calculations are performed weekly to account for changes in fruit supply and processing goals. While t his s cheduling scenario is simple for a large grower with many thousands of boxes under contract it becomes more di fficult when dealing with sma ller growers who cannot make daily or weekly deliveries These smaller growers must negotiate repeatedly with processors for a few delivery dates throughout the season. Scheduling of fruit for processing is a key determinant of processed product quality and quantity levels because the timing of delivery determines the timing of harvest. The timing of the harvest directly impacts economic returns to both processors and growers because of the dynamic nature of the agricultural crop. Quality and quantity parameters are continuously changing as fruit hangs in the field and quickly deteriorates after harvest. Recognizing these quality and quantity changes, growers and processors attempt to schedule fruit for harvest and delivery at a date that produces the largest and highest quality juice yield (Searcy, Roka, and Spreen, 2008). Scheduling harvest at the optimal yield is not always possible, however, due to the narrow window of peak fruit production and juice extraction and stora ge capacity constraints at the processing plant. Instead of harvesting at the peak of production, fruit is harvested and processed at somewhat regular intervals over the course of the season. Some of the crop gets harvested at optimal times but other por tions are harvested at suboptimal times as juice characteristics change with fruit maturation. While this system of scheduling the harvest may
51 not capture all available pound solids, it has been used throughout the industry for many years and is viewed by industry players as an acceptable system of managing fruit inputs. Awarding processing slots based on a remaining balance basis is seen by many industry way of a llocating loads, even though it allows them little influence over scheduling of loads. P rocessors argue that f ull control over plant operations, including delivery scheduling, is required to maximize plant operations 8 9 While processing plant managers d o not allow input schedules to be dictated by the growers th is system is flexible enough to allow for scheduling changes due to extenuating circumstances or occasional load swapping between growers. For example, g rowers are not forced to process immature fruit just to fill load allotments, and fruit on the verge of spoilage is often given a higher scheduling priority. Even the largest growers have little influence over load allocations and are seldom able to exert market power over scheduling. Instead o f haggling over schedules large growers typically will bargain for a higher price floor and a higher percentage of the market price. Juice extraction Juice extraction is defined for this analysis to include all steps required to convert delivered traile rs of orange inputs into NFC or FCOJ ( Figure 2 5). While this definition of juice extraction far exceeds the single step of juice removal that takes place in the processing plant, it is intended as a catchall term for the steps required to convert fruit t o storable juice. The major steps in this conversion include fruit unloading, washing, grading, juice extraction, finishing, either evaporation or pasteurization, and removal of waste or byproducts. Extraction is completed 8 Personal communication with anonymous manager (11) of a Florida orange processing facility, April 2006. 9 Personal communication with anonymous manager (6) of a Florida orange processing facility, April 2006.
52 when the juice or concentrate i s placed into a short term storage holding tank. A detailed flow diagram is shown for canned FCOJ production in Figure 2 6. Bulk FCOJ production skips the canning steps and goes directly from blend tanks to bulk tanker trucks. NFC production differs in that juice moves from the finishers to pasteurizers and then into s torage tanks. Juice extraction decisions are heavily influenced by the quality objectives of the end product. Minimum quality constraints on fruit inputs and juice outputs are mandated by the government. Tables 2 3 and 2 4 give the minimum quality standards for all fruit inputs coming into the processing facility as determined by the Florida Citrus Code (2006) State standards restrict fruit inputs based on peal color break, total soluble solids content, Brix to acid ratio, acid percentage, and minimum juice content levels as measured in gallons per box. Stricter quality parameters are often imposed by individual processors. One Florida processor, for example, requires all fruit processe d before December 15 to yield juice at or above a 14 Brix to acid ratio. This minimum ratio is considerably higher than the 9 to 10.5 ratio required by the Florida Citrus Code. Juice scoring below a 14 ratio but above the state required minimum is accept ed and 10 Table 2 5 shows additional quality regulations imposed by the federal government for processed juice sold at retail or wholesale. While these standards do not regulate the quality of frui t inputs, juice leaving the plant must meet the federal guidelines. The Federal Regist e r lists quality parameters for degree Brix minimums, ratio limits, oil concentrations maximums, and a minimum required USDA score, which includes juice color, defects, and flavor. Processors must monitor the juice qualities of fruit inputs so that juice outputs will meet required quality standards for future sales. Similar to the stricter quality standards required for some fruit inputs, individual processors often imp ose 10 Personal communication with anonymous manager (2) of a Florida orange processing facility, July 2008.
53 stricter quality parameters on juice outputs to ensure the juice meets quality requirement of s ubsequent juice buyer or end consumer. D ecisions made by citrus processing companies are heavily dependent on the final quantity and quality needs of the com pany regardless of whether they are a bulk processor, branded p roduct seller, or a cooperative Their individual needs determine plant production goals and influence facility operations. For example, a bulk processor may set a production goal of process ing as much frozen concentrate orange juice as possible at the lowest achievable unit cost. The quality of the processed juice may be less of a concern to this processor than maintaining high processing volumes and minimizing the unit cost of production. Later blending of concentrate could then be used to meet the quality demands of wholesale buyers. Conversely, a branded product seller may be more concerned with producing NFC juice that meets certain quality standards to maintain a consistent retail pro duct. The branded seller may be willing to pay higher unit processing costs and operate at suboptimal processing volume levels to achieve a higher juice quality. A dditionally, a cooperative may adjust plant operation t o focus more on the needs of grower members, allowing delivery of fruit at times that decrease plant operating efficiency Because processing revenues are returned to growers, decisions are made to return the greatest combined profits from growing and processing operations. These three dif ferent production goals will influence plant operations and plant managers will adopt varying production s trategies to meet these goals. The percentages of FCOJ and NFC that an individual plant produces affect the operation of their processing system. Co nversion of fruit inputs to either FCOJ or NFC is determined by a was based almost entirely on FCOJ production from its first commercial distribution in 1946
54 through the 1980s (Zellner, 1992). More recent advance s in pasteurization and storage technologies, along with changing consumer preferences, have pushed the industry toward more NFC production (Goodrich and Brown, 2001). The industry has gradually inves ted in and converted capital to produce larger quantities of NFC ( Figure 2 7). The move to greater NFC production has altered the incentives linking fruit and juice production. The price per pound solid input purchasing system currently used by the indu stry affords FCOJ processors a simple accounting system of costs and revenues. FCOJ processors buy fruit from growers in units of pound solids and sell concentrate to wholesalers in units of pound solids. An NFC processor, however, buys inputs in pound s olids but sells in gallons. Because pound solids and gallons are not directly comparable, the cost of fruit inputs per gallon can vary based on the juice characteristics. For example, high degree Brix inputs cost processors more per gallon than low juice inputs because the concentration of sugar is greater in a gallon of high degree Brix juice. Additionally, NFC processors have fewer options than FCOJ processors when blending juice after processing due to product specific production practices. These lim itations require NFC processors to more closely monitor quality aspects of fruit inputs than would FCOJ processors. Processing capacity and the expected crop size both influence the decision to process either NFC or FCOJ. Estimates of statewide processing capacity and capacity level changes help explain how processing costs will change as volumes processed change. The U.S. International Trade Commission ( US ITC) estimated all U.S. orange juice production, production capacity, and utilization for the 2001 0 2 through 2004 05 seasons (Table 2 6) ( US ITC, 2006). The US ITC survey of processors revealed that total production capacity of almost 1.7 billion pound solids was available in 2004 05, but less than 60% of capacity was utilized. The disparity in NFC and
55 FCOJ production, 75% utilization verses 50% utilization respectively, and their volume changes from the previous season suggest that FCOJ production decreases proportionally more when fruit input supplies are short. Further evidence of FCOJ production dec reases are shown in Figures 2 8 and 2 9, which detail recent Florida processing volumes from the 2000 01 through 2005 06 seasons for NFC and FCOJ respectively. NFC production volume, reported in the Florida Department of Citrus Processor Report as chilled juice from fruit, remained relatively unchanged during these six seasons, while FCOJ production levels decreased by more than half during years of smaller crops. The variation in these production levels suggests that changes in fruit production levels ac ross years leads to changes in FCOJ production levels but does not change NFC levels. One explanation for this result is that fruit inputs go into NFC production first, and the remaining fruit supply is processed into FCOJ. This would suggest that statew ide NFC annual production of approximately 575 million gallons is optimal for processors. A pproximately 225 million gallons of 48 degree Brix equivalent FCOJ were produced during the 2003 04 season. While not all plants open during this season are still operating, this processed volume estimates a feasible statewide capacity for the quantity of FCOJ that could be processed if fruit input were available. The fluctuations in FCOJ production suggest that it is optimal for FCOJ producers to process available fruit inputs up to this volume of output. Storage NFC juice or concentrate enters a buffer storage tank at the completion of extraction. Production managers then determine the method of storage to be used until the juice leaves the plant. Due to the sea held in long term storage for later sale. FCOJ is stored long term in either large bulk tanks or 55 gallon drums (Hui, 2004). Bulk tank storage is the most common and least costl y form of concentrate storage. With FCOJ production down substantially the past few years ( Figure 2 10),
56 bulk storage capacity is readily available. FCOJ Bulk tanks typically hold between 250,000 and 1 million gallons of concentrate (Enerfab, 2009), equi valent to 1,750,000 and 7 million gallons of single strength juice. NFC juice is commonly stored in large aseptic bulk storage tanks, typically holding 1 .0 million to 1.8 million gallons of single strength juice. Some storage tanks are physically capabl e of holding either end product, but the sevenfold increase in volume required to hold a comparable quantity of NFC usually necessitates product specific storage tanks. Cost to construct new bulk aseptic NFC storage tanks has been estimated from $2.00 per gallon to as high as $4.00 per gallon, including all setup costs. 11 12 Converting FCOJ storage tanks to NFC storage tanks is possible, and in some cases more cost efficient, than constructing new tanks. 13 The higher volume requirements of NFC storage incre ase the cost of storage, as compared to bulk FCOJ storage costs. This higher cost creates an incentive for processors to store only the minimum volume of NFC required. Processors therefore attempt to directly bottle as much NFC as possible, instead of mo ving it into long term storage. Flows of fruit inputs are managed throughout the season to allow for direct bottling and to reduce the volume of juice entering bulk aseptic storage, including holding fruit on the tree until late in the processing season. FCOJ and NFC bulk storage tanks are constructed in groups of four or more tanks housed within a single refrigerated enclosure called a tank farm. 14 Processors usually operate one or more tank farms year round to facilitate the distribution of stored juice independent of processing operations. Storage tanks are also used to hold additional product used to insure the processor 11 Personal communication with anonymous manager (11) of a Florida orange processing facility, September 2007. 12 Personal communication with anonymous manager (1) of a Florida orange processin g facility, October 2007. 13 Personal communication with anonymous manager (1) of a Florida orange processing facility, October 2007. 14 Personal communication with anonymous manager (2) of a Florida orange processing facility, July 2008.
57 against an inventory stock out, in case input supplies decrease. Extra product in storage, referred to as carryover inventory, is me asured as the juice volume remaining in storage at the beginning of the processing season. Carryover inventory may also be required for blending with juice during the following season. The cost to store this carryover product is therefore considered a fi xed cost of doing business. Off site commercial storage facilities are another storage option for processed product. While storage at off site facilities is not commonly used, processors making delivery to the futures market can deliver to an Intercontine ntal Exchange approved tank facility. Handling and storage rates for 15,000 pound solid FCOJ contract at these facilities are $500 per contract for handling and $300 per contract per month for storage. This monthly storage rate is equivalent to $0.02 per pound solid per month. Before NFC futures contracts were delisted in December of 2007, published NFC storage costs ranged from $1,440 to $1,500 per month per 12,000 pound solid contract, or about $0.125 per pound solid per month (Intercontinental Exchang e, 2008). This six fold cost increase from FCOJ to NFC is due to the increased storage volume required to hold a pound solid equivalent volume of single strength juice. Similar to processing and extraction capacities, the storage capacity of individual pr ocessing facilities is proprietary information and therefore not available. Estimates can be developed for the total state capacity using the previous volumes of product reported in storage facilities. Figure 2 11 shows the gallons of c hilled j uice on ha nd from 2000 01 through 2005 06 (FDOC, FCPA, 2007 ). This quantity represents the volume of NFC in storage each week during the six year period. The maximum reported volume was over 320 million gallons during the 2003 season. The following two seasons sa w major decreases in fruit production but NFC production experienced very little decreases, peaking at 310 million and 290 million gallons
58 during 2004 and 2005, respectively. FCOJ volumes on hand peaked during the 2003 season at almost 200 million gallons of 48 degree Brix concentrate during the same period ( Figure 2 12). Gallons of concentrate on hand at the end of the six season period were at their minimum of about 70 million gallons. These inventory levels suggest that statewide NFC storage capacity is at least 320 million gallons, and that FCOJ capacity is about 200, both varying due to storage capacity changes since 2006. Figures 2 13 and 2 10 show storage and production levels from the 2000 01 through 2005 06 seasons for NFC and FCOJ respectively. Again, NFC storage and production are relatively consistent despite changes in input supplies, while FCOJ levels vary widely. Potential Changes to Industry Coordination and Operation Industry wide adoption of mechanical harvesters has the potential to c h ange the coordination and operation of the Florida orange juice processing industry Harvester adoption could result in changes to the volume of fruit harvested during any given time period, the length of the harvest season, the harvest labor market, grow er costs, scheduling of load allocations, processor costs, plant capacities and bottlenecks, the incentive structures for NFC and FCOJ production, and the late season harvest. The remainder of this chapter d escribes each of these potential changes and the expected impact on industry operations and total industry returns by addressing the views and concerns of individual growers and processors. While much of the data used to substantiate these concer ns is qualitative, the views expressed by industry decisi on makers are important factors in the potential adoption of mechanical citrus harvesters. Changes in Fruit Harvest Volume Mechanical harvesters are capable of picking large volumes of fruit in a short period of time. This productivity advantage allows a mechanical harvesting crew to pick about four times the volume of a typical hand harvest crew, with a greater than tenfold increase in per worker
59 labor productivity (Roka and Emerson, 1999; Roka and Hyman, 2004). Volume increase afforded through mechanic al harvesting brings about additional production challenges however As the percentage of mechanically harvested fruit increases, growers and processors will adjust their operations to take full advantage of the additional harvesting capacity from mechan ical units. The added capacity allows for larger volumes of fruit to be harvested during times of peak fruit maturity using only a fraction of the laborers need for hand harvest. This added capacity could also be beneficial if hand labor becomes scarce a nd wages increase. Potential disadvantages from the added harvest capacity include extra costs associated with handling higher harvesting volumes at a single grove site and the increased cost to mechanically harvest small groves. Higher daily fruit volu mes from a single grove could increase the cost and difficulty of fruit hauling. While a typical hand crew needs only three fruit trailers per day, one trailer every two to three hours, a typical canopy shake crew needs one to two trailers per hour. This fourfold increase in trailer activity within a single grove can be difficult to manage, possibly leading to greater transaction costs from dealing with multiple hauling companies. If trailers are not available for roadsiding fruit, all harvesting operati ons must cease. The high cost of harvesters and skilled laborers makes such a shutdown costly. The current piece rate harvest labor system avoids the high capital investment and payment to idle workers. Mechanical harvesters are designed to harvest lar ge volumes of fruit. The most cost efficient operation requires large, continuous groves, avoiding extra shutdown and transportation expenses. Harvesting small, fragmented groves is often much less efficient when using high capacity mechanical harvesters A small grove can be mechanically harvested quickly, but the downtime required for mach ine setup and disassembly, over the road transportation, and
60 cleaning greatly decreases productivity. The currently used hand harvest system is easily scaled to fit any size grove. Harvest Season Length The addition of mechanical harvesters could add additional harvesting capacity to the industry. Total industry harvest capacity (the summed capacity of mechanical harvesters and hand harvesters) could then increase i f the industry adds mechanical harvesters while maintaining current hand labor. This additional harvest capacity has the potential to alter the traditional timing of the harvest, leading to changes in the length of the harvest season. Mechanical harveste rs may enable the industry to lengthen the harvest season by picking more fruit later in the season, a time when hand labor traditionally migrates nort h to harvest other crops and citrus labor becomes difficult to secure in Florida Mechanical harvesting may also allow more flexibility in scheduling the harvest. Mechanical h arvester crews are much smaller and use highly skilled laborers specialized in the operation of harvesting equipment. Crew members are paid higher wages than hand laborers and possess skills that are not as easily transferable to other agricultural crops. An upward sloping labor supply curve would predict that mechanical crews would be more willing to supply additional labor hours at the increased hourly wages. These high skilled lab orers could produce a more predictable harvest capacity and the ability to harvest on weekends, holidays, during inclement weather, and potentially at night. Any additional harvesting flexibility gain from the adoption of mechanical harvesters could enabl e increased harvesting volumes earlier and later in the season, as well as planned shutdowns between peak harvest times. Conversely, mechanical harvesting could facilitate a shorter harvest and juice processing season. The increased capacity allows more f ruit to be harvested in a shorter period of time, reducing the time required to harvest the annual crop. Some Florida processors have expressed
61 concerns about decreasing the length of the harvest season, citing the possibility that unit processing costs c ould increase. 15 If the Florida orange harvest season, typically from November through June, was shorted or segmented, a larger volume of fruit would have to be processed weekly. The increased weekly volumes would require more intensive processing and add itional juice storage capacity, possibly forcing additional capital investments in juice extraction equipment and storage facilities. Processing plants have been designed and equipped to process a specific fruit volume determined in large part by the comp crop volume, and the historic length of the harvesting season. Some flexibility is incorporated into this design, but changing the season length would significantly impact the conditions under which these plants a re co nstructed. Additional storage costs from changing the length of the harvest season may be borne by the processors. 16 Each week of reduced processing requires an additional week of storage capacity to maintain annual juice output levels. For example, if a processor expects to bring in one million gallons of NFC during the last week of operation, and the season is shortened by a week, then this fruit must be processed and stored a week earlier. Processing the juice earlier will require that an additional one million gallons of storage capacity be available due to the shortened processing season. Statewide, the additional storage capacity needed to hold four weeks of extra inventory is about 40 50 million gallons. The estimated cost to processors to const ruct the extra storage capacity required to hold this increased inventory could be as much as $100 million. 17 15 Personal communi cation with anonymous manager (11) of a Florida orange processing facility, April 2006. 16 Personal communication with anonymous manager (2) of a Florida orange processing facility, July 2008. 17 Personal communication with anonymous manager (11) of a Florid a orange processing facility, April 2006.
62 Changing the length of the processing season could result in juice quality changes due to changing the date of harvest. A change in juice qualit y would force processors to adjust blending operations to ensure a consistent output quality. Such production changes could be costly for processors and viewed negatively unless these additional costs are offset. One such offset may be the additional ben efits gained from receiving a more homogeneous, higher quality input product that would require less blending. Some processors believe that the processing cost savings from harvesting more fruit closer to its optimal juice quality and operating their plant s for fewer weeks could outweigh the added cost of juice storage. 18 One central Florida processor argued that harvesting quickly and at the peak of fruit maturity will result in juice production of 19 The addition of mechanical harvesters could allow for this shortenin g of the harvest season and picking of fruit for optimal juice yields. Searcy, Roka, and Spreen (2008) show that harvesting and processing at biologically optimal times can yield greater pound solid volumes per acre of grove. Grower revenues, based on pound solid production, would increase as additional pound solids are captured. In addition to greater pound solid production, gro wers may also experience less fruit loss because of the shorter harvest season. Fruit left on the tree after May runs an increased risk of being damaged or removed by weather or pests than fruit harvested and processed earlier in the season. The faster h arvest could therefore reduce the higher rates of fruit loss seen late in the harvest season due to the increasing rate of fruit drop. Finishing the harvest earlier may also lessen the effects of alternate bearing characteristics seen 18 Personal communication with anonymous manager (1) of a Florida orange processing facility, May 2006. 19 Personal communication with anonymous manager (1) of a Florida orange processing facility, May 2006.
63 in Valencia oranges, where leaving fruit on the tree late in the season reduces the following Mechanical harvesting will not necessarily dictate the length of harvest season, but could instead offer flexibility as to how and when fruit is harves ted, potentially increasing total pound solids captured and reducing operating costs. Industry players have differing expectations of the costs and benefits associated with changing the length of the harvest season. However, an analysis estimating the to tal impact on the processing industry as a whole has not been published. By summing all gains and losses to the industry from a change in harvest season length, predictions can be formulated estimating the individual an d cumulative financial impacts o n in dustry players. Labor Supply and Demand Mechanical harvesting technology is a substitute for hand harvest labor. I ncreased use of mechanical harvesters w ill therefore decrease the demand for hand harvest labor. Economic theory predicts that a decrease i n the demand for labor will lower wages paid to hand labor and push members of t he current labor force to seek others m eans of employment, most likely outside the citrus industry. In addition to the negative welfare impacts expected on hand harvest labore rs due to mechanization of the harvest (Schmitz and Seckler, 1970), concerns have arisen among some growers and processors that the reduction in labor demand may decrease to the point that laborers will not have sufficient employment to remain in agricultu ral labor. Remaining hand picking crews may not be large e nough to harvest fruit that can not be harvested mechanically. These industry players fear that w hile m echanical harvesting could lesse n the total labor required for harvest, there may be som e repe rcussions for growers who cannot use the
64 mechanical harvesters 20 Growers who do n ot or cannot use mechanization c ould fa ce increasing labor costs as a result of labor scarcities brought about by the adoption of harvesters For example, m echanical harvest ing may not b e cost efficient for some growers due to grove size, loca tion, or configuration. Growers owning trees in which mechanical harvesting is neither cost efficient nor physically possible with th e current technology may then face an increase in ha rvest costs. Such a scenario assum es that there is a critical quantity of fruit that must be hand picked to keep workers from leaving the citrus industry. Without this steady level of employment, current harvest labor ers will be forced to seek employment el sewhere, reducing the available supply of hand harvest labor. In this scenario, additional costs are imposed on non adopting growers due to other growers mechanical harvesters The potential reduction of labor and increased costs o f such a scenario is a real concern for some in the citrus industry but is outside th e scope of this research Grower Costs and Operations Cost savings to growers from mechanical harvesting have been estimated at up to half of current harvest costs, savi ng growers $0.60 to $1.35 per box harvested (Roka, 2007). Some of these gains could be offset, however, due to additional costs imposed on growers using mechanical harvesting. Additional costs include grove preparation costs and indirect costs from chang es in harvest timing. New costs arise because mechanically harvested g roves are managed differently than groves harvested by hand. Preparing a n established grove for mechanical harvesting requires retrofit ting the current irrigation system, raising tree skirts, and hedging and topping trees. Rouse and Futch (2007) estimate the cost of skirting, hand pruning, and brush removal at $66 to $96 per acre, and the cost of moving irrigation systems at $50 to $60 per acre. 20 Person al communication with anonymous manager (8) of a Florida orange processing facility, July 2006.
65 The irrigation changes are one time cos ts, while tree cutting costs are recurring. A more costly option is to raze existing groves and plant new groves specifically designe d for mechanical harvesters. In addition to typical replanting costs, mechanically harvested groves should be planted wit h high headed trees, costing an additional $0.50 to $1.00 per tree (Rouse and Futch, 2009). Additional costs to the grower could also in clude increased transactions costs from dealing with a new harvesting company that uses mechanical harvesters or incre ased safety training and injury risks associated with working around additional farm machinery Growers are further concerned that mechanical harvester adoption will les sen their control over harvest timing and alter the current system of fruit delivery ba fruit balance. The c urrently accepted system of fruit delivery allows growers to harvest a percentage of their remaining fruit with hand crews at staggered time period s throughout the season with the quantity delivered at ea ch harvest date determined by the percentage of their fruit remaining on tree 21 This allows the grower to sell fruit at several different fruit maturity periods producing a total juice yield and quality level averaged across the harvest season Mechanic al harvesting could force a change to this system requir ing that all trees in a single location be harvested and processed within a few days annual crop being harvested within a few days. If this harvest were to ta k e place at a less than optimal time during the season, the grower may see significant financial losses whereas harvesting at the optimal would yield higher returns than the average harvest schedule The additional variability imposed by mechanical harve sters due to the timing of harvest and leading to f ruit quality changes or lower recovery rates is not viewed favorably by growers 21 Personal communication with anonymous manager (6) of a Florida orange processing facility, April 2006.
66 The potential threat of industry wide adoption of mechanical harvest may be holding down hand harvest prices. While the rel ationship between growers and hand harvesters in the Florida tomato industry continues to be an issue of contention (Palacios, Emerson, and Vansickle, 2006), the citrus industry has recently managed to avoid such heated labor disagreements. One difference between the two industries that could be allowing citrus growers to avoid such confrontations is the viable threat of mechanizing the citrus harvest. The hand harvest labor forces are similar between the two crops but the technology for mechanical harves ting of tomatoes for fresh consumption is not nearly as advanced as that for citrus. This threat of mechanization could therefore be allowing citrus growers additional market power in the harvest labor market, depressing citrus harvest labor wages. Load A llocations Efficient operation of mechanical harvesters will require changes in the system of assigning load allocations and scheduling delivery of fruit. The cost structure of m echanical harvesting requires large quantities of fruit to be harvested daily from a single geographic location. Transporting machines over the road is costly and requires extensive setup and cleaning, in addition to reducing the time available for operation. This is a change from the current system of scheduling and may not be a allocating loads. The current system of assigning load allocations is based on allowing each grower with outstanding shares of fruit to deliver a fraction of their remaining supply. This allocation system was developed under hand harvesting technology to take advantage of the highly divisible piece rate cost structure of hand harvesting. Weekly allocations, based on the input volume needed tree supply, are viewed by industry pla yers as an acceptable way of conducting business. Growers accept this system because seasonal risks and quality
67 changes are spread across the season. Processors approve because they retain control of input flows, and weekly input volumes are smoothed to avoid large swings in volume. Replacing the current load allocation system with a harvest schedule that optimizes mechanical harvester operation could increase costs for both growers and processors. The most obvious explicit cost would be the loss from harvesting fruit during less than optimal time periods. Elimination of the remaining balance scheduling system would also be seen as a cost to growers if mechanical harvesting forces harvest of all adjacent groves during a single time period. Processing plants, currently spreading load allocations across a large number of growers and grove locations, will have to accept fruit from a limited geographic region to allow harvesters to operate near peak efficiencies. While some processors believe a larger har vest from a more concentrated area will have no significant effect on fruit quality, other processors are concerned juice quality will suffer due to the processing of immature fruit. 22 23 Processors are in agreement that they must maintain control over frui t input volumes. Plant managers maintain the need for flexibility to increase or decrease input flows and avoid entering into guaranteed delivery contracts 24 These adjustments allow for plant managers to utilize input flows to optimize weekly and seasona l processing operations. Developi ng a new load allocation system that allows growers to retain the benefits of the remaining balance system and allow s pr ocessors to control input flows would encourage the adoption of mechanical harvesters. Processing Co sts and Operations The adoption of m ec hanical harvesters could change the current methods of fruit procurement, scheduling and processing, resulting in increased processing costs In an industry 22 Personal communication with anonymous manager (3) of a Florida orange processing facility, May 2008. 23 Personal communication with anonymous manager (11) of a Florida orange processing facility, April 2006. 24 Personal communication with anonymous manager (11) of a Florida orange processing facility, April 20 06.
68 that constantly strives for cost minimization, additional p rocessing costs and costly operational changes are not expected to be acceptable without compensation. Increased costs may arise from excess debris removal from mechanical harvested loads, more intensive pro cessing requirements, and increased storage capa cities. Additionally, changes in load scheduling will receive resi stance from processors if it reduces their control over input flows and the management of processing operations. Managers are comfortable operating unde r the current input scheduling syste m, and relinquishing control of input supply flows could expose them to fluctuations in input volumes, resulting in increased processing costs 25 Processors accepting mechanically harvested loads face the added costs of removing and disposing of excess le af and tree litter from the fruit Loads of mec hanically harvested fruit can include an increase in the amount of debris of 20% to over 250%, as compared to hand harvested loads (Spann and Danyluk, 2010) 26 The processor must deal with the se additional le aves and sticks before fruit can be processed The add ed cost s include the expense of extra line graders to remove the material from the fruit and the expense of removing the material from the facility 27 Increased harvesting capacity afforded by mechanic al harvesting could push processors to increase juice extraction and storage capacities, directly increasing their cost of operation. Adding extraction and storage capital requires large upfront investments on the part of the processor. These capital out lays are not viewed favorably by processors, and agreeing to such investments will require the potential for a favorable return on those investments. Additional returns, however, do not necessarily accrue to processors through the adoption of mechanical 25 Personal communication with anonymous manager (11) of a Florida orange processing facility, April 2006. 26 Personal communication with anonymous manager (10) of a Florida orange processing facility, July 2008. 27 Personal communication with anonymous m anager (1) of a Florida orange processing facility, May 2006.
69 h arvesters. This lack of a direct financial benefit to processors will deter processors from completely backing industry wide adoption of mechanical harvesters. Processors are not expected to willingly accept additional processing costs without the potent ial to offset the se cost s through increases in revenues. While processors have not yet refuse d mechanically harvested loads their financial impact from accepting mechanical harvesting needs to be further explored. 28 While this analysis does not attempt t o estimate the change in processing costs from accepting mechanically harvested loads, it does estimate financial gains to the industry from mechanical harvesting that could then be used to offset added costs. Capacity Constraints Adding mechanical harves ters to current industry operation could increase weekly harvest capacity. The degree to which this harvesting capacity increase is financially beneficial depends on the processing capacity available to handle additional fruit. Potential bottlenecks cons training juice production include the limited capacity of hand harvesting, trailer availability, evaporator capacity, processed juice storage capacity, and feed mill capacity. Mechanical harvesters allow for more harvesting possibilities when large hand cr ews are not able to harvest. For example, hand crews are limited to working during daylight hours and only when weather conditions are suitable. As daily temperatures and the frequency of rain increase toward the end of the harvest season, hand harvest b ecomes more difficult. Demand for harvest labor also begins to increase for other agricultural crops in northern states, pulling laborers out of the citrus labor market. It may be possible for mechanical harvesters to operate in adverse weather condition s, on holidays, or even at night. This additional harvesting capacity is 28 Personal communication with anonymous manager (1) of a Florida orange processing facility, May 2006.
70 only beneficial if other constraints limiting production are not binding. The availability of bulk fruit trailers to deliver harvested fruit from the grove to the plant is one such potentially constraining factor identified by industry players. Managing the available trailers becomes more difficult when output increases from a single region and can lead to harvesting delays. 29 Such delays have been experienced with harvesters runni ng as few as six to seven hours per day. 30 Increasing the availability of trailers could be accomplished through more efficient trailer management, faster unloading times at the processing facility, or increasing the quantity of operational bulk trailers. Bottlenecks in plant operation eventually occur as daily fruit volumes increase. These bottlenecks can arise at any step in processing, depending on plant size and configuration, but evaporation, storage, and feed mill operation have been identified as probable bottlenecks due to increased processing intensity. Processing facilities that produce large quantities of FCOJ are often constrained by evaporator capacity. Increasing the volume of weekly fruit inputs requires more intensive evaporation, either from operating evaporators more hours or running additional evaporators. This increased fruit volume could begin stressing evaporation capacity, which is based on the pounds processed per hour. Given lower FOB prices for FCOJ compared to NFC, it is unli kely that processors will increase capital investments in extraction capacity, instead investing in NFC processing capacity. Increases in fruit inputs at processing plants primarily producing NFC are most likely to cause bottleneck s in juice storage facili ties 31 Figure 2 13 shows the combined industry annual storage of chilled product peaking at about 300 million gallons at the end of each season from 29 Personal communication with anonymous manager (4) of a Florida orange processing facility July 2006. 30 Personal communication with anonymous manager (5) of a Florida orange processing facility, July 2006. 31 Personal communication with anonymous manager (6) of a Florida orange processing facility, April 2006.
71 2001 to 2006, regardless of the annual crop size. This trend suggests that total industry capacity for NF C is approximately equal to this m aximum storage value. Increased processing intensity due to mechanical harvesting may require processors to increase NFC storage capacity. Additionally, some processing managers attempt to bypass long term storage by dir ectly packaging a large portion of processed product. At peak periods during the season, juice going direct to the package can account for up to 70% of daily production Increased daily or weekly fruit input volumes may require such large portions of dai ly production be directly packaged for much of the season to avoid capital investment in long term storage. Another common bottleneck in many citrus processing plants is the feed mill The feed mill converts solid wastes remaining from production into pel letized cattle feed, allowing fruit inputs to continuously flow into the plant. If waste products are not moved quickly through the feed mill, juice production must be slowed or stopped because there is no place to store newly produced waste A breakdown of feed mill equipment can lead to a complete stoppage of operation. 32 Feed mill capacity dictates the volume of fruit inputs entering the plant during any given time period. The natural gas powered dryers used in feed mills at most processing facilities are large, indivisible factors of production. Because of the high expense of operation, processors attempt to run feed mill dryers as efficiently as possible, typically at full capacity. Processors adjust fruit input levels so that the expected waste vo lumes match the number of dryers in operation, maximizing dryer efficiency. Under such processing conditions, a small increase in input volume may be avoided if it requires operating an additional dryer to process the extra waste. 33 32 Personal communication with anon ymous manager (11) of a Florida orange processing facility, April 2006. 33 Personal communication with anonymous manager (11) of a Florida orange processing facility, September 2007.
72 The capacity of juice extractors does not usually limit total plant capacity because processors typically have the ability to adjust the quantity and speed of extractors. E xtractor capacity is considered in which the number of pieces and size of the inc oming fruit determines capacity. The number of extractors needed is set to match the expected fruit volume and fruit size. For example, if fruit being processed is smaller than expected the number of pieces per box will increase and the number of extrac tors set to process small fruit may need to be increased. Increasing the number of small pieces of fruit and not adding small fruit processing capacity could l ead to a plant slow down, decreasing juice yield. Late Harvest Limitations Mechanical harvesti ng of Valencia oranges after mid May is problematic because many of yields to such a large degree that mechanical harvesting typically ceases after May 15. 34 Abscission chemicals are currently being researched that lessen the negative yield impacts of late harvesting, but this technology is currently not commercially avai lable (Blanco, 2009). Until abscission chemicals are approved for use that reduce the force needed to remove mature fruit, all late harvestin g must be done by hand crews. The inability of mechanical harvesters to operate effectively during the late seas on Valencia harvest period is a hindrance to widespread adoption of mechanical harvesters. Growers are concerned that large scale mechanical harvesting early in the season could displace much of the hand labor force, leaving an inadequate labor force to h arvest fruit remaining after 34 Personal communication with anonymous manager (9) of a Florida orange pr ocessing facility, July 2006.
73 mechanical systems are forced to shut down. 35 One potential solution to the late season problem is to shorten the harvest season, ending the harvest by mid May. Some processors are concerned that ending the harvest early would significantly increase juice processing and storage costs. Juice from fruit traditionally harvested in late May, June, and July would have to be processed sooner and stored by the processor, instead of being stored on the tree. Additional juice storage capacity would be require d capacity, potentially requiring large capital investment by processors. Summary The current system of operation has served the Florida orange juice industry well for the past few decades. Producers continually reexamine these operational practices in an effort to remain competitive in world markets, looking for opportunities to lower their overall costs. Converting from a labor intensive hand harvest to a capital int ensive mechanical harvest has the potential to lower production costs. While several hindrances continue to stand in the way of mechanizing the harvest, necessary industry changes allowing for adoption of mechanical harvesters may allow the Florida citrus industry to remain competitive in world markets. The following chapter examines how some other agricultural industrie s have addressed the decision to adopt new harvest technology. 35 Personal communication with anonymous manager (8) of a Florida orange processing facility, July 2006.
74 Figure 2 1. F resh and processed orange utilization in Florida from t he 1988 89 season to the 2007 08 season, boxes ( USDA, FASS, 2009 a ). Figure 2 2 Florida c itrus i ndustry p rocessing i ntensity for 2005 06; monthly processing as a percentage of total season production (Florida Citrus Commission, 2007).
75 Figure 2 3. Flow of fruit and juice m ovement.
76 Figure 2 4 Pick and r oadside h arvest r ates for Valencias from a s ample h arvesting c ompany for 2003 04 s eason ( unpublished data collected by B Hyman and F. Roka through personal interviews wit h harvesters, 2004).
77 Figure 2 5. Process flow diagram of orange juice production, rectangles represent work stations and trapezoids represent storage points.
78 Figure 2 6 Steps in the production of canned frozen concentrate d orange juice (Matthews, 1994)
79 Figure 2 7. Chilled o range juice e xpressed in 's t andard g allons' (11.8 degree Brix) for 2007 ( Florida Department of Citrus Processor Report 20 08 ) Figure 2 8. Chilled juice, n et annual production from fruit, 2000 01 to 2005 06 (FCPA Statistical Summary, 2006).
80 Figure 2 9. FCOJ Net annual production from fruit, 2000 01 to 2005 06 g allons at 48 degree Brix (FCPA Statistical Summary, 2006). Figure 2 10. FCOJ production and s torage, 2000 01 through 2005 06 g all ons at 48 degree Brix (FCPA Statistical Summary, 2006).
81 Figure 2 11. Chilled juice, g allons on hand from 2000 01 through 2005 06 (FCPA Statistical Summary, 2006) Figure 2 12. FCOJ g allons on hand from 2000 01 through 2005 06 g allons at 48 degree Brix (FCPA Statistical Summary, 2006).
82 Figure 2 13. Chilled juice production and s torage, 2000 01 through 2005 06 (FCPA Statistical Summary, 2006).
83 Table 2 1. Estimated average picking, roadsiding, and hauling charges for Florida citrus 2003 04 thro ugh 2005 06 Fresh fruit Processed fruit 2003 04 2004 05 2005 06 2003 04 2004 05 2005 06 Picking Charges: ----$/Box --------$/Box ----Early and Mid Season Oranges 0.84 0.954 1.048 0.772 0.829 0.982 Valencia Oranges 0.84 0.938 1.094 0.791 0.87 1.055 Pink/Red Grapefruit 0.658 0.739 0.78 0.59 0.668 0.736 White/Marsh Grapefruit 0.633 0.744 0.741 ------Temples/Tangelos 0.95 1.163 1.167 0.851 1.043 1.203 Tangerines 1.563 1.529 1.598 ------Add for Spot Picking --0.314 0.175 ------Roadsiding Charges: ----$/Box --------$/Box ----Early and Mid Season Oranges 0.86 0.895 1.038 0.801 0.817 0.963 Valencia Oranges 0.868 0.899 1.074 0.817 0.836 1.011 Pink/Red and White/Marsh Grapefruit 0.755 0.847 0.835 0.62 0.793 0 .877 Temples/Tangelos 0.938 1.003 0.92 0.833 0.89 1.062 Tangerines 1.155 1.095 1.143 ------Hauling Charges: ----$/Box --------$/Box ----0 30miles 0.41 0.417 0.459 0.392 0.393 0.428 31 50miles 0.46 0.512 0.561 0.457 0.464 0.498 51 80 miles 0.553 0.573 0.697 0.53 0.515 0.632 81 100miles 0.625 0.64 0.792 0.57 0.632 0.735 100+miles 0.687 0.746 0.989 0.625 0.728 0.84 SOURCE: Ronald P. Muraro, University of Florida/IFAS, Citrus Research and Education Center, Lake Alfred, FL, August 2006
84 Table 2 2. Largest Florida orange juice processors o perat ing during the 2003 04 s eason. Processor Location Percentage of production Cargill Juice, North America Frost Proof, Avon Park 11.35% Citricuco, North America Lake Wales 9.26% Citrus Belle La belle 2.26% Lake Wales, Bartow 8.11% Cutrale Citrus Juices U.S. A Auburndale, Leesburg 16.30% Holly Hill Fruit Products Davenport 11.13% Juice Bowl Lakeland 0.58% Louis Dreyfus Citrus Winter Gardens, Indiantown 0.29% Pe ace River Citrus Products Bartow, Arcadia 6.66% Silver Springs Citrus Howey in the Hills 1.60% Southern Gardens Citrus Clewiston 9.22% Tropicana Bradenton, Ft. Pierce 23.25% SOURCE: R. Hunter, 2006, Unpublished data of annual production estimates for F lorida citrus processors Table 2 3. M inimum maturity standards for Florida oranges for p rocessing. Date Minimum standards August 1 October 31 50% color break Total soluble solids not less than 9.0% Brix/Acid Ratio as defined in s. 601.20 Acid Percen tage not less than 0.4% Juice Content not less than 4.5 gallons per box = A juice percentage of 43% as a weight/weight measure should meet minimum requirements of 4.5 gallons per box (Wardowski et al., 1999, p.3, Quality Tests for Florida Citrus) Novemb er 1 November 15 50% color break Total soluble solids not less than 8.7% Brix/Acid Ratio as defined in s. 601.20 Acid Percentage not less than 0.4% Juice Content not less than 4.5 gallons per box November 16 November 30 50% color break Total soluble s olids not less than 8.5% Brix/Acid Ratio as defined in s. 601.20 Acid Percentage not less than 0.4% Juice Content not less than 4.5 gallons per box December 1 July 31 Total soluble solids not less than 8.0% Brix/Acid Ratio as defined in s. 601.20 SOURC E: Florida Statutes, 2006. Title XXXV (Agriculture, Horticulture, and Animal Industry), Chapter 601 (Florida Citrus Code), 601.19 Oranges; maturity standards
85 Table 2 4 M inimum Brix/Acid R atio s tandards, as related to total soluble solids ; minimum r atios of solids to acid Total soluble s olids Minimum Brix/Acid R atio 10.50 to 1 10.45 to 1 10.40 to 1 10.35 to 1 10.30 to 1 10.25 to 1 10.20 to 1 10.15 to 1 % 10.10 to 1 10.05 to 1 10.00 to 1 9.95 to 1 9.90 to 1 9.85 to 1 9.80 to 1 9.75 to 1 9.70 to 1 nd < 9.8% 9.65 to 1 9.60 to 1 9.55 to 1 9.50 to 1 9.45 to 1 9.40 to 1 9.35 to 1 9.30 to 1 9 .25 to 1 9.20 to 1 9.15 to 1 9.10 to 1 9.05 to 1 9.00 to 1 SOURCE: Florida Statutes, 2006. Title XXXV (Agriculture, Horticulture, and Animal Industry), Chapter 6 0 1 (Florida Citrus Code), 601.20 Oranges
86 Table 2 5. Minimum maturity standards for Florida processed juice and c oncentrate Processed product S tandards for Grade A Juice Total Score minimum of 90 points Ratio from 12.5:1 to 20.5:1 Degree Brix minimum of 11.0 degree Recoverable oil maximum of 0.035% FCOJ Total Score minimum of 90 points Ratio from 12.5:1 to 19.5:1 Concentrate degree Brix minimum of 41.8 degree Reconstituted degree Brix minimum of 11.8 degree Recoverable oil maximum of 0.035% FCOJ for Manufacturing Total Score minimum of 90 points Ratio from 8.0:1 to 24:1 Concentrate degree Brix minimum of 20.0 degree Reconstituted degree Brix minimum of 11.8 degree Juice from Concentrate Total Score minimum of 90 points Ratio from 12.5:1 to 20.5:1 De gree Brix minimum of 11.8 degree Recoverable oil maximum of 0.035% SOURCE: U.S. Federal Register Table 2 6. Certain orange juice: r eported U.S. production capacity, production, and capacity utilization, crop years 2001/02 2004/05 Crop y ear 2001/02 2002/03 2003/04 2004/05 FCOJM: Capacity ( 1,000 pounds solids ) 1,031,378 970,967 1,063,520 972,247 Production ( 1,000 pounds solids ): 938,152 669,838 934,019 421,083 Capacity utilization ( percent ) 91.0 69.0 87.8 43.3 NFCOJ: Capacity ( 1,000 poun ds solids ) 614,262 674,674 627,120 718,393 Production ( 1,000 pounds solids ): 467,385 556,265 531,322 544,323 Capacity utilization ( percent ) 76.1 82.4 84.7 75.8 Total: Capacity ( 1,000 pounds solids ) 1,645,640 1,645,641 1,690,640 1,690,640 Productio n ( 1,000 pounds solids ): 1,405,537 1,226,103 1,465,341 965,406 Capacity utilization ( percent ) 85.4 74.5 86.7 57.1 SOURCE: US ITC, 2006, c ompiled from data submitted to Commission questionnaires
87 CHAPTER 3 CONCEPTUAL AND THEOR ETICAL FRAMEWORK: AN AP PROACH TO ANALYZING THE ECONOMIC IMPACTS OF ADOPTING MECHANIC AL CITRUS HARVESTERS Introduction The objective of this study is to evaluate coordination issues and assess the expected economic impact of adoption of mechanical citrus harvesters on the Florid a orange juice industry This analysis draws from and expands upon analytical techniques developed previously to investigate markets for perishable agricultural commodities used for processing, including oranges, tomatoes, asparagus, and grapes. First, a n outline of previously published research foundations for this study is presented. Then, specific examples of empirical estimation techniques used in evaluating similar problems are discussed. Lastly, assessment techniques used to evaluate estimation re sults are reviewed. The first part of this chapter addresses published literature on new technology adoption, operations research, and harvest operations. F undamental economic theory, the foundation upon which this analysis is built, is identified, and examples are provided of how this theory has been applied to similar problems. Additionally, because this analysis is part of a larger systems approach to investigating mechanical harvester adoption, previous examples are presented of researchers success fully using a systems approach to problem solving The techniques used in this analysis to estimate the economic impacts of adoption are then described. These techniques require an understanding of both mathematical programming models and the ability to properly define the explanatory variables used in the estimation. Previously published works using mathematical programming approaches to assess the economic impacts of harvest are presented, including models of the California tomato harvest, the Brazili an orange harvest, and the Washington asparagus harvest ( Durham, Sexton, and Song, 1996; Caixeta Filho 2006; Cembali et al., 2007) Economic models used in these and similar
88 estimations are described in detail. Expressions for total revenue, labor and no n labor processing costs, harvest costs, transportation costs, and product storage costs are replicated from available literature to be used as bases for the empirical estimations. Lastly, empirical results from this analysis are assessed under the theore tical framework of optimality and efficiency. The Kaldor Hicks compensation criteria used to assess the economic outcome of the adoption decision are explained in this section. The incremental development of these criteria is detailed, and a graphical ex ample comparing two different outcomes is presented. New Technology Adoption The theory of new technology adoption and diffusion of innovations has been reported on extensive ly with Rogers (1983) developing much of the foundation for this work. Rogers de fines diffusion as the process through which new innovations are communicated over time among individuals within a social network. T h r ough the innovation decision process i ndividuals learn of these new innovations, form an opinion about the innovatio n, d ecide to accept or reject, implement the innovation and finally confirm their decision. Adoption is then defined as the step in the innovation decision process where an individual decides to make full use of the in novation. The rate of adoption is descr ibed for both the relative speed at which an individual adopts the innovation and the relative speed at which a particular innovation is adopted. The relative speed with which individuals adopt a new product allows for the characterization of each adopter as to their innovativeness Rogers de fines innovators as the first individuals to adopt, followed by early adopters, early majority, late majority, and laggards. A cumulative frequency plot of individual adopters typically follow s an s shaped function o ver time, varying due to the unique characteristics of each new innovation. Characteristics determining the rate of adoption include the innovations relative advantage, compatibility,
89 complexity, trialability, and observability. Relative advantage, or pr ofitability, and compatibility are generally agreed to be the most influential attribute characteristics of an innovation in determining new technology adoption rates by U.S. farmers (Dixon, 1980). The time required for individuals to adopt has been furth er studied, with Bass (1969) presenting a growth model for the timing of new product purchase The relative speed with which an innovation is adopted is determined by the specific attributes of the innovation. Relative advantage, compatibility, complexit y, trialability, and observability are defined as the five characteristics of an innovation that determine the speed of adoption. The relative advantage of an innovation is the perceived improvement it provides as compared to the superseding technology. Estimating the degree of relative advantage that adoption of mechanical harvesting brings to the orange juice processing industry, expressed as decreased cost or increased profitability, is the central theme of the following discussion. Operations Research The operations research approach to optimization is used to measure the impact of mechanical harvesting to the Florida citrus industry. This framework is tailored to applied problem solving involving complex coordination structures within an organization or industry. These techniques apply naturally when examining the intricate workings of the citrus processing industry, melding grower, harvester, and processor financial objectives with the biological, mechanical, and logistical constraints. Hillier and Lieberman (1974) describe operations research style studies as usually beginning with extensive observation of the organization, with the goal of fully understanding the nature of its operational structure. From the knowledge gained through observation a nd investigation, a mathematical programming model is formulated to capture the fundamental characteristics of operation, incorporating operational aspects of the problem being addressed. The mathematical programming approach is commonly used for
90 economic problems consisting of a large number of variables and restrictions on those variables which can be imposed as inequality constraints. Resource availability and production capacity limitations are examples of constraints more easily expressed with math p rogramming than with differentiable relationships (Thompson and Thore, 1992). It can then be hypothesized that this economic model sufficiently represents an actual situation such that conclusions drawn from modeling results are assumed to be valid insigh ts to a real problem. McCarl and Spreen (2007) operations research techniques and modeling tools to be used as formal planning aids for agricultural production managers has increased in recent years (Ferrer et al., 2007). Ferrer credits increased adoption of more capital intensive agricultural production technologies and improved mathematical programming techniques as stimuli for the continued development of such tools. Referenced models include a planning model for the fruit industry (Masini, 2003) a delivery planning model for a large meat processor (Bixby, Downs, and Self, 2006) and a model to determine optimal postharvest handli ng of fresh vegetable crops (Alcotti, Araujo, and Yahya, 1997) The original operations research paper making the stretch into agricultural research was C. ( 1953 ) one o f the first reports to focus on adjusting industry operations to optimize returns from harvest. His initial observations of Seabrook revealed a production operations system in which the harvest logistics were poorly planned and inefficient. Harvest typic ally started too early and available capital was strained by un reasonable daily capacity objectives. Harvesting and processing schedules quickly fell behind. A poorly orchestrated harvest resulted in large portions of labor being overpaid for overtime wo rk, lost product due to lack of processing
91 capacity, and product yield far from optimal. After observing a season of operation, Thornthwaite addressed these problems by developing a harvest schedule that would optimize yield quantity and quality. He then adjusted farm operations to fulfill this optimal schedule. Processing capacity was the initial binding constraint to be addressed. Working backwards from a given and fixed processing capacity, estimates were developed for the crop acreage to be harveste d during each period. The targeted acreage of harvested crop coincided with processing and harvesting volume capabilities. A planting schedule was then developed to produce a crop of the desired acreage and quality at the scheduled time of harvest. The system ensured full utilization of harvesting and processing equipment, which resulted in increased output quantity, efficiency gains by optimizing planting sch edules, extending the harvest season, reducing capital requirements for harvesting and processing, and minimizing labor costs. first agricultural application of operations research field analysis, which until then had been used predominantly for military logistics. Kreiner defines field analysis as the practice of operations research in which an analyst directly observes operati ons and learns the specific science and engineering constraints of the business process. rvesting as a key cause of inefficient operation. While the industry as a whole viewed this inefficiency as an acceptable cost of operation, Thornthwaite recognized that a system of uncoordinated planting, harvest, and processing used was the root of the production problems. The operating model implemented by Thornthwaite resulted in improvements to the efficiency throughout production agriculture. His
92 production model is still used wholly, or in part, by present day agricultural managers to schedule dai ly operations. Another key characteristic of operations research is the broad systems approach used for problem solving, bringing together experts from a wide knowledge base to simultaneously research different aspects of the problem. This type of systems approach to problem solving was used successfully to mechanize the tomato harvest in California. As described by Rasmus s en (1968), the successful industry wide adoption of the tomato harvester required changes across all facets of industry operation. Th e complete overhaul of the processed tomato industry to incorporate mechanical harvesting technology within a few years was made possible because engineers, horticulturalists, agronomists, and irrigation specialists worked simultaneously to change industry operation in a way that was agreeable to both tomato growers and processors. Fearing an increase in the cost of labor, the California tomato industry was able to end their ithin ten years of harvester develo pment (Chern and Just, 1978). In addition to the technological advancements that allowed for industry wide adoption, processors subsidized the cost of incorporating the new technology within the initial crops and are cre dited with expediting the rate of adoption. Further, processors relaxed quality standards for raw tomato inputs and altered some aspects of processing to advance technological change. A coordinated systems approach to mechanical harvesting by the Califor nia processed tomato industry achieved economic results not possible by any single industry player (Schmitz and Seckler, 1970). Mathematical Programming Model The framework of operations research follows that information gained through observation is mathe matically formalized and analyzed with math programming techniques. Typical operations research problems consist of large sets of constraints and physical capacity limitations
93 resulting in many inequality constraints. The most straight forward way to sol ve such optimization is often through math programming. Examples of optimization with math programming are readily available in the agricultural harvest literature. These models differ from biological studies estimating factors effecting crop yield by in stead assuming a biological growth function and then optimizing economic returns from harvest. The following discussion outlines programming models used to analyze the California tomato harvest, the Brazilian orange harvest, and the Washington asparagus h arvest. The processed tomato industry has been studied extensively with r espect to optimizing production with industry wide adoption of mechanical harvesters (Brandt and French, 1983; Hopper and Folwell, 1999; Durham, Sexton, and Song, 1996; Huang and Sext on, 1996). Durham, Sexton, and Song provide one such example of optimizing the allocation of inputs for the California processed tomato industry using a mathematical programming model. The ir study compares the returns to the industry when operating under an estimation of the actual allocation of inputs verses returns when operating under an estimated efficient allocation, reporting the estimated gains or losses to the industry. Their analysis attempts to optimize input allocations between growers and pro cessors, and transportation efficiency. Durham, Sexton, and Song distinguish bet ween high and low quality products for both inputs and outputs. The quality of raw inputs is determined by the soluble solids concentration of the tomatoes and the quality of output products is determined by the type of production plant, producing either bulk tomato paste or a mix of diversified tomato products. The model allocates tomato inputs among processing plants based on facility size type and location, with the obje ctive of maximizing v ariable profits to the industry. Variable profits are defined to be the aggregate revenue from product sales less
94 processing and transportation costs. This formulation yields Equations 3 1 through 3 4, the mathematical programming mo del f or maximizing California processing tomato industry profits. Maximize (3 1) Subject to (3 2) (3 3) (3 4) In the profit function (Eq. 3 1), with t representing the weekly time period TR represents total weekly revenue from the sale of processed goods and TC is the total operating and processing costs per week with j representing a diversified process ing plant and k representing a tomato paste plant TTC is the total transportation cost per week where n denotes both diversified and bulk paste plants Equation 3 2 simulates processing capacity constraints, where XL is the tonnage of low soluble solids raw tomato inputs and i represent s the geographic region, XH is the tonnage of high soluble solids raw tomato inputs, and CAP represents the capacity of each processing plant For the low and high quality input volume constraints (Eq. 3 3 and 3 4), NL re presents the low soluble so lid input supply limitations and NH r epresents the high soluble solids input supply limitations. Caixeta Filho (2006) presents a similarly constructed linear programming model, shown in Equations 3 5 through 3 8, estimating retur ns from the Brazil orange harvest.
95 Maximize (3 5) Subject to (3 6) (3 7) (3 8) The Brazilian model maximizes a profit function (Eq. 3 5) estimatin g the total margin toward profit (M), defined as the sum of returns from the sale of orange juice solids (RSS), less transportation costs (CTOTTR) and harvest costs (CTOTCOLH) where PSS is the price of 1 kg of soluble solids, PROij is the v alue of product ion for grove i in month j, TSSij is the content of soluble sol ids for grove i in month j, Tij is the proportion of grove i harvested in month j CUT is the transportation cost, DISTi is the distance of grove i from the ind ustry, and CUCi is the harvesting cost of grove i The profit function is constrained by requiring the monthly volume of boxes processed (PROCMES) to be less than or equal to monthly industry capacity (CAP) (Eq.3 6), by forcing the Brix to acid ratio (RATIO) of processed juice to be with in a specified acceptable range (RATIOMIN, RATIOMAX) (Eq.3 7), and by allowing each grove to be scheduled for harvest (CRONOTAL) only once (Eq. 3 8). This model differs from the Durham, Sexton, and Song mathematical programming model because it does not i nclude any costs or revenues accruing to processors. Caixeta Filho instead focuses on grower returns, including revenues from the sale of juice and costs to harvest and transport fruit inputs. Including the harvest scheduling constraint allows the model to simulate the static, single harvest per season
96 nature of the citrus harvest. Inputs were viewed in terms of soluble solids in both models, but Caixeta Filho used an estimated pound solid production value for each month, instead of separating into high and low soluble solid inputs. A third harvesting simulation analysis reported by Cembali et al. (2007) maximizes a profit function for asparagus growers. Equations 3 9 through 3 11 detail returns to Washington asparagus growers in a model similar to the B razilian orange harvest model. Asparagus grower profits are defined as the revenue from sales of asparagus harvested during each time period (H(t)) less the harvest cost at each time period (Eq. 3 9). Maximize (3 9) Subject to (3 10) (3 11) Unit selling price is represented by p and the unit cost of harvest is represented by the function r (H(t)) expressing unit harvest costs as a function of harvest volume. The payable weight of the asparagus harvested during each time period is PW(CR(t)) and a discount rate is included in this formulation to account for the time value of returns. Equation 3 10 constrains the volume of carbohydrates removed during harvest, where W(CR(t) ) is the total weight of spears at time t, Additionally, Equation 3 11 requires that the spears available for harvest equal the difference between the number of newly em erged spears and the number of spears harvested. The bioeconomic model described by Cembali et al. simulates the daily asparagus harvest volume, yield, profit, and cost of harvest for Washington asparagus growers. This model accounts only for the grower profit, excluding revenue and costs associated with processing. An important
97 changes caused by the harvest. A dynamic modeling framework is necessary in this case b ecause only a few mature stalks are removed from the plant at each harvest. The harvest volume in a particular time period, therefore, affects the yield of subsequent harvests. The dynamic biological characteristics differ for agricultural crops harvest ed only once per season, as shown in the static modeling procedures used by Durham, Sexton, and Song, and Caixeta Filho. A dynamic model could, however, be used to analyze cross seasonal effects on the harvest of crops harvested only once per season. Whi le orange trees are harvested only once per season, the date of harvest may affect the following season and holding fruit on the tree may reduce the crop yield in the subsequent season. This alternate bearing of crop yields could be modeled by adapting the dynamic approach used by Cembali et al. to estimate asparagus yield. Explanatory Variables Programming models presented by Durham, Sexton, and Song Caixeta Filho, and Cembali et al. are designed to maximize profit functio n, subject to physical limitations. Each component of the three previously presented objective function s can be further describe d, detailing revenue calculations and each cost component. Understanding how profit components have been estimated with respe ct to different crops and different characteristics of production will aid in development of a function estimating orange juice industry profits. Durham, Sexton, and Song derive the components of total revenue by showing w eekly total revenue to be a funct ion of the value of diversified (D) tomato output per ton of raw inputs, P D the value of processed paste output (P) per ton of raw input, P P the tons of raw inputs, XL and XH and a plant type specific conversion factor for the increased level of output production
98 from high solids inputs, D and P Equation 3 12 represents weekly total revenue to the processor from processed product sales (3 12) Total revenues to the industry are therefore determined by summing of revenues from sales of high and low quality outp ut products. The revenues accruing solely to tomato growers cannot be determined in this model as they are in the Caixeta Filho model for Brazilian orange growers. Returns to Brazilian orange growers are shown in Equation 3 13 as a function of the price per kilogram of soluble solids (PSS), the value of the production function in boxes (PRO ij ), the soluble solid content per box (TSS ij ), and the portion of grove harvested (T ij ). (3 13) This formulation calculates revenue fro m soluble solid production by multiplying the price per kilogram solid by the monthly kilogram solid production in each grove. Processing costs are then defined by Durham, Sexton, and Song as the sum of labor and non labor costs to transform raw tomato in puts into either bulk paste or a diversified mix of processed products. These costs are estimated separately based specifically on the type of output product being produced For diversified production plants, labor costs were estimated by adjusting data from previous processing cost study (Logan 1984) to represent current costs for plants of various sizes. A continuous function for labor costs was then estimated by regressing a log linear average cost function on the point estimates of unit costs, w her e LC D represents the labor cost in a differentiated plant and X represents the sum med volume of low and high soluble so lid inputs This formulation yields the log linear unit cost function for labor shown in Equation 3 14.
99 ( 3 14) This function produces unit labor cost estimates that decrease logarithmically as the weekly volume of production increases. This cost function is consistent with the increasing returns to scale expected as the size of production facilities increase For bulk paste production plants, labor costs per unit output were estimated by assuming a less flexible system of operation requiring full capacity processing throughout the operational season. Because of the relatively fixed, continuous operation req uirement of paste processing plants, the function for average unit labor cost was assumed to more closely approximate a rectangular hyperbola. Labor cost per week for the bulk past plant could then be assumed equal to full labor costs per ton of raw input ( C* ) when operating (Eq. 3 15), and zero direct labor costs when not operating. (3 15) L abor cost as a continuous function of input is estimated by allow ing for the transformation shown in Equation 3 16, w here LC P approa ches the cost per ton as X becomes large and the parameter h is directly related to the rate of asymptotic approach (3 16) Unit labor costs were estimated separately for three different sized plants to account for scale vari ations. N on labor costs of production were also estimated by Durham, Sexton, and Song for California tomato production, again draw i n g costs from the report by Logan and updating to s base year, 1989. Non labor costs were est imated as a constant unit cost for each unit of input. Combining the functions for labor and non labor costs yields the total cost function for weekly operation of a diversified product processing plant (Eq. 3 17), where
100 DL DH are the estimated non labor unit costs o f processing low and high solid inputs respectively. (3 17) Combining labor and non labor production costs yields (Eq. 3 18) the total weekly cost function for a bulk paste plant ( 3 18) Caixeta Filho did not analyze orange processing costs. Harvest costs were included in an objective function of profits from Brazilian orange harvest. Harvest costs (CTOTCOLH) were defined as the unit cost to harvest (CUCi) a box of fruit from grove i times the volume of boxes harvested from each grove (PRO ij T ij ). Summing costs across all i groves and j time periods yields the total season cost of harvest, shown in Equation 3 19. (3 19) Chern (1969) attempted to estima te processing plant cost functions fo r orange juice processing in Florida. Chern estimated both discontinuous and continuous processing cost functions, describing the f available data and th e lack of previous empi the cost function of a food processing facility must be discontinuous due to the indivisibility of capital used in proces sing and fixed nature of the many input costs. This operation al structure p roduces a stepwise total plant cost curve that is not easily incorporated into a mathematical programming model. Like Durham, Sexton and Song, Chern regressed a continuous function across this stepwise cost data to estimate a continuous function approxima ting industry costs.
101 Using a dynamic approach to estimating harvest costs, Cembali et al. defined the total cost to harvest asparagus (Eq.3 20) as the volume harvested each time period (H(t)) times the unit harvest cost (r), where unit cost was a functio n of harvest volume (r(H(t))). (3 20) This formulation can account for the economies of scale often gained as larger quantities of an agricultural product are harvested from a fixed acreage. In such cases, the higher densi ty of product to be harvested will lower the per unit cost of harvest. This dynamic cost is realized in the citrus industry when growers are charged a higher per box price to pick fruit from a low yielding grove than a high yielding grove. Despite this t heoretical formulation of harvest cost, empirical estimations by Cembali et al. assumed a single fixed unit cost of harvest for both manual and mechanical harvesting. No explanation was given for assuming a fixed cost, but such simplification is often nec essary due to computational difficulties and limitations of available data. Finally, Durham, Sexton and Song estimated t otal transportation costs by c alculating the per ton cost, C i n to ship between each region. The total shipping cost can then be est imated by multiplying the cost schedule times the input volume shipped between each region, and summing a cross all regions. Equation 3 21 estimates the total wee kly transportation costs as a function of inputs. (3 21) Simila rly, Caixeta Filho estimates total transportation costs (CTOTTR) as the cost to transport all fruit to the processing facility, shown in Equation 3 22. (3 22)
102 CUTi is the cost function per box to transport fruit one kilomet er, DIST i is the distance of grove i from the processing facility, and PRO ij T ij is the volume of boxes harvested from each grove during each time period. Summing costs across all i groves and j time periods yields the total transportation cost. The costs associated with product storage are an additional cost component important for many food manufacturing operations While not presented in the previously outlined production models, storage of raw inputs, intermediate products, and finished products must b e addressed by production managers producing a wide variety of goods, including agricultural goods, chemical manufacturing, and fuel production. Storage capacity allows processors to smooth seasonal fluctuations in both input availability and output deman d, attempting to optimize facility operations (Hoffman, 1973). Research establishing the foundation of storage costs theory as it applies to food manufacturing is based historically on the theory of dams (Meyer, Rothkopf, and Smith, 1979 ) Dams are model ed under the assumptions of an unknown independent input volume into a reservoir of fixed capacity, and a controllable but constrained output volume. These assumptions from dam theory are then expanded to allow for modeling of the manufacturing process. Meyer, Rothkopf, and Smith list the three key components of modeling storage and inventory as output demand rate (D), input supply rate (M), and storage capacity (X). Figure 3 1 shows four possible cases when viewing storage over time, with varying outp ut rates, input rates, and storages availability. Storage replenishing rates (V) shown in each case are defined as the difference between the input rate and the output rate (Eq. 3 23). (3 23)
103 The cyclical nature of storag e is shown in each case as the volume of inventory on hand undulates over time, where T is defined to be the time of manufacturing operation and T' is defined as downtime. Case 1 shows storage management with seasonal carryover and inventory levels at ful l storage capacity for an extended period of time, requiring input and output volumes to be equal during the time of operation. If Case 1 represented an orange juice production facility, time zero depicts a situation near the end of a processing season du ring which all tank capacity is full and there is no change in storage volume. All output produced during this time production off season, processing ceases and p lant outputs come from stored juice at a rate of D. The beginning of time period T represents the start of the next production season and the difference between production and output is the rate of storage replenishment (V). The volume of juice remaining in storage at the beginning of the season, or inventory carryover, is equivalent to the total available storage volume (X), less the output rate times the length of the off season nventory carryover, conditions most likely realized during NFC producers. Production and storage in Case 2 is similar to Case 1 except that inventory levels are allowed to go to zero, defined as a stockout, during the period of operational shutdown. Stoc kouts are also seen in Case 3, but unlike Case 2 inventory levels never reach full storage capacity. Such cases are not typical of orange juice production, as processors typically avoid complete stockouts which result in the complete disruption of juice o utputs. Inventory levels in Case 4 never reach full capacity and are never fully depleted. The inventory level I o seen in Case 4, represents inventory carryover, stocks that remain in storage at the beginning of the next operational cycle. Current FCOJ inventory cycles
104 are most similar to Case 4, with a base volume of carryover and peak storage volumes below available capacity. Ross (1973) reviews some of the mathematical modeling tools processors use to evaluate storage capacity to make sound decisio ns as to the volume of storage best fittings their operational needs. Defining storage as a profit producing asset used to absorb fluctuations in products due to patterns in receipts, operational downtime, and seasonal storage; Ross explains that incorrec tly sized storage facilities, either too small or too large, will negatively impact profitability. Too much storage ties up excess capital, limiting future investments and increasing costs for holding excess inventory. Too little storage limits plant ope ration, potentially causing plant shutdowns, necessitating temporary and more expensive off site storage and additional management responsibilities. Limited storage capacity may also lead to fundamental changes in non storage related plant operations, suc h as shipping or receiving operations, resulting in decreased profits. In such cases, limited storage capacity can go unrecognized as the cause of profit reductions, while other plant functions are cited as the culprit. Ross explains that while storage c atastrophes such as overflow or stockout rarely occur at processing facilities, processors more often face the cost associated with avoiding catastrophes. These avoidance costs are much more difficult to assess because they are seldom calculated explicitl y and are not usually available or considered when storage capacity decisions are being made. In an attempt to value storage capacity, Ross briefly outlines a Monte Carlo simulation modeling approach that incorporates both the direct costs of building and maintaining storage, as well as indirect costs associated with using alternatives storage when capacity problems arise. This cost structure relates all important plant operating factors to the returns from operation and incorporates variability from unce rtain factors of production. While this model does not attempt
105 to optimize operation, it could be used to develop tools to evaluate the consequences of storage decisions. The agricultural harvesting literature contains vast examples of detailed mathematic al approximations of total revenues, labor and non labor processing costs, harvest costs, transportation costs, and storage costs, each of which is used to estimate grower or processor profits. Previously outlined examples represent only a small sample of the studies most pertinent to estimating the revenues and costs of the orange juice industry. Chapters 5 and 6 of this study build upon these previous studies by combining pieces from each model into a single profit function intended to simulate orange j uice industry profits. Maximizing this profit function under different sets of industry operational assumptions simulates optimal industry operation within each scenario, including various levels of mechanical harvester adoption. Optimality and Efficiency To accomplish the research objective of assessing the economic impact of adoption of mechanical citrus harvesters, an acceptable assessment method is needed to evaluate estimation results. Total changes in revenues and costs to the orange juice industry as a whole are necessary but not sufficient to assess the impact of adoption. The industry is not completely vertically integrated, such that a change in operation may result in increasing industry profits, with one industry group gaining while the other group loses. Determining the impacts on each group, growers and processors, is sufficient to assess the impact on the industry as a whole. One possible assessment method could involve comparing the utility of each party before and after adoption. The use of utility theory to value the economic impacts from technological advancement is problematic however, because of the ordinal nature of the utility concept and the inability to compare across individuals. Working to solve this problem, Vilfredo Pareto s uggested the concept of Pareto optimality, which states that a situation is optimal if by
106 reallocating payouts one party cannot be made better off without making the other party worse off (Pareto, 1906; Fonseca, 2007 ). This concept of efficiency was fur ther explained by Maurice Allais (1943) ( Fonseca, 2007 ), who explained that an allocation is a Pareto optimal allocation if a distributable surplus does not exist. The existence of a distributable surplus would suggest that economic actors could make cha nges to increase efficiency and capture additional surpluses. A somewhat weaker principle to Pareto optimality is the idea of Pareto improvements. A Pareto improvement suggests an incremental reallocation in which one party is made better off without mak ing the other party worse off. This principle makes no claims on optimality, only that the reallocation is favorable or more efficient that the original allocation. The central drawback to viewing economic efficiency and optimality in a Pareto framework i s that very rarely is it possible to make a change in which no party is made worse off. Sir John Richard Hicks (1939) and Nicholas Kaldor (1939) attempted to overcome this shortcoming by adding the idea of hypothetical compensation to the Pareto principle Kaldor proposed that a reallocation is more efficient if winners are hypothetically able to compensate losers, leaving at least on e party better off and neither party worse off. Similarly, Hicks explored the compensation approach and proposed that a re allocation would be considered more efficient if the loser was unable to bribe the winner into not reallocating. The Kaldor and Hicks compensation criteria argue that comparing different allocations across individuals is possible when allowing for hypothe tical compensation. Kaldor Hicks efficiency requires only that it is theoretically possible to compensate the parties made worse off, thus allowing the potent ial for Pareto improvements. For example, in the pure exchange Edgewor th Bowley box shown in Fig ure 3 2, the current allocation is E, where individual A is endowed with X A E and Y A E and individual B is endowed with X B E and Y B E Note that at allocation E individual A is on utility
107 cur ve U A (E) and ind ividual B is on utility curve U B (E). Now suppose a reallocation to endowment C, where individual A is made better off and individual B is made worse off. The Kaldo r compensation criteria support endowment C over endowment E because individual A could hypothetically compensate B by the amount X A C X A F y ielding a final allocation at F. Allocation F is considered superior to the beginning allocation of E because individual A has move d to a more preferential utility curve, U A (F), while individual B remains on the original utility function, U B (E). While li mitations do exist in use of these compensation criteria (Jhunjhunwala, 1974), these tools are used to assess the impact of economic changes (Schmitz and Seckler, 1970). Schmitz and Seckler apply the compensation tests proposed by Kaldor and Hicks to thei r social welfare analysis of the adoption of mechanical harvesters by the California processed tomato industry. They explain that the main losers in the case of the tomato harvester are farm workers, but that there are surely other losers not accounted fo r in their analysis. Their use of the Kaldor Hicks compensation test is therefore a necessary but not a sufficient condition for assessing total economic impacts of the technological improvement. Optimality and efficiency concepts will be applied to the e stimated modeling results from industry return under varying levels of mechanical harvester adoption. Estimated shifts in costs and revenues to each player under the given levels of harvester adoption are summed to yield industry returns. It is the asses sment of these total industry returns from which conclusions are drawn as to changes in industry efficiency due to mechanical harvester adoption. Summary This chapter briefly outlines the fundamental economic thought and previously reported research techni ques upon which this analysis is constructed. Combining and modifying the relevant aspects of literature presented in this chapter, investigative modeling techniques will be
108 developed to specifically analyze the Florida citrus industry. Using o perations research techniques for problem solving, this research will address the coordination and economic issues hindering the adoption of mechanical citrus harvesters as part of a larger systems approach to ha rvester adoption. The next chapter s present the detai led structure of an economic model simulating operation of the Florida citrus industry followed by an assessment of the results based on the previously presented economic theory
109 Fig ure 3 1. Four cases for storage and inventory (from Meyer, Rothkop f, and Smith, 1973 ) Figure 3 2 Kaldor compensation c riteria (from Fonseca, G., The Paretian System, The New School for Social Research, 200 7 )
110 CHAPTER 4 BIOLOGICAL CHARACTER ISTICS OF FLORIDA ORANGES The internal chemical composition of citrus fruit constantly changes as it hangs on the tree, but maturation ceases at the time of harvest (Kader, 1999). As soon as the fruit is removed from t he tree, ripening stops and fruit quality begins to decline (Jackson, 1991). Harvest timing is therefore crucial f or citrus processing because fruit characteristics at harvest determine the characteristics and value of the final product. To ensure the ca pture of large quantities of high quality juice, growers and processors attempt to schedule harvest to coincide with the optimal levels of key juice characteristics. Understanding the changing biological characteristics of fruit is therefore required to e stimate the economic consequences of altering the harvest method or harvest timing. The processing industry uses a few key characteristics to estimate juice quality and maturity. Soluble solid content, acid concentration, fruit weight, and juice weight ar e four of the most important characteristics that can be directly measured. From these measurements, the Brix to acid ratio, pound solids per box, and percent juice can be calculated. Additionally, juice color, flavor, and the presence of defects are imp ortant characteristics for juice processing and in determining the final value of the juice. This chapter details these biological characteristics affecting juice quality and quantity, and describes how each impacts the Florida orange juice processing ind ustry. In addition to quality characteristics, fruit drop is an important biological component of the orange harvest. Fruit drop through the natural abscission process is described where mature fruit drops, decreasing juice quantity yield. The chapter concludes with an estimation of how several of these characteristics change over the course of the season, including how pound solid production per acre changes specific to tree type, tree age, and growing region.
111 Biological Characteristics One of the mos t important biological characteristic indicating fruit quality and maturity is the concentration of total soluble solids in the juice, about 70% to 80% of which are sugars. Concentrations of these soluble solids, measured as the specific gravity of the ju ice and reported in degrees Brix, increase as fruit remains on the tree and serve as a key indicator of maturity (Davies and Albrigo, 1994). The concentration of total soluble solids has been shown to decrease in overripe fruit, but this decrease was show n to occur very late in th e harvest season (Harding Winston, and Fisher, 1940). Figure 4 1 plots all of the total soluble solid data collected by Harding et al. for Valencia oranges during the 1935 to 1938 seasons. While this research was conducted over 70 years ago, it remains as one of the only published articles reporting the biological characteristics of Florida oranges in June and July. The second order polynomial function regression line fitted to these data is consistent with the biological chang es expected during the season and yields a greater coefficient of determination than regressing a linear function. The high importance placed on the soluble solid concentration (degree Brix value) of juice is somewhat of a relic from the days when FCOJ was the dominate output product (Wardowski et.al, 1995). For FCOJ, concentrate of a known degree Brix and volume can be valued based on the volume of product it will yield when reconstituted. This quantification method works well for juice reduced to a stan dard sugar concentration but is less applicable for NFC. Despite the lack of a direct volume equivalent, the soluble solids measure remains a key indicator of NFC quality, and juice is still traded based on its total soluble solid concentration. The solu ble solids concentration of NFC juice is also important in juice blending practices to ensure a palatable end product.
112 In addition to sugar concentration, acid concentration is an important juice characteristic for both growers and processors. Acids, co nsisting primarily of citric acid, are measured and reported as the titratable acid concentration. The percent acid concentration decreases gradually throughout the harvest season (Davies and Albrigo, 1994). Figure 4 2 depicts the average acid percentage concentrations for Valencia oranges sampled in Florida by Harding et al. from 1935 to 1938 (1940). The regression estimating acid concentration is assumed to be a quadratic function and is shown over time to decrease at a decreasing rate, consistent with Harding et al. Early and mid season oranges show similar properties (Chen, 1990; Harding, 1947; Rouse and Atkins, 1953). Average fruit and juice weights are commonly used in conjunction with sugar and acid concentrations to evaluate oranges used for pr ocessing. Fruit weight represents the average weight of a single uncut piece of fruit and juice weight is the average weight of juice removed from a single piece of fruit by standardized means. Average fruit and juice weights are indicators of tree produ ctivity and are used to predict annual juice production levels. Juice weight per piece of fruit increases rapidly during early stages of fruit development as fruit size increases (Davies and Albrigo, 1994). Juice weight eventually reaches a maximum and then declines toward the end of the season as fruit begins to dry. An example of seasonal changes in the juice weight of Valencia oranges is shown in Figure 4 3, expressed in pounds of juice per 90 pou nd box of fruit, (Harding Winston, and Fisher, 1940). This figure shows the juice weight per box. Changes in weight per piece are expected to follow a similar pattern. A quadratic function is regressed across the sample data to estimate juice weight weekly during the season. Changes in juice weight for e arly and mid season fruit varieties follow a similar pattern but occur earli er in the season (Harding Winston, and Fisher, 1940).
113 The average fruit weight per piece follows a pattern similar to juice weight changes in which weight increases rapidly early in the season, reaches a maximum weight, and declines if fruit is left on the tree very late in the harvest season. Equation 4 1 mathematically specifies these changes in average fruit weight per piece ( FTWT ) as a function of time. (4 1) More explicitly, fruit weight per piece changes over time can be represented by the second order polynomial function shown in Equation 4 2. (4 2) Remaining consistent with the biological fruit characteristics, the quad ratic parameter ( FTWT ) is expected to be negative and the linear term ( FTWT ) should be positive. A function estimating juice weight is omitted here but is expected to have the same functional form. Estimated parameter values for either function will va ry with fruit variety, geographic location, and tree age, but the assumed functional form should remain constant for all oranges throughout the state. Data collected on soluble solid concentration, percent acid, f ruit weight, and juice weight are used to c alculate additional parameters representing juice quality. One of the most used calculated predictors of quality is the Brix to acid ratio. While no single quality measurement perfectly predicts fruit maturity, the Brix to acid ratio is commonly accepted as the most useful indicator of juice quality and flavor (Kimball, 1984). Brix to acid ratio measures the ratio of sugar levels to acid levels in the juice, calculated by dividing the degree Brix by the percent acid concentration. The Brix to acid ratio is low early in the season because the sugar concentration is low and the acid concentration is high. The ratio increases over time as soluble solids increase and acid concentrations decrease. Solids and acids are both changing to directly increase the Brix to acid ratio as the season progresses, causing the ratio to eventually increase at an
114 increasing rate. Figure 4 4 shows the average Brix to acid ratio for Valencia oranges sampled by Harding Winston, and Fisher (1940). All juice oranges grown in F lorida exhibit similar ratio changes (Westbrook and Stenstrom, 1964). Pound solids are the central volume measurement used by the Florida citrus industry for processed fruit inputs. Total pound solids are calculated by multiplying the total soluble soli ds, measured as degree Brix, by the juice weight in pounds (Jackson, 1991). Since pound solids make up the foundation of the fruit payment system, it is important to understand how pound solid volumes changes over the course of the harvest season. Figure 4 5 uses data collected during the 1936 to 1938 citrus seasons to show how average pound solids per box ( PS ) increase during the early part of the season, reach a maximum concentration, and then decrease late in the season for Florida Valencia oranges (Ha rding Winston, and Fisher, 1940). Similar changes in pound solids per box were reported by Ramirez (1977 ). Such biological changes over time follow the functional specifications shown in Equation 4 3. (4 3) Similar to fruit an d juice weight changes, pound solids per box accumulation over time can be represented by the second order polynomial function shown in Equation 4 4. (4 4) The functional form depicted in Equation 4 4 is consistent with expect ed changes in sugar production of oranges grown in Fl orida, as shown by Harding, Winston, and Fisher, and Rami rez. The quadratic parameter ( PS ) is expected to be negative and the linear term ( PS ) positive. The assumed functional form will remain consta nt for all oranges throughout the state, while the estimated parameter values will vary for specific fruit varieties, geographic locations, and tree ages.
115 Juice percentage is an important quality parameter that indicates both quantity and quality charact eristics of the jui ce. J uice percentage is a weight/weight calculation, determined by dividing the weight of extracted juice by the total fruit weight of the sample. Juice percentage increases as the season progresses and the fruit accumulates water. A maximum is typically reached late in the season, after which fruit begins to lose juice, causing the percentage of juice to decrease (Harding and Lewis, 1941). Because of the increasing nature of juice in the beginning of the production season, this para meter is used by the industry as an indicator of fruit maturity. Color, flavor, and the presence of defects in the juice are monitored as fruit matures. These three factors are important in determining the value of juice for retail sale and are used by t he USDA to certify the quality of processed product. Juice color and flavor change during the harvest season and across fruit variety. Color usually increases, or darkens, as the season progresses, similar to seasonal sugar changes (Sinclair, 1961). Jui ce from early season fruit is often a pale yellow, contrasting the deep red orange juice typically extracted from late season fruit. Flavor is an extremely complex characteristic of juice and is dependent on many factors, including the concentrations of s ugars and acids, ratio, and many other flavor components. Flavor typically increases in quality during the season as acid concentrations are reduced. Flavor quality will then decrease as the Brix to acid ratio increases exponentially near the end of the harvest season. The late season Valencia orange is commonly accepted as having more desirable flavor characteristics than early and mid season fruit varieties (Jackson, 1991). Finally, the presence of defects lowers the quality and value of juice. Juice defects include seed particles, discolored specks, or clouding and can be associated with either fruit quality or processing techniques.
116 Pre harvest fruit drop is a key determinant of fruit yield and juice production volume. Fruit abscises from a tree at different times of the season and for different reasons. Shortly after fruit set, some fruit naturally drops to lower the fruit count to a level the tree is able to sustain (Davies and Albrigo, 1994). This early drop, in Florida termed the June drop, is an unavoidable biological process and does not directly affect the harvest decision. Additional drop occurs throughout the season due to weather events, fruit damage, and biological changes experienced during later stages of ripeness and senescence (B orroto et al., 1981). For example, fluctuations between dry and rainy periods can increase fruit loss, especially in groves lacking adequate irrigation capacity. Heavy precipitation causes rapid uptake of water, splitting fruit and abscising it from the tree. It is this rate of fruit drop that is especially important to growers during the harvest season, because once fruit has matured and is ready to harvest, additional drop results directly in decreased yields. Growers therefore monitor fruit loss clos ely during this part of the season and make harvest decisions based on expected drop rates. Fruit drop is usually reported as the cumulative percentage of fruit abscised from the tree after a given date. August 1 is typically used as the baseline fruit count to avoid including fruitlets lost during the June drop. FASS calculates fruit drop by counting the pieces of fruit remaining on sample trees at specified dates during the production season, subtracting the remaining pieces from the base line fruit c ount, and then dividing pieces lost by the base count. Subtracting the cumulative fruit drop percentage from 100% gives the percentage of fruit remaining from the original count. The average rate of fruit drop during the season has been estimated at 2.6% per month for early and mid season fruit and 3.3% loss per month for late season fruit (Albrigo, 2006). These estimates, however, are based almost exclusively on drop data collected from August through April, ignoring the last few months of the harvest s eason.
117 Some additional drop data is presented in Figure 4 6 to show drop late in the season, expressed as the percent of fruit dropped since the previous sampling during the 1937 1938 season (Harding, Winston, and Fisher, 1940). These drop rate estimates should be discounted due to the lack of modern irrigation systems in sample groves. Drop rate studies for Valencia oranges in Cuba show drop increasing from December through July (Borroto, Garcia, and Rodriguez, 1977; Borroto et al., 1981). Both Cuban s tudies show drop increasing at an increasing rate during the harvest season. The 1977 study estimates a function for fruit drop over time using a third degree polynomial. A function estimating the cumulative fruit drop rate over time should differ from th at of pound solids per box and fruit weight per piece. The fruit drop rate percentage, calculated from the August 1 fruit count, begins at a value of zero. The y the origin. The cumulative drop percentage ( DROP ) then increases over time as fruit falls from the tree (Albigo, 2006), as shown in Equation 4 5. (4 5) More explicitly, cumulative fruit drop over time can be represented by the second order polynomial function (Eq. 4 6) where the q uadratic parameter DROP DROP ) are expected to be positive, and the y DROP ) is set to zero. (4 6) Cumulative drop parameter values will vary with variety, location, and age, but the assumed functional form will remain constant. To calculate the pieces of fruit remaining on the tree, the drop percentage is subtracted from 100% and multiplied by the August 1 fruit count. Growers and processors monitor specific changes in each key biological characteristic as th e production season progresses, because these characteristics determine the end product
118 quality. To be legally marketable, product quality must meet juice quality parameters set by state and federal governmental agencies. For example, the Florida Departm ent of Citrus requires that the acid concentration and the Brix to acid ratio of fruit inputs must be within defined limits before harvesting and processing can proceed. As shown in Chapter 2, Table 2 3 and 2 4 outline basic quality parameters for fruit i nputs as determined by the State of Florida and Table 2 5 shows the basic requirements for juice sold at wholesale and retail as determined by the U.S. Federal Governme nt (FDOC 2008 a ). More stringent quality parameter boundaries are often set by fruit an d juice buyers, and these stricter parameters are usually included in the terms of contracting. Factors affecting the biological characteristics of fruit include tree variety, rootstock, age, soil type, growing region, weather conditions, irrigation and fe rtilization methods, and a multitude of additional climatic and production variables. While many of these factors are beyond the scope of this research, fruit variety and tree age are central to this analysis and therefore are described in further detail. Fruit variety determines biological characteristics of fruit and juice as well as the date of fruit maturity. Primary orange varieties used for processing in Florida are separated into early season, mid season, or late season fruit varieties. Early sea son fruit consists mainly of Hamlin oranges that are typically ready for processing from October through March. Mid season oranges consist primarily of Pineapple oranges that are usually processed from December through March. The late season orange crop in Florida consists almost entirely of Valencia oranges, typically processed from March through July. Navel oranges and mandarins are produced for fresh marketing and are not included in oranges used for processing. Figure 4 7 depicts the estimated dates
119 ( USDA, FASS, 2007 ). Tree age also impacts internal fruit quality and yield, with older trees generally producing higher quality juice. Young trees typically produce juice with lower soluble solids and Brix to acid ratio while older trees produce juice with higher soluble solids and ratio scores (Zekri, 2000). Tree age also affects tree size and productivity, both of which determine fruit yield per tree. Under commercial growing practices, fruit yields increase for the first 25 years after planting (Parvin, 1971). Beyond this age, yields reach a maximum and gradually decline. Biological Modeling The biological cha racteristics of orange growth are used to construct a mathematical model estimating pound solid production per acre ( PSPA ) as a function of time for a grove of fixed bearing acreage. Optimizing pound solid yields then becomes a matter of timing fruit harv est to capture the maximum available pound solids per acre. Pound solid production per acre can be estimated as the product of the expected pound solids produced per box ( PS ) and the expected boxes produced per acre ( B ), both estimated as functions of tim e ( t ). Box yield per acre, al ong with the pound solids per box are the yield component s used to determine total pound solid s production per acre. Defined as the quantity of 90 pound units of fruit harvested per acre, box yield per acre is also estimated as a function of time. Similar to pound solid changes over time, box yield is constantly changing over time and is expected to increase as fruit increases in size and maturity, until reaching a maximum yield per acre. As fruit continues to hang on the t ree past the date of optimal yield, boxes per acre decline due to fruit drop and fruit drying, both of which negatively affect yield (Jackson, 1991; Albrigo, 2006). This section details how pound solids per box and boxes per acre are calculated and combin ed to estimate the total pound solids per acre available for harvest at any time during the season.
120 The quantity of boxes available for harvest at any time can be calculated using four primary factors; the pieces of fruit on the tree at the beginning of th e production season ( PPT ), the droppage rate of fruit during the season, the average weight of each piece of fruit, and tree planting density ( DEN ). These four factors vary across tree variety, geographic growing region, tree age, and time of the season, each affecting box yield during the season. For this analysis, the pieces of fruit per tree represent the August 1 piece count estimates for each variety, region, and tree age group. The cumulative drop percentage is a function over time estimating the c umulative percentage of fruit dropped since the August 1 fruit count by variety and region. Fruit weight per piece is represented by a function estimating the average weight of a single piece of fruit over time, specific to variety and region. Tree plant ing density estimates the number of trees being grown per acre by variety, region, and age group. Equation 4 7 combines each of these variables to estimate the boxes available for harvest per acre of grove, specific to tree variety (v) growing region (i) age group (a) and harvest time (t) (4 7) Multiplying this function for boxes per acre by the pound solids per box yields the in grove fruit maturity model representing pound solids produced per acre (Eq. 4 8). (4 8) Pound solids per acre production is expressed as a function over time, estimating the average pound solid yield per acre of grove specific to fruit variety geographic region, and tree age Fruit Maturity Data All fruit maturity data used f or pound solid production estimates are compiled from Florida Agricultural Statistics Service (FASS) sample grove test results of on tree fruit quality and quantity. The FASS data set used contains over 85,000 observations from sample groves across
121 Florid a. Data were collected monthly over eleven seasons from 1996 97 to 2006 07 and include sample measurements of total soluble solids (degree Brix), percent acid, fruit weight, juice weight, and fruit drop, each sampled throughout the season from groves segr egated by fruit variety, geographic location, and tree age (Table 4 1). Data on maturity and drop for early and mid season fruit varieties were collected from August 1 through January 1 of each year. Late season fruit maturity data are collected from Sep tember 1 through May 1 and for late season fruit drop from August 1 through April 1. Each recorded observation consists of a 15 piece fruit sample tested for total soluble solids measured as degrees Brix, percent acid measured as the percentage of all t itratable acids, sample fruit weight in pounds, and sample juice weight in pounds. From these four measured values, additional maturity variables can be calculated, including juice weight per box, pound solids per box, and Brix to acid ratio. Juice weigh t per box is defined as the pounds of juice per 90 pound box and is calculated as the ratio of sampled juice weight to sampled fruit weight times 90 pounds. Total soluble solids and juice weight per box are multiplied to calculate pound solids per box. P ound solids per box equals the total soluble solids times the pounds of unfinished juice per 90 pound box [PS = TSS (degree Brix) x Juice weight (lbs) / 90lb box]. Brix to acid ratio is calculated as the total soluble solids divided by the acid percentage Maturity data are specified by fruit variety, geographic growing region, and tree age. Sweet orange varieties are separated into three categories: early, mid, and late season oranges. Early season fruit consists mainly of Hamlin oranges but may also i nclude Parson Brown, Navel, and Ambersweet oranges. Mid season fruit consists almost entirely of Pineapple and Midsweet oranges and late fruit is entirely Valencia oranges. The five geographic growing regions within the state include the Indian River reg ion, the northern region, the central region, the western
122 region, and the southern region ( Figure 4 8). While these geographic regions are not perfectly separ able by county boundaries, FASS defined Florida commercial citrus production areas can be closely approximated by counties. The Indian River region consists mainly of Brevard, Indian River, Martin, Palm Beach, St. Lucie, and Volusia counties; the northern region includes Citrus, Hernando, Lake, Marion, Orange, Pasco, Putnam, and Seminole counties; th e central region includes Highlands, Osceola, and Polk counties; the western region includes DeSoto, Hardee, Hillsborough, Manatee, and Sarasota counties; the southern region includes Charlotte, Collier, Glades, Hendry, Lee, and Okeechobee counties. Beari ng trees are separated into five age groups: Group 1 includes 3 to 5 years old trees; Group 2 includes 6 to 8 years old trees; Group 3 is 9 to 13 years old trees; Group 4 is 14 to 23 years old trees; and Group 5 includes all trees 24 years and older. All d ata sampled across time is segregated into weeks, with August 1 designa ted as the beginning of Week 1 This transformation is consistent with both the biological characteristics of fruit development and industry operations. August is considered the offse ason for processing as Florida oranges do not yet satisfy minimum state maturity standards. Weekly intervals are consistent with typical fruit scheduling to the processing facility and the reported processing data (FCPA, 2007). Regression analysis is use d to estimate time dependent functions for fruit weight, fruit drop, pound solids produced per box, Brix to acid ratio, and total soluble solids. Parameter coefficients are estimated for each variable using ordinary least squares estimation within the SAS software program. Estimates assume the functional form expectations previously discussed and are specific to geographic production region and fruit variety but are aggregated across all tree age groups. This simplification imposes equivalent values acro ss age groups for this set of
123 variables. For example, 6 to 8 year old trees may reach their optimal pound solids production at a different date than 24 year old trees of the same variety and region; however, functions are estimated across all bearing age groups. While this modification may stray from actual production yields for some tree groups, limited data availability forces these simplifying assumptions. The complete set of parameter estimates and the SAS programming codes used to generate these par ameters are presented in Appendix A (Table A 1 to A 7) and Appendix B, respectively. Point estimates for average piece count, average tree planting density, and bearing acreage planted are calculated specific to region, variety, and tree age categories. T ree planting density, reported as the average trees per acre, and bearing acreage data were collected from publicly available FASS datasets and specified by region, variety, and age ( USDA, FASS, 2005 ). These estimates are consistent with the expected prod uction yields. While the piece counts and average tree densities are fixed for any given season, fruit weights and drop rates are estimated as functions over time, similar to the estimates of pound solid per box. The functional form chosen to approximate these changes is again assumed to be a second order polynomial. Combining point and functional estimates enables the calculation of in grove maturity equations specific for each combination of variety, geographic region, and tree age, using the set of p arameter estimates. In grove maturity equations include the 75 time dependent functions estimating pound solids per box (Table A 3) and the 75 time dependent functions estimating box yield per acre (Table A 7). The mathematical product of the estimated c oefficients for pound solids per box and boxes per acre results in a sixth degree polynomial expression estimating average pound solids produced per acre as a function of harvest date (Table 4 2). Each equation represents the pound solids that an acre of variety, region and age specific grove is estimated to
124 produce at any given week during the harvest season. Table 4 3 shows the estimated optimal week of harvest for each tree variety, region, and tree age to maximize pound solid production per acre. Tab le 4 4 lists the designated W eek numbers with the corresponding calendar week. An example of the estimated in grove maturity function for pound solid production per acre is shown in Figure 4 9, detailing changes in late season fruit in the central region for all age groups. The function is consistent with the assumption that pound solid production per acre increases early in the season due to increases in pound solids per box and increasing fruit weight, reaches a maximum later in the season, and then dec reases due to decreasing fruit weight and increased drop. All region, variety, and age specific graphs for pound solid per acre production are similar in form and can be constructed from Table 4 2. Figure 4 10 expands this example to include all three tr ee varieties for central region trees 14 to 23 years old to show the variation in pound solid production per acre over time. The FASS defined harvest windows for each variety are shown below the estimated functions for pound solid production. Some limitat ions exist for the estimated in grove maturity functions. Most notable is the limited availability of biological data late in the harvest season. FASS sampling is only conducted during the traditional harvest and processing season. Data is, therefore, n ot available for all of the estimated time periods. For example, maturity and drop estimates after January 1 for early and mid season fruit and after May 1 for late season fruit are forecasts based on available data and functional form assumptions. While these data on yield and drop are limited in observations, they are the best in grove yield data publicly available. Summary The quality and quantity of available juice is constantly changing as fruit hangs in the tree. The timing of harvest is the decis ion variable that stops this ripening process and sets the potential characteristics of processed juice. The 75 in grove maturity functions for the five
125 growing regions of Florida estimate the pound solids per acre that are expected to be harvested during any given week of the season. Recognizing this peak production yield and the consequences of harvest outside of the optimal should allow decision makers to be better informed when scheduling the harvest. The following chapters use these maturity functio ns to estimate the pound solids captured by the orange juice processing industry and the resulting revenue.
126 Figure 4 1. Total soluble solid percentage concentration for Valencia oranges, sampled from 1935 1938 (Harding, Winston, and Fisher, 1940 ). Figure 4 2. Acid percentage concentration for Valencia oranges, sampled from 1935 1938 (Harding, Winston, and Fisher, 1940 ).
127 Figure 4 3 J uice weight per box for Valencia oranges, sampled from 1935 1938 (Harding Winston, and Fisher 1940 ). Figu re 4 4 Brix to acid ratio for Valencia oranges, sampled from 1935 1938 (Harding Winston, and Fisher 1940 ).
128 Figure 4 5. Pound solid s change for all Valencia oranges, averages from 1935 1938 ( Harding Winston, and Fisher, 1940 ). Figure 4 6 Per cent f ruit d roppage of tagged fruit in 1937 1938, assumed to be monthly drop rates (Harding Winston, and Fisher 1940).
129 Figure 4 7. Florida j uice o range h arvesting s eason s by month, from August through July ( USDA, FASS 2007 )
130 Figure 4 8. Florida commercial citrus production a reas as defined by the Florida Agricultural Statistics Service ( USDA, FASS 2009 a )
131 Figure 4 9. Pound solids per a cre during the harvest s eason, late fruit by age group in the central region, weeks from Aug 1 ( USDA, FASS, 2009a ) Figure 4 10. Pound solids per acre during the harvest season, early, mid, and late v arieties, 14 to 23 year old trees in central re gion, weeks from Aug 1 ( USDA, FASS 2009 a )
132 Table 4 1. Mean and standard variation for e ach biological characteristic by fruit variety and region, including the number of samples (N). ACID (%) TSS (Brix) RATIO (Brix/%) JUCPER (%) PS (lbs/box) FRUWT (lbs) JUCWT (lbs) A1FASSALL (N=21044) 1.287 10.755 9.758 56.441 6.108 5.552 3.151 (0.498) (1.534) (4.089) (5.718) (1.257) (1.144) (0.776) V10I1 (N=519) 1.031 10.697 11.581 53.273 5.724 4.864 2.601 (0.351) (1.195) (4.004) (5.835) (1.038) (0.912) (0.596) V10I2 (N=248) 1.115 10.411 10.501 54.488 5.690 5.131 2.806 (0.39 8) (1.186) (3.704) (5.779) (0.983) (0.935) (0.629) V10I3 (N=1302) 0.994 10.399 11.831 53.780 5.607 4.743 2.563 (0.363) (1.092) (4.245) (5.948) (0.938) (0.909) (0.598) V10I4 (N=2291) 1.033 10.472 11.391 54.149 5.689 4.882 2.650 (0.363) ( 1.180) (3.999) (5.899) (0.995) (0.951) (0.604) V10I5 (N=1585) 0.955 10.160 11.864 53.758 5.477 5.096 2.747 (0.324) (1.102) (4.080) (5.954) (0.943) (0.923) (0.603) V20I1 (N=630) 1.118 10.704 10.715 55.565 5.986 5.343 2.979 (0.373) (1.506 ) (3.875) (5.945) (1.236) (1.050) (0.690) V20I2 (N=21) 1.268 11.140 9.689 54.776 6.131 5.299 2.913 (0.399) (1.537) (3.434) (4.434) (1.139) (0.972) (0.627) V20I3 (N=530) 1.166 10.570 10.162 54.901 5.824 5.199 2.866 (0.383) (1.303) (3.756 ) (5.842) (1.048) (1.123) (0.727) V20I4 (N=1157) 1.158 10.588 10.277 55.876 5.946 5.118 2.870 (0.391) (1.356) (3.799) (5.906) (1.135) (1.042) (0.683) V20I5 (N=695) 1.022 10.254 11.377 54.881 5.653 5.381 2.967 (0.366) (1.245) (4.310) (5. 498) (1.030) (0.940) (0.647) V30I1 (N=2100) 1.491 11.331 8.816 58.800 6.714 5.829 3.435 (0.522) (1.806) (3.902) (5.399) (1.463) (1.082) (0.725) V30I2 (N=380) 1.537 11.162 8.389 58.470 6.566 5.526 3.241 (0.526) (1.686) (3.640) (4.919) (1 .318) (0.945) (0.647) V30I3 (N=3142) 1.488 10.952 8.462 57.573 6.340 6.025 3.481 (0.508) (1.612) (3.588) (4.752) (1.234) (1.083) (0.730) V30I4 (N=2786) 1.495 11.009 8.486 58.168 6.442 5.741 3.349 (0.513) (1.649) (3.638) (5.157) (1.302) (1.072) (0.716) V30I5 (N=3658) 1.408 10.797 8.844 57.574 6.254 6.135 3.540 (0.481) (1.680) (3.839) (4.959) (1.286) (1.102) (0.725) VarReg represents variety and region, including early fruit (V10), mid season fruit (V20), late fruit (V30), India n River (I1), north (I2), central (I3), west (I4), south(I5). N is the number of observations
133 Table 4 2. Estimated parameter results for p ound solids production per acre. VarRegAge I nt t t^2 t^3 t^4 t^5 t^6 V10I1G1 336.9 26.17 0.073 0.0245 1.23E 04 5.02E 06 4.06E 08 V10I1G2 623.5 48.44 0.136 0.0454 2.28E 04 9.29E 06 7.52E 08 V10I1G3 819.7 63.68 0.178 0.0597 3.00E 04 1.22E 05 9.88E 08 V10I1G4 1245.7 96.78 0.271 0.0907 4.57E 04 1.86E 05 1.50E 07 V10I1G5 968.9 75.28 0.211 0.0706 3.55E 04 1.44E 05 1.17E 07 V10I2G1 166.3 16.19 0.001 0.0128 9.67E 05 1.27E 06 1.16E 08 V10I2G2 913.1 88.87 0.008 0.0705 5.31E 04 6.99E 06 6.34E 08 V10I2G3 1128.9 109.87 0.010 0.0872 6.57E 04 8.64E 06 7.84E 08 V10I2G4 1551.5 151.00 0.014 0.1198 9.02E 04 1.19E 05 1.08E 07 V10I2G5 1160.6 112.95 0.010 0.0896 6.75E 04 8.88E 06 8.06E 08 V10I3G1 496.6 49.05 0.083 0.0414 3.17E 04 3.62E 06 3.31E 08 V10I3G2 755.5 74.62 0.127 0.0630 4.83E 04 5.51E 06 5.03E 08 V10I3G3 1154.5 114.04 0.193 0.0963 7.38E 04 8.42E 06 7.69E 08 V10I3G4 1686.2 166.56 0.282 0.1407 1.08E 03 1.23E 05 1.12E 07 V10I3G5 1820.8 179.86 0.305 0.1519 1.16E 03 1.33E 05 1.21E 07 V10I4G1 442.5 30.23 0.268 0.0098 9.12E 05 2.72E 07 2.81E 09 V10I4G2 843.5 57.62 0.510 0.0187 1.74E 04 5.18E 07 5.35E 09 V10I4G3 1510.8 103.21 0.914 0.0335 3.11E 04 9.28E 07 9.59E 09 V10I4G4 2022.6 138.17 1.223 0.0448 4.17E 04 1.24E 06 1.28E 08 V10I4G5 1956.0 133.62 1.183 0.0433 4.03E 04 1.20E 06 1.24E 08 V10I5G1 431.8 32.52 0.183 0.0256 2.94E 04 1.31E 06 1.94E 08 V10I5G2 978.8 73.71 0.415 0.0580 6.67E 04 2.96E 06 4.40E 08 V10I5G3 1487.2 112.00 0.631 0.0882 1.01E 03 4.50E 06 6.68E 08 V10I5G4 1954.7 147.21 0.830 0.1159 1.33E 03 5.92E 06 8.79E 08 V10I5G5 1804.6 135.91 0.766 0 .1070 1.23E 03 5.46E 06 8.11E 08 V20I1G1 241.2 21.94 0.065 0.0189 7.41E 05 2.89E 06 1.91E 08 V20I1G2 448.4 40.78 0.121 0.0351 1.38E 04 5.37E 06 3.55E 08 V20I1G3 552.5 50.25 0.149 0.0432 1.70E 04 6.61E 06 4.37E 08 V20I1G4 801.4 72.89 0.217 0.062 7 2.46E 04 9.59E 06 6.34E 08 V20I1G5 978.9 89.03 0.265 0.0765 3.01E 04 1.17E 05 7.75E 08 V20I2G1 198.8 27.35 0.322 0.0429 3.06E 04 6.02E 06 5.58E 08 V20I2G2 193.0 26.55 0.313 0.0417 2.97E 04 5.85E 06 5.41E 08 V20I2G3 569.0 78.26 0.923 0.1228 8. 75E 04 1.72E 05 1.60E 07 V20I2G4 635.0 87.33 1.030 0.1370 9.77E 04 1.92E 05 1.78E 07 V20I2G5 849.3 116.81 1.377 0.1833 1.31E 03 2.57E 05 2.38E 07 V20I3G1 320.5 39.01 0.172 0.0469 5.01E 04 3.08E 06 4.26E 08 V20I3G2 720.2 87.67 0.387 0.1055 1.13E 03 6.92E 06 9.57E 08 V20I3G3 1004.5 122.27 0.540 0.1471 1.57E 03 9.66E 06 1.34E 07 V20I3G4 1476.2 179.70 0.794 0.2162 2.31E 03 1.42E 05 1.96E 07 V20I3G5 1511.3 183.97 0.813 0.2214 2.36E 03 1.45E 05 2.01E 07 V20I4G1 241.6 21.90 0.125 0.0087 5. 30E 05 5.80E 07 3.20E 09 V20I4G2 481.4 43.64 0.249 0.0173 1.06E 04 1.16E 06 6.38E 09 V20I4G3 1030.6 93.43 0.532 0.0370 2.26E 04 2.47E 06 1.37E 08 V20I4G4 1580.4 143.27 0.816 0.0567 3.47E 04 3.80E 06 2.09E 08 V20I4G5 1793.8 162.62 0.926 0.06 44 3.94E 04 4.31E 06 2.38E 08 V20I5G1 292.3 25.06 0.088 0.0145 1.59E 04 1.22E 07 4.45E 09 V20I5G2 675.9 57.95 0.204 0.0336 3.67E 04 2.83E 07 1.03E 08 V20I5G3 1295.0 111.03 0.391 0.0644 7.03E 04 5.42E 07 1.97E 08 V20I5G4 1518.5 130.18 0.459 0.0756 8.24E 04 6.35E 07 2.31E 08 V20I5G5 1682.7 144.26 0.509 0.0837 9.13E 04 7.04E 07 2.56E 08 V30I1G1 152.5 18.14 0.147 0.0123 9.08E 05 2.76E 07 3.01E 09
134 Table 4 2. Continued VarRegAge int t t^2 t^3 t^4 t^5 t^6 V30I1G2 275.8 32.80 0.266 0.0222 1.64E 04 5.00E 07 5.44E 09 V30I1G3 503.4 59.87 0.485 0.0405 3.00E 04 9.12E 07 9.93E 09 V30I1G4 565.8 67.29 0.546 0.0455 3.37E 04 1.02E 06 1.12E 08 V30I1G5 537.3 63.90 0.518 0.0433 3.20E 04 9.73E 07 1.06E 08 V30I2G1 395.5 40.6 3 0.231 0.0209 1.27E 04 6.40E 07 4.93E 09 V30I2G2 453.7 46.61 0.265 0.0239 1.45E 04 7.34E 07 5.65E 09 V30I2G3 679.4 69.79 0.396 0.0358 2.17E 04 1.10E 06 8.46E 09 V30I2G4 881.7 90.58 0.514 0.0465 2.82E 04 1.43E 06 1.10E 08 V30I2G5 794.3 81.60 0. 463 0.0419 2.54E 04 1.28E 06 9.90E 09 V30I3G1 203.0 26.57 0.417 0.0228 1.67E 04 3.46E 07 4.39E 09 V30I3G2 407.1 53.28 0.835 0.0457 3.35E 04 6.94E 07 8.81E 09 V30I3G3 540.5 70.74 1.109 0.0607 4.44E 04 9.22E 07 1.17E 08 V30I3G4 735.6 96.29 1.510 0.0826 6.05E 04 1.25E 06 1.59E 08 V30I3G5 959.4 125.58 1.969 0.1078 7.89E 04 1.64E 06 2.08E 08 V30I4G1 247.9 31.89 0.450 0.0267 1.58E 04 1.28E 06 9.82E 09 V30I4G2 514.1 66.13 0.933 0.0554 3.27E 04 2.65E 06 2.04E 08 V30I4G3 609.4 78.38 1.105 0. 0656 3.88E 04 3.14E 06 2.41E 08 V30I4G4 755.6 97.18 1.371 0.0813 4.81E 04 3.90E 06 2.99E 08 V30I4G5 870.0 111.89 1.578 0.0937 5.54E 04 4.49E 06 3.45E 08 V30I5G1 180.5 24.32 0.193 0.0161 1.25E 04 2.30E 07 3.25E 09 V30I5G2 405.7 54.65 0.435 0.036 1 2.80E 04 5.17E 07 7.30E 09 V30I5G3 564.5 76.03 0.605 0.0503 3.90E 04 7.20E 07 1.02E 08 V30I5G4 690.8 93.04 0.740 0.0615 4.77E 04 8.81E 07 1.24E 08 V30I5G5 830.9 111.91 0.890 0.0740 5.74E 04 1.06E 06 1.49E 08 ion, and age, including early fruit (V10), mid season fruit (V20), late fruit (V30), Indian River (I1), north (I2), central (I3), west (I4), south(I5), age 3 5 (G1), age 6 8 (G2), age 9 13 (G3), age 14 23 (G4), and age 24+ (G5). Table 4 3. Optimal w eek of h arvest to m axi mize pound solid production, weeks from Aug. 1 Variety Indian River North Central West South Early 21 25 25 29 22 Mid 24 21 23 29 28 Late 35 38 36 34 36 Table 4 4. Designated w eek numbers with corresponding calendar date. Week Dat e 1 Aug 1st 5 Sep 1st 9 Oct 1st 14 Nov 1st 18 Dec 1st 23 Jan 1st 27 Feb 1st 31 Mar 1st 35 Apr 1st 40 May 1st 44 Jun 1st 48 Jul 1st 53 Aug 1st
135 CHAPTER 5 THEORETICAL MODELING : SCHEDULING OF THE OR ANGE HARVEST TO OPTIMIZE RETURNS This chapter uses the fundamentals of economic theory to describe how harvest timing affects fruit production and profits of the Florida orange juice industry. Intensity and duration of harvest and processing are defined, including a description of their effects on in dustry operations and returns. The returns to growers, processors, and the total industry are then outlined by defining a separate profit function for each, including an explanation of how the adoption of mechanical harvesters is expected to change profit s through changes in harvest timing and product yield. The chapter also uses comparative statics to describe how changes in pound solids and box volumes change grower, processor, and total industry returns. Harvest timing, defined as the specific date o f picking, is a key decision variable determining crop quantity and quality. Harvest timing is central to the optimization problem because it influences the quality of fruit and juice captured, the quantities delivered for processing, and the total return s from growing and processing. The harvest timing of an individual grove determines juice characteristics, pound solid yield per box, and fruit yield per acre (Searcy, Roka, and Spreen, 2008). For example, harvesting on the date of peak pound solid produ ction will typically yield growers the greatest returns per acre. Conversely, harvesting ahead of the optimal harvest date yields smaller fruit containing less juice and sugar, while harvesting after the optimal harvest date yields fewer pieces of fruit. Either suboptimal harvest scenario results in decreased pound solids and decreased financial returns. It is the collective harvest timing for all individual groves that directly determines the processing schedule at the receiving juice plant. This coll ective harvest schedule also dictates the duration and intensity of juice processing for the season. Growers and processors jointly
136 schedule harvest and processing, with each attempting to adjust the harvest schedule to optimize returns. Optimizing indi vidual returns requires adjusting the intensity and duration of the annual harvest schedule to account for the physical and biological constraints of fruit production, harvesting, and processing. Harvest capacity is a key constraint influencing both the i ntensity and duration of the season. Adding mechanical harvesting has the potential to alter current harvest capacity, thereby changing harvest timing, product yield, and returns to industry players. These changes could influe nce both the individual and collective timing of harvest, as well as lowering the unit harvest costs, a necessary condition for mechanical harvester adoption. The remainder of this chapter examines how the timing of harvest determines industry returns, including changes to the harve st schedule and returns caused by the adoption of mechanical harvesting. Intensity and Duration Intensity and duration of the citrus harvest season are central to optimizing the harvest schedule and maximizing industry returns. Intensity and duration dire ctly affect the volume of juice processed, and the costs to harvest, extract, and store juice. For this analysis, intensity ( I ) is defined as the number of boxes harvested each period (boxes/week) and duration ( D ) is defined as the length of the overall h arvest and processing season (weeks/year). The product of intensity and duration is the seasonal box yield ( TB ), shown in Equation 5 1. (5 1) If the maximum box yield is assumed fixed for a given year, as determined by the numb er of bearing trees and environmental conditions of that season, then harvest of the entire crop requires that intensity and duration vary inversely and proportionally. Increasing the duration of the
137 season, forces a proportional decrease in the average w eekly intensity of processing. Conversely, increasing weekly harvesting intensities forces a proportionate increase in processing intensity and the volume of juice going into storage. These concepts are paramount to understanding inherent tradeoffs betwe en weekly processing volumes and weeks of processing operation. The intensity and duration of harvesting and processing are directly impacted by the daily capacity of processing equipment and the capacity to store juice. Overriding both harvesting and pr ocessing capacity is the biological reality that fruit quality and quantity is constantly changing on the tree and begins to deteriorate immediately after being pulled from the tree. Intensity and duration are realized through the scheduling of weekly loa d allocations. The summation of all weekly load allocations determines the annual harvest and processing schedule, processed, and stored each week. The mea ns of control over duration and intensity lie in the deliveries. Load allocations must be available to the grower before harvest occurs to ensure picked fruit ca n be processed before spoiling. Attempting to optimize processing operations, the processor determines a target quantity of fruit to move through the plant each period. The target quantity is dependent on plant capacities, annual crop characteristics, an d the ability to secure fruit inputs. The processor divides available loads among growers based on the percentage of on tree fruit the grower has remaining, allotting the portion of the needed fruit each is to deliver. The extent to which the timing of f realized by the industry are maximized.
138 Cost I mpacts Costs of harvest, extraction, and storage are dependent on the intensity and duration of the harvest and processing season. The unit cost to harvest changes as the duration and intensity of the harvest change, depending on the harvesting system being used and the market conditions of the harvesting i ndustry. When using hand labor, variable costs have the most impact on total harvest costs. Piece rate wages tend to fluctuate according to labor market conditions. A more intense harvest over a shorter duration increases the demand for labor, causing a n increase in harvest costs. A less intense harvest over a longer period tends to decrease unit harvest cost. With mechanical harvesters, the fixed costs of investment in the mechanical harvesting equipment drives unit harvest costs. A more intense harv est does not necessarily alter the variable costs of harvester operation, but it does require additional mechanical harvesters, increasing fixed costs. A less intense harvest over a longer duration would reduce the number of mechanical harvesters needed, lowering fixed costs. The unit cost of extraction changes with increases in the intensity of extraction. An orange processing facility is expected to have a processing cost function much like any other factory with fixed factors of production ( Figure 5 1). Increasing returns to scale are expected until production nears plant capacity, where decreasing returns to scale are expected, forming a U shaped marginal cost curve. Under these production conditions, the direction of change in extraction cost due to increasing intensity will depend on specific plant production and extraction capacity levels. The large upfront capital costs for storage are similar to the capital requirement for extraction. With storage, however, increasing returns and a decreasing average cost curve are expected until storage capacity is full ( Figure 5 2). The next unit stored then requires a large capital investment, so the marginal and average cost curves spike. For example, the marginal
139 cost of adding juice to a one million gal lon storage tank is close to zero for each additional gallon, until the tank is full. Attempting to store one gallon above capacity, however, requires a large investment in new bulk storage. The marginal cost for this additional gallon is substantial. S imilar to extraction costs, the marginal change in storage costs due to increasing intensity depends on specific plant storage capacity and production levels. Total P ound Solids C aptured Intensity and duration of the season determine the timing of harvest and, therefore, the percentage of available pound solids captured within a given season. The pound solid volume captured and the timing of harvest and processing are key factors in determining grower and processor profits. Figures 5 3 and 5 4 describe g rower and processor returns as a function of pound solid production. All factors of fruit and juice production are assumed fixed for the season, with only pound solid volume and fruit yield allowed to fluctuate due to harvest timing. Changes in pound sol ids and fruit yield are expected to follow seasonal changes described in Chapter 4 (Tab. 4 2, Figure 4 10). 1 2 ) increase as the pound solids captured increase and reach a maximum if al l available pound solids are captured (TPSg ). Attaining a yield of TPSg would require harvesting all fruit during the few weeks of peak pound solid yield. This represents capturing the theoretical maximum pound solids for a season. Returns using mecha nical harvesting are assumed to be greater per pound solid because of harvest cost savings. Both functions are increasing because if capture of the marginal pound solid gives a negative return, then the grower will cease the harvest. For example, Figure 4 10 and Tables 4 3 and 4 4 show that a grove of mid season fruit in the central region should be harvested within a few weeks of January 1 (Week 23). Harvesting outside of this optimal window will decrease the pound solids, therefore decreasing grower pr ofits.
140 Pro and then decrease as higher volumes of fruit and juice cause processing and storage costs to escalate. A solution at TPSp would give the highes t possible returns by making efficient use of plant capital to process the optimal volume of fruit for the season. Processing and storing more than the optimal level of pound solids pushes processing operations beyond the most efficient input volumes. Excess throughputs lower efficiencies and returns by increasin g stoppages due to bottlenecks, requiring overtime pay, and pushing the limits of storage capacity. These additional variable costs can outweigh the marginal gains in revenue, causing processor returns to decrease. Pound solid volumes less than TPSp do not allow for spreading of fixed costs across as many units as possible. Potential revenue is lost as labor and capital sit idle or operate at less than full capacity. Processors manage the flow of fruit throughout the season to capture TPSp pound solid s in an attempt to maximize annual returns. Total industry returns as a function of total pound solids (TPS) can be mathematically expressed as the sum of grower returns and processor re turns, as shown in Equation 5 2 (5 2) The total pound solid volume producing optimal returns to the industry is determined by differentiating industry profits with respect to total pound solid volume and setting equal to zero, as shown in Equation 5 3. (5 3) Total industry returns are shown graphically in Figure 5 5 when using either hand harvesting 1 2 ). Optimizing industry returns under hand harvesting occurs when producing pound solid volume of TPS i 1 Returns are optimized at TPS i 2 when using mechanical harvesting. Neither indu stry optimal maximizes individual returns, but the summed
141 returns to both players are maximized. Pound solid production volumes between TPSp and TPSi 1 suggest processor market power, while production between TPSi 1 and TPS g suggest grower market power assuming operation with hand harvesting With mechanical harvesting, production between TPSp and TPSi 2 suggests power is held by processors, while production between TPSi 2 and TPSg suggests growers hold market power. In addition to describing indus try returns, Figure 5 5 provides insight into the incentives Because growers are responsible fo r all costs of harvest, decreases in the unit cost of harvest afforded b y mechanical harvesters will cause Growers, therefore, should favor the use of mechanical harvesters unless mechanical harvesters could decrease the volume of pound solids captured enough to decrease grower returns. The case for processor acceptance of mechanical harvesters is somewhat more complicated. Harvest cost savings from mechanical harvesting do not directly impact processors because they do not typically pay for the harvest. The processor profit function is then the same before a nd after harvester adoption. The addition of mechanical harvesters does, however, shift the new industry optimal pound solid produc tion volume from TPSi 1 to TPSi 2 If the timing of harvest is adjusted to capture TPSi 2 then total industry returns and grower returns will increase, but processor returns will decrease. This decrease in processor returns is due to diseconomies of scale caused by increasing volumes. C ost increases arise from processing higher than optimal volumes of fruit and attempting t o store larger volumes of juice than processing facilities are designed to handle These extra costs can be manifested through additional processing operations such as less than optimal feed mill dryer operation, paying employees overtime
142 wages, and high cost offsite juice storage. Processors are not expected to favor mechanical harvesters if adoption pushes pound solid production away from TPSp If mechanical harvesting and a change in the harvest schedule cause lower processor returns, processors coul d be expected to agree to accept mechanical harvesting if they are compensated for their financial losses. By the Kaldor Hicks compensation criteria, TPSi 2 is preferred to TPSi 1 because it is theoretically possible for growers to fully compensate proce ssors, making one party better off and keeping the other party as well off. In this example, the industry as a whole can be made better off, while transfer payments could help to ensure no party is made worse off due to the adoption of mechanical harveste rs. Total Industry Returns As a starting point for a concept ual model, consider the Florida orange juice production IND ) can then be G P ), shown in Equation 5 4. (5 4) Grower R eturns Grower returns can be defined as total revenue from the sale of fruit (TRF) less total fruit production costs (TPC) and total harvest costs (THC), which include transporting fruit to a processing plant( Eq 5 5 ) (5 5) Growers are paid for the total quantity of pound solids they deliver to the processing plant. Total fruit revenue ( TRF ) is equal to the product of the delivered in price per pound solid ( DIP ), the pound solids per box ( PS ), the box yield per acr e ( B ), and the total acres ( A ) of bearing trees owned by the grower (Eq. 5 6).
143 (5 6) Note that the product of pound solids per box, box yield per acre, and total acres is equal to the total pound solid production for the sea son ( TPS ), and that the product of the box yield per acre and the total acres is equal to the total boxes produced during the season ( TB ). The two main cost components determining annual grower returns are fruit production and harvest costs. Production co sts include capital, fertilizer, pest control chemicals, labor, and other input costs associated with grove management and caretaking. Production costs per acre are the harvest until the current harvest, with production ending for the season at the time of harvest. The total cost of fruit production is calculated as the per acre unit cost of production ( UPC ) times the acreage of grove holdings (Eq. 5 7). (5 7) The total cost of production often accounts for a large percentage of grower costs but is considered a sunk cost at the time of harvest and does not factor into the harvest decision. Total harvest costs are paid by growers and incl ude pick, roadside, and hauling costs to deliver fruit to a processing plant. Picking and roadsiding are separate costs associated with hand harvesting crews. The pick charge is paid directly to a worker to remove fruit from a tree and place into collect ion tubs. Roadside costs are incurred to move fruit from the collection tubs to bulk trailers at the edge of a grove. When using mechanical harvesters, pick and roadside costs are combined into a single cost. Hauling costs account for the cost of removi ng loaded fruit trailers from the grove and delivering them to the processing facility. Hauling costs increase as the travel distance increases but should not vary with the method of harvest.
144 The total cost to pick, roadside, and haul is equal to the per box unit harvest costs times the quantity of boxes harvested. Incorporating separate functions for hand harvesting and mechanical harvesting can be done by including the number of boxes picked with each technology and the respective unit harvest cost. Eq uation 5 8 expresses total harvest costs as the sum of total hand harvesting costs ( THC HH ) and total mechanical harvesting costs ( THC MH ), which is equivalent to summing the unit cost of hand harvest ( U HC HH ) times the total boxes hand harvested ( TB HH ) and t he unit cost of mechanical harvest ( U HC M H ) times the total boxes mechanically harvested ( TB M H ). (5 8) Unit harvest costs are usually dependent on the expected boxes per acre in the grove to be harvested for both hand and mechanica l harvesting systems. The harvest cost per box typically decreases as the number of boxes available per acre increases. This cost difference is manifested rate and demanding higher piece rates for low yielding groves. Lower piece rates are accepted for higher yielding groves because fruit can be picked at a much faster rate, enabling harvesters to pick more fruit while exerting the same effort. Similarly, mecha nical harvesters will accept lower per box picking rates for groves producing higher box yields per hour of operation. The lower rates are accepted because mechanical harvesting effort depends more on the number of trees to be harvested than the fruit yie ld. The time required to mechanically harvest a tree is independent of the quantity of hanging fruit. Higher yielding trees are therefore less costly per box to shake than low yielding trees. While this dynamic cost structure is practiced by the Florida citrus industry, estimations for this analysis are simplified by assuming constant harvest costs. Unit costs for hand and mechanical harvesting are assumed to be different and are held
145 constant with respect to per acre yields. This simplification is mad e due to a lack of data on yield specific harvest costs. This simplifying assumption does, however, keep harvest costs consistent with average fruit yields, which should produce results in line with average industry costs. Total returns to the grower ar e represented in Equation 5 determined by the delivered in price of fruit, pound solid production, box yield, grove acreage, unit production costs, hand harvest costs, and mechanical harvest costs. (5 9) Total returns to the grower are graphically represented as a function of total pound solid production in Figure 5 3 by assuming fixed values for fruit prices, production costs, and bearing acreage. Pound solids per box and boxes per acre are not fixed bu t are determined by the biological characteristics of fruit at the time of harvest, as shown in Figure 4 11 where the time of harvest directly impacts the volume of pound solids harvested. Unit harvest costs are allowed to increase with increasing product ion volume, causing profits to increase at a slower rate as production reaches the maximum. Variable unit harvest costs are not imperative to this analysis and not included in Equation 5 9, but are likely to be realized in the labor market. Attempting to maximize profits, growers time the harvest of a grove to coincide with peak pound solid production, yielding the greatest production and profits per acre. 1 represents grower returns using hand harvesting ( TB MH = 0) as a function of the quantity of pound solids captured. Grower returns are shown to be negative when few pound solids are captured because of the high fixed cost of fruit production. As add itional pound solids are sold, returns eventually become positive and continue to increase with the sale of each additional unit. Operating at TPS g results in the capture of the maximum possible pound solids and the greatest returns to the grower, repres ented
146 as 1 TPS g represents timing of the harvest such that all groves are picked at peak pound solid production, yielding the maximum total pound solids. At quantities less than TPSg profits are less than optimal. Growers may switch from hand harvest to mechanical harvest ( TB H H = 0) if mechanical harvesters have the potential to lower the cost of harvest. The adoption of mechanical harvesters shifts the function for 1 2 as adoption affords a decrease in the unit cost of harvest The function 2 yields higher returns to the grower at each pound solid value due to the decreased harvest cost The production volume yielding maximum returns 2 is again realized at TPS g Growers attempting to maximize returns will therefore t ry to capture all possible pound solid s, independent of the type of harvesting used. Processor R eturns Processors convert whole fruit into a marketable juice product. Processor returns are defined as the total revenue gained from juice sales ( TR J ) les s total fruit input costs ( TFC ), total extraction costs ( TEC ), and total storage co sts ( TSC ), shown in Equation 5 10 (5 10 ) Note that the total fruit input cost paid by the processor is equivalent to the total revenue received by the grower (Eq.5 6) for the sale of fruit ( TFC = TR F ). To simplify Equation 5 10 the total revenue from juice sales and the total cost of fruit inputs can be combined and expressed as the total value added from juice production ( TVA ) shown in Equatio n 5 11 (5 11 ) Additionally, the total value added from production can be calculated on a per unit basis by multiplying the total pound solids ( TPS ) produced during the processing season times the unit processing margin ( PM ), as shown in Equation 5 12
147 (5 12 ) The processing margin is defined as the additional unit value added to juice when whole fruit inputs are converted to marketable juice or concentrate. This calculated variable represents the value that processing adds to juice. The processing margin is the difference between the price paid for a pound solid entering the plant as whole fruit and the price received for a pound solid leaving the plant as juice. The two remaining costs asso ciated with fruit processing are juice extraction and storage costs. For the purpose of this analysis, extraction costs are defined to include all variable costs associated with fruit handling, unloading, cleaning, grading, extracting juice, juice finishi ng, pasteurizing or evaporating, and processing of byproducts. Under this definition, extraction activities begin when fruit is dropped at the processing facility and are completed once juice enters storage. Total extraction costs are typically expressed as the unit cost to extract a box of fruit inputs times the total boxes processed (Eq. 5 13 ). (5 13 ) The operational objective of extraction is to capture juice meeting specified quality parameters at the lowest possible uni t costs. Unit processing costs mirror costs associated with many other manufacturing facilities that handle a variable quantity of inputs subject to fixed capital constraints. While very little information is publicly available pertaining to unit extract ion costs in modern orange juice processing facilities, economic theory argues that economies of scale will cause unit extraction costs to be non linear. If plant capital is fixed at the start of the processing season, then there should exist an optimal t hroughput of boxes that minimize the average unit extraction cost. Box volumes above or below this optimal will increase unit costs. Further analysis of Figure 5 1 depicts this relationship graphically by
148 showing the expected shapes of the m arginal and a verage unit c ost c urves for processing in a t ypical o range j uice p rocessing p lant. The high fixed costs of a processing plant greatly inflate marginal cost when only a few units are being processed. As the processing volume increases, fixed costs are spr ead across more units of output and marginal cost decreases. At some production value, marginal costs will reach a minimum and then begin to increase as output increases, due to the decreasing economies of scale experienced when increasing production volu mes begin to tax fixed processing equipment. At a still greater production level, marginal cost will equal average cost, and unit processing costs are minimized. Beyond this production volume, unit costs increase as physical plant capacities are strained and the cost to produce additional output increases quickly. Increasing production further requires investment in an additional processing facility which will have similar marginal and average cost functions. The total cost of processing as a function of output, equal to the product of average cost and total output quantity, is expected to start at the level of fixed costs, increase at an increasing rate as inputs are increased, slow to a decreasing rate, and then spike with increases in capital. Storag e costs are defined as all costs involved with holding juice from the completion of extraction until the juice leaves the processing facility. The costs of juice storage, either construction or maintenance costs, directly impact operational decisions made by production managers. Storage capacity is necessary for two reasons. First, a processor must bridge the time gap between the fruit harvest period (typically less than nine months) and a twelve month juice sales cycle Second, storage enables processo rs to mask natural quality variations occurring during a typical harvesting season and blend juices throughout the calendar year, produc ing a consistent end product that meets consumer quality specifications For example, storag e and blending allow proces sors to mix some poor quality, low ratio juice from early season fruit with
149 higher ratio juice captured later in the season, enabling processors to buy fruit inputs of varying juice quality while maintaining a consistent quality end product Total storag e costs can be expressed as the unit cost to store a pound solid of processed juice times the volume of juice in storage (PSS) (Eq. 5 14 ). (5 14 ) Figure 5 2 graphically represents the expected average cost of juice storage. The cost to store the first gallon is extremely high, but as the volume stored increases, the cost to store an additional unit is close to zero, decreasing average storage costs. Average costs should decrease at a decreasing rate until the bulk storage tank is full. Storing an additional unit of juice would then require use of an additional tank, resulting in a significant increase in cost. The average cost to store additional juice again decreases as the volume stored increases. The total cost of ope rating storage annually is expected to be a large fixed cost with small variable cost increases from storing additional units. Processors adding additional storage tanks will incur large capital expenditures, followed by fixed annual operating costs and v ariable unit costs comparable to the variable costs of operating existing storage tanks. Processor returns can be expressed in terms of unit processing margin, unit extraction costs and unit storage costs, as shown in Equation 5 15, using Equation 5 10 and the relatio nships defined in Equations 5 11 through 5 14 (5 15 ) Total returns to the proc essor, as shown in Equation 5 15 are graphically represented in Figure 5 4 as a function of total pound solid production. This function assumes that processing capacity is fixed at the beginning of the season, and managers attempt to minimize the cost of processing the optimal volume of pound solids. Similar to grower returns, processor returns are
150 negative if very few pound solids are pr ocessed during the season due to the high capital costs associated with owning a juice plant. As the volume of pound solids run through the plant increases, processor returns increase. Total returns became positive when the pound solid volume is above th e breakeven point. At this breakeven volume, processing margin revenues are sufficient to cover fixed costs. Continuing to increase pound solid volumes above the breakeven point allows processor to earn positive returns. Returns eventually reach a maxim um, shown in Figure 5 when the pound solid volume is optimal for the processor, TPS p Increasing pound solid volume beyond this optimal level will cause processor returns to decline due to the eventual decreasing marginal returns inherent with fixed factors of production. These decreasing returns are realized when the marginal cost of processing, shown in Figure 5 2, is greater than the processing margin revenue earned on each additional pound solid produced. Total returns will eventually beco me negative again as processing volumes continue to increase. C hanges in harvest costs afforded by the adoption of mechanical harvesters do not directly enter the function estimating processor returns unlike the function for grower returns Any changes in harvest timing caused by the addition of mechanical harvesters could, however, alter pound solid production and processor returns. If mechanical harvesting allows for harvest timing closer to peak pound solids production, the pound solid volumes captu red will increase. Processor returns will increase if processing volumes had been less than optimal, below TPS p and returns will decrease if volumes had been at or above the optimal quantity. Conversely, if mechanical harvesting causes the timing of ha rvest to occur during less optimal periods, pound solid volumes will decrease. In this case, processor returns will decrease if processing volumes had been less than optimal and increase if volumes were at or above the optimal pound solid quantity.
151 Indu stry R eturns Combining grower returns (Eq. 5 9) and processor returns (Eq. 5 15 ) into the original function for total industry returns (Eq.5 4) gives a complete representation of total industry returns (Eq. 5 1 6). (5 16 ) Total retur ns to the industry as a whole are graphically represented as a function of pound solid production in Figure 5 1 2 represent total industry returns as a function of pound solid volume under hand harvest and mechanical harvest systems r espectively. The increased returns from using mechanical harvesting result from the unit harvest cost savings realized by growers, as shown in Figure 5 1 are achieved with a pound solid volume of TPSi 1 Under mechanical harvesting, maximum returns 2 are achieved with a pound solids volume of TPSi 2 Figure 5 5 shows that changing the cost of harvest changes both the optimal industry returns and the optimal pound solid volume required to realize the higher returns. To fully understand the motivation for this change, the functions expressing grower and processor returns ( Figure 5 3 and 5 4) are examined together. Pound solid production volume, which must be equivalent across both players, is shown to be a central driver of industry revenues, affecting the individual returns of each player. The optimal pound solid volume under hand harvesting is found by examining marginal pound solid increases. Both parties gain revenue with each additional unit until rea ching TPSp beyond which processor returns decrease as grower returns continue to increase. The optimal industry pound solid production volume is found by increasing production until the marginal loss in returns to the processor equals the marginal gain in revenue to the grower, TPSi 1 Marginal increases in pound solid production greater than
152 TPSi 1 will result in decreases to total industry returns because the processor loses more than the grower gains. As previously discussed, if mechanical harvester adoption lowers the cost of 1 2 The increased slope of the new grower returns function increases the pound solid volume required to maximize total industry returns. The new optimal pound solid volume is then TPSi 2 with grower returns 2 2 The previous example solves for the optimal industry returns and is directly relevant to a vertically integrated cooperative. If growers and processor s are not integrated but separate business entities, the harvest timing decision must still be made jointly. It is through the bargaining and negotiations of this joint decision making process that growers and processors reconcile differences in harvest d ate, and consequently determine the pound solid volume captured. When one party is attempting to maximize its returns at the expense of the other example, suppose pr oduction is at TPS p in Figures 5 3, 5 4, and 5 5. To move pound solid production to the industry optimal level, harvest and processing intensity must be increased during biologically optimal periods. Moving to TPS i will increases grower returns and dec rease processor returns, as shown in Figures 5 3 and 5 4. This shift in production levels provides an opportunity to make Kaldor Hicks improvements in the market, improving the returns to at least one player while not making any other player worse off. S uch an improvement could involve the winning player, in this case the grower, making a payment to the losing player, the processor. The gains to the winning player are shown to be large enough to retain some additional revenue after offsetting all losses. The industry as a whole would be made better off, and no player would be made worse off.
153 Industry operation away from the industry optimal production level suggests that one group of players has market power. As described, growers would like pound solid production volume as high as possible, TPSg because their individual returns increase when industry production increases. Production volumes above the industry optimal would then suggest that growers are exerting market power over processors. Converse ly, processor returns increase when production volumes are below the industry optimal. Production volumes below the industry optimal suggest that processors are exerting market power over growers. Additionally, pound solid production levels below the pro cessor optimal, TPSp suggest that outside factors such as freeze, hurricane, or disease are reducing pound solid production. Summary The timing of harvest determines the intensity and duration of the processing season, directly impacting the volume of j uice captured and the returns to industry players. Growers and processors negotiate to jointly determine harvest timing, each attempting to maximize individual returns. The addition of mechanical harvesting has the potential to alter these returns and ch ange the volume of pound solid production required to maximize total industry returns. The following chapter expands upon this theoretical model to empirically estimate returns to the orange juice processing industry.
154 Figure 5 1 Expected m arginal and a verage unit c ost curves for processing in a t ypical o range j uice p rocessing p lant as a function of quantity Figure 5 2. Average cost curve for juice or concentrate storage in a typical orange jui ce processing plant as a function of quantity $ Q
155 Figure 5 3. Grower returns as a function of total pound solids Figure 5 4. Processor returns as a function of total pound solids $ 1 2 T PS p TPS i1 TPS i2 TPS g TPS 2 1 g i 1 1 g i 2 2 0 $ TPS i2 TPS g TPS p i 1 p i 2 0 TPS p TPS i1
156 Figure 5 5. Total industry returns as a function of total pound solids $ 1 TPS i2 T PS g TPS 2 1 2 0 TPS p TPS i1
157 CHAPTER 6 EMPIRICAL MODELING A ND RESULTS Harvesting and processing operations of the Florida orange juice industry are analyzed using mathematical programming mo dels to estimate industry profits as a function of an a nnual harvesting schedule Industry production models simulate operation during a typical harvesting and processing season, with constraints designed to mimic actual industry conditions. For example, load allocation restrictions assumed in the model are set based on weekly averages allowed by processors to simulate an actual processing season. Models are initially estimated under conditions attempting to replicate current market conditions and then a ltered to mimic changing industry conditions. Changes to conditions allow for operation with fewer limitations, including the use of mechanical harvesters, altering the length or continuity of the processing season, or changing processing or storage capac ities. The models compare gains from harvesting and processing under varying constraints to determine the schedule that will deliver the greatest monetary returns to the industry. Comparison of these returns indicates whether changes in industry operatio ns could allow for improvements to total industry returns. Modeling results are used to predict how changes to industry operation and capital availability change total revenues. Additionally, minimum capital requirements can be established for mechanical harvesting capacities, as well as the required extraction and storage capacity needed to match this harvesting capacity. Grower Decision Model Equation 5 9 defines the variables that determine grower returns from fruit production. Further specification of the dependent variables with respect to geographic location, tree variety, and time of harvest is shown in Equation 6 1.
158 (6 1) D efinitions for all variables used in this chapter are listed in Table 6 1. T he variables used in Equ ation 6 1 include returns to a G ) that are summed across all fruit varieties ( v ), geographic regions ( i ), tree age groups ( a ), and time periods ( t ) as a function of the delivered in price per pound solid ( DIP ), pound solids per box ( PS ), boxes per acre ( B ), fruit production co st per acre ( UPC ), percentage of fruit harvested by hand ( ), hand harvest cost per box ( UHC HH ), mechanical harvest cost per box ( UHC MH ), and the acres of grove harvested ( A ). A harvest constraint variable ( X viat ) is used to represent the harvest status o f each grove aggregate during each time period, w here each grove aggregate denote groves of the same variety, region, and age. A harvest constraint variable value of 0 repr esents no harvest during the given time period, while a value of 1 represents compl ete harvest of the aggregate For example, if 75% of fruit from a certain variety, region, and age is harvested during Week 34 and the remaining 25% is harvested during Week 35, then the value of the harvest constraint variable would be 0.75 for Week 34, 0.25 for Week 35, and 0.00 for all other t ime periods. Growers maximize returns by choosing the harvest date of a specific grove aggregate specified by variety, region, and tree age, during the optimal harvest time period ( t* ) Assuming acreage is fixe d during the harvest season and that all prices and costs are exogenously determined by market forces, harvest timing and harvest method are the only endogenous variables determining grower returns that are influenced by individual grower decisions. The h arvest timing decision impacts total pound solid production and box yield, because both are constantly changing throughout the season.
159 The type of harvest used, either hand or mechanical, also influences grower returns because of differences in unit cost s of harvest for each method. Defining to represent the percentage of boxes harvested by hand, total hand harvest costs are calculated as the unit harvest cost to hand harvest ( UHC HH ) times the quantity of boxes harvested by hand ( ( B A )). Similarly, the total cost to harvest mechanically is calculated as the unit harvest cost to mechanically harvest ( UHC MH ) times the boxes harvested mechanically ((1 ) ( B A )). Equation 6 2 describes total harvest costs (6 2) Constraining the grower decision model are the physical lim itations of harvest, including biological limitations and processor capacity limitations. If the grower were the sole decision maker determining the harvesting schedule, harvest timing would be based only on the fruit maturity Harvest would occur when p ound solids production was at its maximum, and the least cost harvest method would be employed. The only constraint is the limitation that each grove could only be harvested once per season (Eq. 6 3). Subject to for each v i, a nd a where (6 3) Equation 6 particular grove aggregate and requiring each aggregate to be harvested no more than once. The harvest constraint all ows for partial harvesting of a grove aggragate across several time periods by permitting any X viat to be less than one during any time period and requiring all X viat for a specific variety, region, and age to sum across all time periods to less than or e qual to one. While not tested in this analysis, t he harvest decision could be further restricted to require an entire grove aggregate to be harvested during a si ngle time period by requiring X viat to equal one if aggregate via is harvested in time t and z ero at all other periods. Requiring complete
160 harvesting within a single period could be used to simulate the high cost of disassembling and transporting mechanical harvesters over the road The physical limitations of juice processors are included in the factors constraining the grower decision model. Growers must harvest weekly fruit volumes corresponding to the volume requirements of processing facilities due to the quick rate of fruit spoilage after harvest. Growers and processors must come to an agre ement on the weekly volume exchanged. Agreed upon volumes are manifested through weekly load allocations assigned by processors to growers. Equations 6 4 and 6 5 represent the lower and upper bound load allocation constraints respectively. (6 4) (6 5) The load allocation constraints represent the quantity of fruit growers are permitted to deliver to the processor. UB represents the upper bound quantity of 90 pound boxes of fruit allowed to be delivered to the processor during each time period and LB represents the lower bound quantity of boxes that must be delivered to the processor in each time period. Loads delivered beyond the upper bound limit may not be accepted. Conversely, growers who consistently de liver below their allocation may have their weekly load allocations reduced. Processor Decision Model programming model executed in GAMS. The processor model analyzes pro duction decisions as a function of prices, pound solids volume, box yield, juice extraction costs, and bulk juice P ) by selecting the optimal
161 volume of daily fruit deliveries throughout the season, and then minimizing the cost of processing these volumes of fruit Assignments of load allocations di ctate daily processing volumes, setting the duration and intensity of harvest, extraction, and storage. Under this model, growers deliver fruit to meet load allocations as processors attempt to optimize plant operations and minimize costs subject to the p hysical operational constraints and fruit maturity. Equations 6 6 through 6 16 depict the processor decision model. (6 6) Subject to for each v i, and a where (6 7) (6 8) (6 9) (6 10) (6 11) (6 12) (6 13) (6 14)
162 (6 15) (6 16) Variables representing the unit processing margin ( PM ), the cost per box to extract juice ( UEC ), the unit cost to store juice for one period ( USC ), and the volume of juice in storage during each period ( P SS ), for both NFC production ( N ) and FCOJ production ( C ), are introduced into the processor model. Constraints on processor operation s are more numerous and complex than constraints on grower operations. The first three constraints listed in the process or model, the harvest variable constraint (Eq. 6 7) and the lower and upper bound load allocation constraints (Eq. 6 8 and 6 9), are identical to the grower model. Additional constraints represent processing plant capacities for both NFC and FCOJ weekly t hroughput extraction capacities, total storage capacities for NFC and FCOJ, and juice quality constraints. Equations 6 10 and 6 11 represent the weekly throughput extraction limitations of processing plants for NFC and FCOJ respectively. The actual volume of boxes harvested and processed during a given week is represented by Equation 6 17. (6 17) Calculated similarly to the upper and lower bound constraints, TB is the estimated sum of boxes processed weekly. Further, processors dete rmine which percentage of fruit inputs are processed to NFC ( NFC %) and which percentage of inputs are processed to FCOJ ( 1 NFC % ). Multiplying these percentages by the weekly box input volumes allows for the calculation of the total boxes
163 processed weekly to NFC ( TB N ) and the total boxes processed weekly to FCOJ ( TB C ), as shown in Equations 6 18 and 6 19. (6 18) (6 19) The constraints shown in Equations 6 10 and 6 11 require that the maximum box t hroughput of each product remains below weekly NFC processing capacity ( UTB N ) and weekly FCOJ processing capacity ( UTB C ). Total weekly capacities of each product are determined by the physical capacity limitations of the processing industry. Storage capa city for juice and concentrate end products also constrain processor operation. Processors have a limited storage capacity specific to each product. Equations 6 12 and 6 13 limit the volume of NFC and FCOJ in storage during any time period, and to be less than or equal to the total storage capacity available for each product, UPSS N and UPSS C respectively. The volume of juice or concentrate in storage during any given time period is calculated by summing the volume in storage at the end of the previous period, or with the volume change during the current period, or Equations 6 20 and 6 21 represent the volume change of stored product during the period, for juice and concentrate respectively, by subtracting the output volume during the period ( OUT ) from the total pound solids processed during the period ( TPS ). (6 20) (6 21) The output volume for each period represents any juice or concentrate removed from the processing plant through sales, transfers, or disposal. The total pound solids of each product type
164 produced during a period is calculated by multiplyin g the total boxes of each type processed, by the average pound solids, as shown in Equations 6 22 and 6 23. (6 22) (6 23) constrained by juice quality characteristics. The terms of sale for most juice products are dependent on quality; therefore, processors strive to achieve a juice product that meets required quality parameters. Two closely monitored juice quality attribu tes are the soluble solids content, measured in degrees Brix, and the Brix to acid ratio. Total soluble solids and the Brix to acid ratio of juice can be estimated as time dependent functions, similar to estimates developed in the grower section of the mo del where pound solids per box, fruit drop rate, and fruit weight are estimated as functions over time. The functional form chosen to approximate these changes is assumed to be a second order polynomial. Explicitly, total soluble solids ( TSS ) changes ove r time can be represented by the second order polynomial function shown in Equation 6 24, and Brix to acid ratio changes over time can be expressed by the function shown in Equation 6 25. (6 24) ( 6 25) FASS sample data are used to estimate the quality parameters of processed juice. Early in the processing season, soluble solids have not had time to accumulate, so processors try to run only fruit with soluble solid levels above a given degree Brix value. While the actual lower limit degree Brix value required by the State of Florida to be met before fruit can be delivered may change with seasonal conditions, federal guidelines require a minimum degree Brix between 8 degree and 9 degree (USDA, FASS 2007), and processors have reported
165 requiring at least 10 degree Brix 1 2 Soluble solids measured in degree Brix are numerically equivalent when reported as the percentage of juice (Matthews, 1994). For example, a total soluble solids concentration of 11% is equivalent to a juice of 11 degree Brix. Equation 6 14 constrains fruit inputs to be greater than or equal to minimum soluble solid requirements. Brix to acid ratio quality requirements are also included in the modeling constraints. Florida stat e requirements hold that fruit must achieve a minimum ratio between 9 and 10.5 before the fruit can be processed. Processors often have stricter lower bound requirements on the ratio of fruit inputs to e nsure juice quality. One processor in Florida typic ally requires a minimum 14 ratio for early season fruit, or payments are reduced due to low juice quality. While no upper bound constraint on ratio is specifically outlined, processors discourage juice of exceedingly high ratio. The state required 0.4% m inimum acid concentration effectively sets a maximum ratio at about 20 prohibiting the sale of insipid tasting juice (Wardowski, 1995) Equation 6 15 constrains the lower bound ratio ( LRATIO t ) and the upper bound Brix to acid ratio ( U RATIO t ) of fruit inp uts to conform to both legal requirements and typical processor requirements. In addition to state requirements on fruit inputs, federal standards for processed juice and concentrate require U S D A Grade A juice to have a ratio of 12.5 to 20.5 for single s trength juice and 12.5 to 19.5 for concentrate. Again, these standards are often stricter for processors demanding a high quality end product. Equation 6 16 restricts the average Brix to acid ratio for the entire season within lower and upper bounds ( LAV GRAT and UAVGRAT ), meeting USDA, Florida Department of Citrus, and typical processor requirements. 1 Personal communication with anonymous manager (1) of a Florida orange processing facility, July 2008 2 Personal communic ation with anonymous manager (2) of a Florida orange processing facility, July 2008
166 This model depicted in Equation 6 6 through 6 16 represents processor returns and includes several variables processors have the ability to alter, thus a ffec ting their economic gains or losses. Most important among these variable are the operational decisions that impact the total pound solids of juice captured for the season. In addition to volume of fruit purchased, the timing of harvest and extraction cha nges the total pound solid volume intake. Processor s work to minimize these extraction and storage costs. Extraction costs are subject to the increasing returns to scale seen in most large manufacturing operations with high amounts of fixed capital. Proc essors attempt to spread the high fixed cost of capital across as many boxes of fruit as possible to lower unit processing costs. The extent to which these costs can be reduced depends nd marketing strategy. Similarly, storage costs are impacted by production decisions. Processors attempt to minimize these costs with efficient use of existing capital and optimal inventory management. Industry Decision Model The grower and processor m odel s are combined into a decision model for the entire Florida orange juice processing industry. As previously shown, total industry returns ( IND ) equal the sum of grower and processor returns (Eq. 3 4). Total industry returns are maximized by summing Equations 6 1 and 6 6, subject to all the operational and quality constraint equations, while allowing the programming model to choose the profi t maximizing harvest period for each grove aggregate as shown in Equations 6 26 through 6 37.
167 (6 26) Subject to: for each v i, and a where (6 27) (6 28) (6 29) (6 30) (6 31) (6 32) (6 33) (6 34) (6 3 5) (6 36)
168 (6 37) The industry decision model is solved under varying market conditions to simulate the economic effect of changes in the duration and intensit y of harvesting and processing. Altered load allocations, harvest costs, and harvesting methods simulate varying levels of mechanical harvester adoption. The resulting changes in industry costs and revenues allow for a better understanding of the impacts that mechanical harvesting will have on the Florida citrus industry. Use of mechanical harvesting is simulated by lessening the percentage of the crop constrained to be harvested by hand (Eq. 6 31). The model then adjusts the scheduling of harvest to ma ximize total industry returns. For example, total and unit costs to harvest are expected to decrease when harvesting mechanically (Roka and Rouse, 2004a), and pound solids yields and storage requirements are expected to increase due to harvesting closer t o the pound solids production optimum (Searcy, Roka, and Spreen, 2008). If the duration of the season is decreased due to more intense processing, then higher storage costs are being offset by decreases in processing and harvest costs. While the focus of this modeling process will be to quantify further analysis to address additional issues such as industry changes from greater NFC production or new fruit varietie s. The programming model selects the optimal date of harvest for each grove aggregate by balancing pou nd solids yield, box yield, harvest cost processor constrains, and juice quality parameters The model is run using grower level data on grove acreage and characteristics, as well as estimates of historic load allocations allowed by the processor. Estimated equations for
169 fruit weight, juice weight, pound solids per box, juice yield, total soluble solids, and the Brix to acid ratio are embedded in the mo del as functions of harvest date, allowing for variations across region, variety, and tree age. The model calculates the maximized returns to the industry, the pound solids returned for each time period and each grove aggregate and the annual harvest sch edule yielding the greatest industry returns. Additionally, weekly yield estimates are combined with machine capacities to estimate the minimum harvesting and extracting capital requirements needed to operate under the specified market conditions. Estima ting the minimum number of harvesters needed is calculated by dividing the weekly box peak volume by the average weekly harvesting capacity per machine. Similarly, the minimum extraction capital needed can be calculated by dividing the weekly box peak vol ume by the average weekly extraction capacity per processing line. This modeling exercise assumes unifor m fruit and juice quality within each period, with all variability coming between fruit variety, region, and processing time periods. This assumption assures that all fruit of the same variety, region, and harvest date, will be of equal quality and will yield equal quality juice. Fruit processed the following week will have different quality and yield characteristics as dictated by biological changes in the fruit. The model further assumes that box yields are not affected by the method of harvest. Therefore, all else being equal, a grove picked with hand labor or mechanical harvesters should produce equivalent yields. Maturity modeling estimates jui course of a single representative season. While based on data collected by FASS to predict annual fruit production, the maturity model presented in this paper is designed as part of a simulation model and is not intended to be used as a production forecasting model. Rather, it estimates total industry returns from a static, single season. No similar model is publicly
170 available that views total returns to the entire Florida orange j uice industry. Calculating revenues as the summed value of pound solids production also differs from other citrus production models that view yield in terms of field boxes ( USDA, FASS, 2007 ). The revenue calculation difference is intended to be a more pr ecise revenue measure because Florida growers are paid on the basis of total pound solids delivered. Description of Data Processing Data Orange processing data used for this analysis is compiled from two primary sources, the Florida Citrus Processors Ass ociation (FCPA) and personal interviews with citrus processors. Summed state level processing data are publicly available from FCPA on fruit inputs delivered to processing facilities, including information on box yield, pounds solids, percent acid, ratio, and harvest date. FCPA data include weekly reports on FCOJ and NFC production, industry storage levels, and juice movement. These data are used to estimate weekly and annual processing and storage capacities for the industry, as well as estimating quali ty parameters of juice entering the plant. Firm specific data of plant operations are proprietary and not publicly available from individual orange processors. All firm data were, therefore, collected through personal contacts with processing plant manag ers. Because of this limited release, most data reported by processing managers are estimated values and are not linked to an identifiable source due t o confidentiality agreements. Processing data, including price data, used for this analysis are reported in Table 6 2. Delivered in prices, as reported by the FCPA, have averaged $1.0 4 per pound solid over the 10 seasons from 1998 to 2008 The lowest seasonal average of $0.71 per pound solid was recorded during the 2003 04 season, and the highest average p rice per pound solid of $2.11 was recorded
171 during the 2006 07 season. Early and mid season fruit delivered in prices averaged $0.94 per pound solid and late season fruit averaged $1.16 per pound solid, a difference of $0.22. Free on board (FOB) bulk FCO J prices are reported by Florida Citrus Mutual (FCM), Statistical Division, for the same ten year period. FOB prices represent the value of finished juice loaded for transporting away from the processing facility. The lowest seasonal average was $0.85 pe r pound solid during the 2003 04 season and the highest price per pound solid was $2.07 during the 2006 07 season. Average seasonal prices resulted in a n estimate of the average processing margin over the 10 year study period to be $ 0 .15 per pound solid This estimate was derived as the average difference between the delivered in price s reported by the FCPA and the FOB price s reported by FCM. While similar data on NFC is not available, an average increase of $0.20 per pound solid in the selling price of NFC was estimated for NFC over FCOJ. 3 The resulting average processing margin for NFC is estimated at $0.35 above the delivered in price. In addition to the value of processed juice, processors also receive revenues from the sale of byproducts remaining f rom juice production. These byproducts are estimated to be valued at approximately 5% of the delivered in price of fruit. 4 By this estimate, an ave rage delivered in price of $1.50 would coincide with a byproduct value of at least $0.05 per pound solid. While the actual prices realized will change with market conditions, processors use similar pricing estimates to make production decisions. There are no current, publically available estimates of the unit cost of extraction, defined for this analysis as th e unit cost to convert raw fruit inputs into a marketable juice product. The average cost to wash, grade, and juice fruit is estimated at approximately $0.02 per pound 3 Personal communication with anonymous manager (1) of a Florida orange processing facility, July 2008 4 Personal communication with anonymous manager (2) of a Florida oran ge processing facility, July 2008
172 solid. 5 6 7 Processing managers estimate that pasteurization, a step in the NFC extrac tion process, has an average variable cost of approximately $0.02 per pound solid, and the average variable cost of evaporation, required to produce FCOJ, is estimated at approximately $0.03 per pound solid. 8 Economic theory suggests that the assumed cost s of extraction, pasteurization, and evaporation should experience economies and diseconomies of scale consistent with production using large capital investments These dynamic cost changes are experienced in the industry but are not fully captured in thi s model due to the limited availability of processing cost data. Processing capacities are estimated for this analysis based on historical processing levels reported by the FCPA. The maximum weekly volume of NFC processed during the 2007 08 season was 3, 471,700 pound solids, and the maximum volume of FCOJ processed was 3,169,800 pound solids. Juice storage costs are estimated using the storage cost tables from the ICE futures contract fees schedule ( Intercontinental Exchange, 2008). Monthly storage, hand ling, and inspection prices represent the cost to store product with an outside party, typically at a much higher cost than the actual storage costs for processors who own storage capacity. The cost to store FCOJ is $500 for the input and output handling fee, $300 per month for storage, and $100 for re inspection for each 15,000 pound solid contract. The monthly storage fee equates to $ 0 .02 per month or $0.005 per week of storage. Per contract fixed fees are $600 per contract or $0.04 per pound solid. N FC futures contracts, defunct as of January 2008, had a similar ICE storage schedule before trading ceased (ICE, 2008). The last reported fee schedule for NFC storage was 5 Personal communication with anonymous manager (2) of a Florida orange processing facility, July 2008 6 Personal communication with anonymous manager (1) of a Florida orange processing facility, July 2008 7 Personal commu nication with anonymous manager (10) of a Florida orange processing facility, July 2008 8 Personal communication with anonymous manager (2) of a Florida orange processing facility, July 2008
173 $1,440 per month storage per 12,000 pound solid contract. This monthly storage fee is equivalent to $0.12 per pound solid, or $0.03 per pound solid per week. No additional information was given as to handling and re inspection fees, so fixed fees are assumed equivalent to fixed fees for FCOJ of $600 per contract, which is equivalent to $0.05 per pound solid of NFC. Processor estimates of the variable cost of storage for NFC juice average $0.04 to $0.06 per gallon of NFC when stored in available tanks. 9 This estimate is the average unit cost to store and does not include a weekly cost co mponent. Assuming an average storage length of six months and a cost of $0.06 per pound solid, the variable cost of NFC storage to a processor with available storage capacity is approximately $0.0025 per pound solid per week. These estimates further assu me an average degree Brix value of 11.5 degree to allow one gallon of juice to equal to one pound solid (FDOC, 2008 a ). While similar cost estimates are not available for FCOJ storage, cost estimates are generated by assuming a one sixth reduction in the c ost of storage, equivalent to the difference in ICE storage fees. FCOJ cost estimates calculate to approximately $0.0067 to $0.01 per pound solid annually. Again assuming an average storage time of six months, the weekly cost of FCOJ storage to a process or with available storage capacity is approximately $0.0004 per pound solid per week. Available storage capacity is estimated from FCPA data representing on hand juice volume in storage during the 2003 through 2008 processing seasons ( FDOC, FCPA, 2009). T otal NFC capacity is estimated at 324,200,000 pound solids, equivalent to 315,100,000 gallons at 11.8 degree Brix (1 gallon = 1.029 pound solids) (FDOC, FCPA, 2009). Total FCOJ capacity is estimated at 831,200,000 pound solids, equivalent to 200,000,000 g allons of concentrate at 9 Personal communication with anonymous manager (2) of a Florida orange processing facility, July 2008
174 42.0 degree Brix (1 gallon = 4.156 pound solids) (FDOC, FCPA, 2009). Average carryover volume is subtracted from the average total volume to estimate the volume of storage available at the beginning of a typical processing season. Carryover juice, held across seasons for blending and to protect against a stock out, is assumed to be a sunk cost of operation and is not included in the available capacity estimates. Carryover of 87,000,000 pound solids is estimated for NFC and 415,60 0,000 pound solids for FCOJ ( FDOC, FCPA, 2009). Available NFC storage is estimated at 237,200,000 pound solids and available FCOJ storage is estimated at 415,600,000 pound solids. FCPA juice movement data is also used to estimate weekly outflows of NFC ( ) and FCOJ ( ). Average NFC outflows are estimated at 10,290,000 pound solids per week and FCOJ outflows are estimated at 9,559,000 pound solids per week ( FDOC, FCPA, 2009). Processed juice and concentrate qualit y parameters are set to meet minimum government standards, ICE futures market quality standards, or commonly used industry parameters mandated by processors. These quality guidelines are designed to assure model results stay within the quality bounds that regulate industry operations. Minimum and maximum quality parameters assumed for this analysis are presented in Table 6 2. Harvesting Data Harvest cost and volume data are gathered from published literature on current hand harvest costs, as well as curre nt and projected costs of mechanical harvesting. Mechanical harvesting currently plays only a small role in the annual fruit harvest, so most estimated harvest cost data assume industry operation under hand harvest. Estimated annual industry cost average s for hand harvest (Muraro, 2004) included in this model are compared against estimates of expected costs of harvest with adoption of mechanical harvesters. Simulating harvest using 5%
175 and 95% mechanical harvesting produces estimates of potential cost cha nges as the new technology moves towards industry wide adoption. In the hand harvest labor market, Muraro (2006) estimated average picking charges of $0.982 per box for early and mid season oranges and $1.055 per box for late season oranges, with average per box roadsiding charges of $0.963 and $1.011 respectively. Hauling charges for all types of fruit averaged $0.498 per box for hauls of 31 to 50 miles. Research by Polopolus and Emerson (1993) suggests that productivity of harvest workers averages betw een eight and 10 boxes per hour and Roka and Cook (1998) suggest that workers average between 30 and 35 350 boxes per week. Additional evidence indicates that harvest costs vary directly with crop yield and time of year ( u npublished data from personal interviews with selected citrus harvesters during the 2003 04 season conducted by B. Hyman and F.M. Roka, 2004 ). For example, seasonal cost changes occur in part due to the out migration of labor. Workers begin leaving in the late spring, before the end of the harvest season, because of deteriorating work conditions and the greater availability of other higher paying seasonal agricultural jobs. Firm specific da ta are not readily available on the costs and capacities of mechanical harvesting due to the small size and competitive nature of the industry. C ost savings of up to $1.00 per box less than hand harves t costs have been estimated when mechanical harvesting becomes fully adopted (Roka, Brown, and Muraro, 2000). These lower harvest costs assume that the cost efficiencies afforded through mechanical harvesting are captured by the grower and not third party harvesters. The competitive nature of the mechanical harvesting industry could allow some profits to be captured by the harvesting industry, lowering cost saving accruing to growers. The harvest capacity of mechanical harvesters depends on the technology being employed and
176 the number of operational machine s in the field. A set of the continuous canopy shake and catch machines currently being used have shown harvest capabilities of at least 12 highway trucks, approximately equivalent to 6,000 boxes, per eight hour day. A set of trunk shake and catch machin es can harvest about three trailers, approximately 1500 boxes, of fruit per eight hour day (Brown, 2002) Table 6 2 summarizes the estimated costs and constraints of harvesting used for this analysis. Bearing Acreage The 2008 Florida Agricultural Statist ics Service Commercial Citrus Inventory ( USDA, FASS, 2009) lists bearing acreage of orange groves by region. Fruit variety specific plantings in this most current release only presents information on the ten counties with the most acreage. Data are now o nly reported on the ten counties with the most acreage to meet Federal disclosure requirements. To estimate the actual bearing acreage for all counties, the 2008 tree inventory is used in conjunction with the 2004 Commercial Citrus Inventory ( USDA, FASS, 2005), which lists acreage in all counties. Total orange acreage in 2008 was measured at 496,518 acres with 463,994 bearing acres. This is a 20% reduction in total acreage from the 622,821 acres reported in 2004 and an 18% reduction in bearing acreage, 5 62,755 acres. Table 6 3 lists the bearing acreage by region for 2004 and estimated for 2008 using the two most current citrus inventories. GAMS Modeling The industry decision model linear programming problem is solved using the software package GAMS (Gen eral Algebraic Modeling System) and the MINOS algorithm. GAMS/MINOS is a general purpose solver that uses a reduced gradient technique designed to solve linear programming problems by finding a local optimum ( Murtag h, Saunders and Gill 2008). The solve r maximizes total industry returns from constrained production models using g rower and processor data Constraints include biological characteristics of fruit production, as
177 well as estimates of current fruit production, juice extraction, and storage capa cities. All models are based on fruit maturity and drop functions estimated using biological fruit and juice data collected by the Florida Agricultural Statistic Service from 1996 to 2007, as described in Chapter 4. Tree acreage values are estimated from the 2004 and 2008 Commercial Citrus Inventory ( USDA, FASS, 2005 ; USDA, FASS, 2009). Four models are estimated under varying capacity levels and harvesting methods to simulate changes in industry operation. The models maximize the combined profit of gro wers and processors as a single, vertically integrated company would maximize their total profit The solver chooses the weekly acreage to be harvested of each variety, within each geographic region, and each age group. This weekly harvest decision sets the harvest schedule maximizing The complete GAMS programming code is shown in Appendix D. Constrained Industry Production Models Assuming 2007 08 Processing Volumes The first two constrained industry production models (CIPM1 and CIPM2) are es timat ed under conditions constrained to simulate the actual weekly processing volumes for the 2007 08 season, as reported by the FCPA. Setting the weekly box volume processing capacity (Table 6 4), storage capacity, and output volumes of both NFC and FCOJ equal to the volumes reported in the 2007 08 Processors Report ( FDOC, FCPA 2009) assumes this is an optimal schedule as determined by the Florida Citrus Industry. The processing schedule is feasible given the 2007 08 levels of fruit production, harvest capacity, processing capacity, and storage capacity. The first model, CIPM1, constrains harvest to 95% hand harvest and 5% mechanical harvest to simulate an annual harvest using almost all han d harvest and a very small portion of mechanical harvesting. The assumption of 5% mechanical harvest is approximately equal to the 9.6 million boxes (5.78%) harvested mechanically during the 2007 08 season (FDOC, FCPA,
178 2009). Delivered in fruit prices are assumed to be $1.50 per pound solid and the maximum weekly harvest volume is constrained at 6,324,150 boxes, the largest weekly cumulative volume of fruit harvested by all Florida growers during the 2007 08 season. Additional price and constraint values are listed in Table 6 2. The second model, CIPM2, is identical to CIPM1 except that hand harvest is constrained to 5% and 95% mechanical harvesting, simulating an industry shift towards total industry wide adoption of mechanical harvesting. The small per centage of remaining hand harvesting represents groves that cannot be adapted to allow for mechanical harvesting and must be harvested by hand. While the actual acreage of commercial groves that cannot be mechanically harvested is unknown, the 5% hand har vest restriction acknowledges that a small portion of acreage may not be conducive to mechanical harvesting. Mechanical harvester volume estimates for this model assume that one set of mechanical harvesters is capable of removing 35,000 boxes per week, eq uivalent to 7,000 boxes (14 trailers) per day operating five days per week. While some harvest volume estimates predict harvest rates as high as 11,000 boxes (22 trailers) per day (Oxbo International Co., 2009) and over 1,200 boxes per hour (Roka and Hyma n, 2004), a 35,000 box weekly average assumes operating only one shift of operators per day with stoppages due to equipment setup, takedown, transportation, worker breaks, inclement weather, and mechanical breakdowns. Under these constraints, harvesting 9 5% of the 2007 08 crop mechanically would require operating approximately 172 sets of harvesting machines during some weeks of the season. The number of harvesters required is determined by the harvest volume needed during the week of peak processing. Ha rvesting only 5% of the crop mechanically requires nine sets of machines (Table 6 5). Increasing machine operational hours by using multiple shifts and running closer to 24 hours per
179 day could substantially reduce the total number of harvesters required t o meet peak harvesting demands. Models CIPM1 and CIPM2 maximize total industry returns under harvesting and processing constraints designed to approximate industry conditions with first very little mechanical harvesting and then with almost full industry adoption, respectively. Direct comparisons of CIPM1 and CIPM2 results explain the expected industry changes arising when hand harvesting is replaced with mechanical harvesting, holding weekly processing volumes constant. Changes to industry operation an d returns are applicable to a scenario in which mechanical harvesting is used as a substitute for hand harvesting, lower ing the unit cost of harvest. Comparisons between CIPM1 and CIPM2 are intended to simulate operation under lower harvest costs and do n ot attempt to simulate or explain additional costs or benefits of mechanization due to changes in processing operation s Summary results for industry models are shown in Table 6 5. Weekly processing results, including total boxes processed and the shadow values associated with weekly box constraints, boxes processed to NFC, boxes processed to FCOJ, pound solids of NFC captured, and pound solids of FCOJ captured, are shown for CIPM1 in Table 6 6. Table 6 7 shows the same result categories for CIPM2. Figur es 6 1 and 6 2 show NFC and FCOJ pound solid production graphically for each model. Table 6 8 summarizes weekly harvesting results for both models, detailing the weeks each grove aggregate is harvested and the percent harvested each week. Table 6 9 shows the estimated weekly volumes of NFC and FCOJ in storage for both CIPM1 and CIPM2. Weekly NFC and FCOJ storage volumes are also shown graphically in Figures 6 3 and 6 4. These figures include the average weekly volumes carried by the industry from 2000 t hrough 2006 as reported by the FCPA for comparison to model estimates.
180 CIPM1 Results CIPM1 estimates total industry returns of $1,777,637,205 with growers receiving $1,338,329,624 ( 75.3% ) and processors receiving $439,307,581 (24.7 % ). This model is desig ned to simulate an average citrus season and is not attempting to precisely estimate returns during any particular season. Estimated results are, however, in the neighborhood of the citrus crop values of $1.5 billion and $1.2 billion for the 2006 07 and 2 007 08 crops respectively ( USDA, FASS, 2009 a ). Model results do not estimate the same industry components as the FASS crop value estimates, which estimate the on tree equivalent value of all Florida citrus crops. No published values of industry returns a re available that estimate the same components being modeled in this analysis. The annual FASS estimates may be the most similar data available and, therefore, are reported to demonstrate that model estimates are of the appropriate magnitude. CIPM1 resul ts are used as a baseline to which industry changes are compared. Simulation results are designed to be used to analyze the changes in industry returns when changes occur in industry operation. The CIPM1 simulated a processing season that begins in Week 10 and ends in Week 50, approximately from the first week of October until the second week of July. Start and stop dates and weekly processing volumes for CIPM1 are set by model constraints to replicate 2007 08 values. W eekly volume constrain ts limit pr ocessing and cause about 35,000 acres (8%) to not be harvested. NFC production runs from Week 12 to Week 50, halting during weeks 14, 16, 17, and 18. FCOJ production takes place during weeks 10, 11, 14, and 16 through 47. Total box production is 165,999 ,586 with 86,896,842 boxes (52.3%) going to NFC and 79,102,744 boxes (47.7%) going to FCOJ. Total box production was constrained by the model to mimic 2007 08 box volumes. This harvest and processing schedule yields a total pound solid production of 1,1 10,787,043, with 573,853,950 pound solids (51.7%) going to NFC and 536,933,093 pound
181 solids (48.3%) going to FCOJ. The FCPA reported 2007 08 total season production of 1,170,923,336 pound solids, putting the model estimate 60,136,293 pound solids ( 5.1%) below actual production. Actual NFC production was reported at 609,090,877 pound solids and FCOJ production was reported at 561,832,459 pound solids, differences of 24,899,366 ( 4.43%) and 35,236,927 pound solids ( 5.79%) respectively. The volume of pro cessed NFC in storage fluctuated from 90,006,023 pound solids during Week 19 to 345,534,023 pound solids during Week 48. Weekly FCOJ volumes in storage fluctuated from 383,347,049 pound solids during Week 17 to 637,557,049 pound solids during Week 45, vol umes equivalent to 92,040,108 and 153,074,922 gallons of FCOJ at 42 degree Brix. The marginal values, or shadow values, for weekly box volumes increase from $4.24 in Week 1 to $10.84 in Week 49 before beginning to decline. These shadow values estimate th e increase in industry revenue from processin g an additional box of fruit. Low v alues estimated during the earlier weeks in the season reflect the low returns expected because a box of fruit contains very little juice early in the season an d will be in st orage for a longer time Shadow v alues increase later in the season as the pound solids per box increase and the time in bulk storage decreases. Strictly positive values each week indicate a binding capacity constrain and that loosening processing constr aints during any week of the season to allow more fruit to be processed would increase i ndustry returns. Allowing additional processing would enable previously un captured boxes to be captured, with the value of the increase in returns dependent on the tim e of harvest. Shadow values for storage are zero during all weeks except Week 48, which had an estimated shadow value of $ 0 .15. This shadow value represents the increase in industry returns given an additional pound solid of storage capacity. For exampl e, total industry returns would increase by about $150,000 annually if the
182 industry added one million pound solids of additional NFC storage, assuming this estimated shadow value remains constant for each pound solid added (a 0.3% increase in storage volum e). CIPM2 Results Weekly processing results for CIPM2 estimate total industry returns of $1,851,519,436, with growers receiving $1,413,519,334 (76.3%) and processors receiving $438,000,102 (23.7%). Total returns are estimated to increase by $73,882,231 ( 4.2%) over CIPM1 returns, resulting in a $75,189,709 (5.6%) increase in returns for growers and a $1,307,478 ( 0.3%) decrease in returns for processors. The addition of mechanical harvesters is therefore shown to positively impact returns for growers whil e having a very small negative effect on processors, as compared to harvesting prima rily by hand. Grove acreage not harvested increased slightly to about 9%, (2,700 fewer acres), but total boxes processed and pound solid production remained relatively unc hanged between the two models. The volume of NFC in storage fluctuated from 89,106,023 to 345,534,023 pound solids and 384,147,049 to 622,356,049 pound solids for FCOJ, equivalent to 92,232,185 to 149,425,222 gallons of FCOJ at 42 degree Brix. Storage le vels are similar between CIPM1 and CIPM 2, with peak storage levels occurring within a few weeks of each other. NFC production begins Week 15 and then restarts a few weeks later to run continuously from Week 19 to Week 50. FCOJ production runs from Week 1 0 to Week 48, pausing only during Week 15. Season start and stop dates are shown to be fairly consistent between models with FCOJ production beginning and ending a few weeks earlier than NFC production in both cases. Similar to CIPM1, marginal values incr eases from $4.69 in Week 1 to $11.29 in Week 49. Shadow values for CIPM2 are higher, potentially because the lower cost of harvest allows for higher per unit returns.
183 Constrained Industry Production Model s Assuming a Single Fixed Processing Maximum Week ly processing volumes for t he constrained industry production models assuming a single fixed processing maximum (CIPM3 and CIPM4) are constrained by the highest single week ly processing volume observed during the 2007 08 season Production is no longer co nstrained by the volumes observed each week during the season, as with CIPM1 and CIPM2. Loosening this processing volume constraint assumes processing facilities are capable of operating at the 2007 08 maximum weekly volume during every week of the season Th is loose ning of the processing constraints assumes that it is possible for a processor to operate their plant at full processing capacity for many successive months, but does not require or suggest at the availability of fruit, harvest capacity, or st orage capacity This change is intended to allow the processing constraint to better represent only processing capacity. As previously specified, the values constraining processing during the 2007 08 season could have been influenced by other volume limi ting factors experienced during the season, such as fruit availability, harvesting capacity limitations, or hauling capacity limitations. For production models with a single fixed weekly processing maximum, removing the 2007 08 weekly volumes assumes frui t availability is determined solely by the models estimating biological characteristics, and maximum harvest and hauling capacities are assumed sufficient to supply fruit volumes matching the processing capacity. If harvesting, hauling, and processing cap acities are sufficient during the peak week production, then this same capacity is assumed to be available during any week of the season. Storage capacity estimates are based on peak volumes on hand in recent years for each juice product ( FDOC, FCPA, 2009 ). Using only the maximum storage capacity for NFC and FCOJ assumes weekly processing schedules are not impacted by weekly storage management decision, such as adjusting storage volumes for different types and qualities of juice. For example, the process ing volumes may have been limited during certain weeks of the 2007 08 season to allow
184 some storage tanks to be emptied of a certain type or quality of juice to make room for juice of different quality parameters. Such storage management decisions could ha ve influenced the 2007 08 schedule but are not considered in production models limited by a single weekly maximum. The maximum weekly box volume processed during the 2007 08 season was 6 324 150 boxes, run the second week of April (Table 6 4). Hand harves t of 95% of the crop is simulated with model CIPM3 and mechanical harvest of 95% of the crop is simulated with model CIPM4. All other parameter and constraint values from CIPM1 and CIPM2 are repeated for CIPM3 and CIPM4. Storage capacity and output volum es of both NFC and FCOJ are set equal to the volumes reported in the 2007 08 Processors Report ( FDOC, FCPA 2009), and the delivered in fruit price is again assumed to be $1.50 per pound solid. Additional price and constraint values are listed in Table 6 2. Industry returns are maximized subject to these physical constraints to simulate operation with full utilization of weekly processing capacity. Summary results for both CIPM3 and CIPM4 are shown in Table 6 5. Processing results for CIPM3 are shown in Table 6 10 and for CIPM4 in Table 6 11, with pound solid production shown graphically in Figures 6 5 and 6 6. Table 6 12 shows the estimated harvest dates for both models. Table 6 13 and Figures 6 3 and 6 4 show weekly NFC and FCOJ volumes in storage. Mo deling results from CIPM3 are compared to results from CIPM1 to simulate operational changes when weekly processing volumes are allowed to increase to the seasonal peak level, assuming all harvest is done by hand. While operation at a single weekly peak v olume is infeasible under current operational conditions, modeling this theoretical optimum estimates a baseline with which to compare the impact of the constraints prohibiting operation at this optimum. Comparisons between t hese two models are shown to d escribe how the inclusion of
185 weekly processing levels in CIPM1 constrains industry operation and returns from the theoretical optimum. CIPM3 Results The constrained industry production model with a weekly box maximum and 95% hand harvesting (CIPM3) estim ates total industry returns of $1,897,685,870. Estimated returns are $120,048,665 (6.8%) greater than estimated returns from CIPM1. Grower returns are estimated at $1,495,484,295 (78.8% of industry returns), an increase of $157,154,671 (11.7%) over CIPM1 returns. Processor returns are estimated at $402,201,575 (21.2% of industry returns), a decrease of $37,106,000 ( 8.4%) compa red to returns estimated by CIPM 1. The NFC production season begins in Week 21 and runs continuously through Week 49, approximat ely equivalent to the third week of December through the beginning of July. The FCOJ production season is also continuous but begins and ends earlier, with production beginning Week 15 and ending Week 46, about the second week of November through the thir d week of June. From starting date to ending date, the harvesting and processing schedule resulting from CIPM3 is ten weeks shorter for NFC production and six weeks shorter for FCOJ production, as compared to the estimates from CIPM1. Total box yield is estimated at 186,253,800 boxes, an increase of 20,253,324 boxes (12.2%) over CIPM1 estimates. This box increase is allowed by increasing the weekly processing capacity to the maximum capacity utilized during any single week of the 2007 08 season. All be aring acreage is harvested and processed; including acreage containing early fruit in the southern region that was not fully harvested in CIPM1. Additionally, more acres are harvested closer to their biological optimal, capturing higher box per acre yield s with CIPM3. Oranges going to NFC production account for 87,244,800 boxes (46.8% of all boxes), an increase of 347,958 boxes (0.4%) over estimated boxes to NFC production in CIPM1. Oranges
186 going to FCOJ production account for 99,006,600 boxes (53.2% of all boxes), an increase of 19,903,856 boxes (25.2%) compared to CIPM1. Total pound solid production is estimated at 1,242,223,700 pound solids, an increase of 131,436,657 pound solids (11.8%) over CIPM1. NFC production accounts for 574,636,000 pound soli ds (46.3% of all PS) and FCOJ accounts for 667,587,700 pound solids (53.7% of all PS). These estimates show virtually no change in NFC pound solid production, increasing 782,050 pound solids (0.1%), with almost all pound solids gained in the form of FCOJ, increasing in production by 130,654,607 pound solids (24.3%). Most of the additional pound solids captured in CIPM3 are processed into FCOJ because of the slack constraints in production and storage capacity. Similar production practices are shown in Fi gures 2 10 and 2 13. Fluctuations in juice yield cause FCOJ production to vary while NFC production remains fairly consistent. The volume of NFC in storage fluctuated from a low of 69,306,023 pound solids in Week 20 to a high of 345,534,023 pound solids during weeks 46 through 49, filling all available storage capacity. FCOJ in storage changed from a low of 402,747,049 pound solids (96,697,971 gallons at 42 degree Brix) during Week 15 to a high of 765,417,049 pound solids (183,773,601 gallons at 42 degr ee Brix) during Week 46, below the maximum estimated FCOJ capacity of 831,200,000 pound solids (200,000,000 gallons at 42 degree Brix). Shadow values for the marginal box of fruit allowed to be processed estimate the additional returns to the industry fr om harvesting an extra box of fruit during weeks when the processing upper bound constraint binds. Shadow values are zero for all weeks in which processing volumes are below the maximum upper bound. A shadow value of $ 0 .46 is estimated in Week 21, increa sing to $ 0 .69 in Week 26, and then decreasing weekly with the last non zero value of $ 0 .13 in Week 45. Shadow values are much lower than estimates for models using the
187 2007 08 processing schedule because lessening the processing constraint in CIPM3 only c hanges the week of harvest for a box of fruit to a more optimal week. In CIPM1 and CIPM2, not all fruit was harvested due to the scheduling constraint binding at each week. Allowing an additional box to be harvested and processed would then cause much la rger increases in returns. Because all acreage is being harvested in CIPM3, expanding the processing constraint by one box results in a slightly more optimal harvest of a box fruit and relatively small increases in returns. CIPM4 Results CIPM 4 represent s the theoretical optimum harvest using primarily mechanical harvesting. Comparisons of results between CIPM4 and CIPM3 represent the possible changes to optimal operation when switching from 5% mechanical harvesting to 95% mechanical harvesting. Compari ng these models is intended to estimate changes from mechanical harvesting, assuming adoption allows for consistent harvest volumes near the weekly maximum capacity. The constrained industry production model with a weekly box maximum and 95% mechanical ha rvesting (CIPM4) estimates total industry returns of $1,981,335,515. Comparing estimated results of CIPM4 to results from CIPM3, total industry returns increased by $83,649,645 (4.4%). Estimated grower returns are $ 1,564,963,183 (79.0% of all returns), a n increase of $69,478,888 (4.6%) over CIPM3. Estimated processor returns are $ 416,372,333 (21.0% of all returns), an increase of $14,170,758 (3.5%) over CIPM3. Growers received a larger percentage increase in returns than processors, but processors were shown to benefit from the adoption of mechanical harvesting. All acreage is harvested and NFC production runs from Week 18 through Week 52, approximately the first week of December through the last week of July, pausing during weeks 19, 20, and 48. The 35 week NFC processing schedule is six weeks longer than the NFC
188 schedule from CIPM3. FCOJ production runs continuously for 29 weeks from Week 16 through Week 45, or about the third week of November through the second week of June. This schedule is three weeks shorter than the 32 weeks required to process FCOJ in CIPM3. Changes in processing schedule length are likely due to changes in the production mix of NFC and FCOJ. While total box yield is relatively unchanged at 186,090,700 boxes, a decrease of 1 63,100 boxes ( 0.1%), oranges going to NFC production account for 92,972,800 boxes (50.0% of all boxes), an increase of 5,728,000 boxes (6.6%), and oranges going to FCOJ production account for 93,115,900 boxes (50.0% of all boxes), a decrease of 5,890,700 boxes ( 5.9%), as compared to CIPM3. This change in product mix, and shift to producing more NFC, is in part responsible for the incr ease in returns compared to CIPM 3. This shift in production to more of the higher valued NFC product could have been allo wed due to the lower cost of harvest offsetting the higher cost of NFC storage. Tota l pound solid production for CIP M4 is estimated at 1,232,501,000 pound solids, 9,722,700 pound solids ( 0.8%) less than production estimates for CIPM3. The pound solids g oing to NFC increased by 30,870,000 pound solids (5.4%) to 605,506,000 pound solids (49.1% of all PS) and FCOJ production decreased by 40,592,700 pound solids ( 6.1%) to 626,995,000 (50.9% of all PS). During the year, the weekly volume of NFC in storage f luctuates from a low of 83,106,023 pound solids during Week 20 to a high of 345,534,023 for Week s 45 through 52. Weekly FCOJ volumes in storage change from a low of 394,147,049 pound solids (94,633,145 gallons at 42 degree Brix) during Week 15 to a high o f 734,387,049 pound solids (176,323,421 gallons at 42 degree Brix) during Week 45. Similar to CIPM3, NFC volumes in storage were at full capacity for several consecutive weeks, while FCOJ volumes never reached the full storage capacity.
189 Similar to shado w values estimated for CIPM3, the first non zero shadow value for additional processing capacity was $ 0 .29, estimated in Week 21. This value can be interpreted as the increase in returns by allowing an additional box of fruit to be harvested and processed Shadow values increase to $ 0 .52 in Week 31 and then decrease to $ 0 .07 in Week 44. These potential increases to returns from expanding capacity are relatively small, as compared to CIPM1 and CIPM2, because all available fruit is being harvested in CIPM4 An expansion in the scheduling constraint allows fruit to be harvested during a more optimal week but does not allow more fruit to be harvested, yielding small increases to industry returns. Finally, comparing processing operations and results for CIPM1 and CIPM4 estimates total changes to the industry from adoption of mechanical harvesters when adoption allows for processing at a single weekly volume. This comparison is practically relevant if this change eliminates some aspects of hand harvesting that are limiting weekly processing volumes. Eliminating these constraints and allowing harvest at a lesser unit cost could shift industry operation away from the assumptions of CIPM1 and allow operation under constraints similar to CIPM4. Comparing the resu lts of these two models estimates the greatest change in operation and the largest possible gains from mechanical harvester adoption. Total industry returns for CIPM4 are $203,698,310 (11.5%) greater than returns from CIPM1. These increases are coming fr om the costs savings of mechanical harvesting and the increase in fruit and juice production allowed by the relaxing the constraint on processing capacity. Estimated grower returns are $226,633,559 (16.9%) greater than with CIPM1 and e stimated processor returns decrease $22,935,248 ( 5.2%) in CIPM4, predicti ng that all gains accrue to grower s, while processors absorb some additional costs. These results predict that while gains could accrue to the industry from the adoption of mechanical harvesting, proc essors will see only losses under the current
190 payment system. Processors are therefore not likely to support the move to adoption without financial compensation. Start and stop dates for NFC processing schedules estimate CIPM1 starting earlier with a few inconsistent weeks of processing and CIPM4 running a few longer. FCOJ production f or CIPM4 is estimated to begin six weeks later and end three weeks earlier tha n CIPM1, shorter by nine weeks. Total box yield increases by 20,090,224 (12.1%),with oranges going to NFC increasing by 6,075,958 boxes (7.0%), and oranges going to FCOJ production increasing by 14,013,156 (17.7%), as compared to CIPM1. Increases in box yields in CIPM4 are caused by two factors: the complete harvesting of all acreage and the cap turing of additional weight by harvesting more acreage during periods of peak yields. Total pound solid production increases by 121,713,957 (11.0%) pound solids for CIPM4, with 31,652,050 (5.5%) additional pound solids going to NFC and 90,061,907 (16.8%) additional pound solids going to FCOJ. Again the majority of the increase occurs in FCOJ production because of the available processing and storage capacity. The wee kly volume of NFC in storage is at capacity for one week with CIPM1 and for several cons ecutive weeks with CIPM4, while FCOJ volumes never reached the full storage capacity for either model. One of the most striking contrasts from estimates of CIPM1 and CIPM4 are the 20 million additional boxes and the122 million additional pound solid captur ed in CIPM4, shown weekly in Figures 6 11 and 6 12. All of this variation can be see occurring during the middle of the harvest season, with operation at the beginning and end of the season being equal in both models. Production in the tails of both mode ls are equal at 240 million pound solids, while weekly pound solid volumes captured before Week 22 and after Week 44 vary weekly. The additional 122 million pound solids captured in CIPM4 occur from Week 22 to Week 44, with over half of the
191 additional pro duct captured (65 million PS) from Week 29 to Week 34. These are the weeks during which processors typically reduce input volumes as they transition from processing early fruit varieties to late varieties. Maturity models estimated in Chapter 4 show thi s time period to be near optimal for harvest of many trees throughout Florida. CIPM 4 therefore continues harvesting and processing near full industry capacity during these weeks, resulting in much grea ter yields. Additionally, CIPM 4 harvests more boxes a nd captures more pound solids just before the typical early season peak and a few weeks after the typical late season peak. Summary Grower and processor decision models are combined into a single industry decision model. This linear programming model representing the industry is estimated to simulate operation of a single, vertically integrated juice production firm operating under constraints designed to mimic current production and market conditions. Constraints are then adjusted to estimate operati on using mechanical harvesting. Industry simulation estimates show increases in total returns when operating with mechanical harvesting. The use of mechanical harvesting does not, however, necessarily increase returns for all players. Constraints are fu rther lessened to estimate optimal production based only on estimated levels of fruit availability, processing capacity, and storage capacity, omitting historical processing trends. Modeling results and their implications for the Florida orange juice indu stry are discussed further in the following chapter.
192 Figure 6 1 CIPM1 weekly pound solid production for NFC and FCOJ Figure 6 2. CIPM2 weekly pound solid production for NFC and FCOJ.
193 Figure 6 3 NFC weekly storage volume estimates and weekly av erage from 2000 2006, reported in pound solids. Figure 6 4. FCOJ weekly storage volume estimates and weekly average from 2000 2006, reported in gallons at 42.0 degree Brix.
194 Figure 6 5 CIPM3 weekly pound solid production for NFC and FCOJ Fig ure 6 6. CIPM4 weekly pound solid production for NFC and FCOJ.
195 Figure 6 7 CIPM1 and CIPM3 weekly total b ox production Figure 6 8. CIPM1 and CIPM3 weekly total pound solid production.
196 Figure 6 9 CIPM2 and CIPM4 weekly total b ox production Fi gure 6 10. CIPM2 and CIPM4 weekly total pound solid production.
197 Figure 6 11 CIPM1 and CIPM4 weekly total b ox production Figure 6 12. CIPM1 and CIPM4 weekly total pound solid production.
198 Table 6 1. Definition of variables for grower, processor a nd industry models Variable Definition IND Industry returns G Grower returns P Processor returns V Fruit variety I Geographical growing region A Tree age T Time period in weeks t Optimal harvest week A Acres X Harvest variable TSS Total soluble solids RATIO Brix to acid ratio PPT Average pieces of fruit per tree DROP Cumulative fruit drop percentage FTWT Average fruit weight per piece DEN Average tree planting density in trees per acre DIP Delivered in price of fruit PS Pound solids per box B Boxes per acre LB Lower Boun d quantity of boxes to be delivered for processing UB Upper Bound quantity of boxes to be delivered for processing UPC Grove production cost per acre UHC HH Hand harvest cost per box UHC MH Mechanical harvest cost per box Minimum percentage of fruit hand harvested PM N Processing margin for NFC PM C Processing margin for FCOJ TB N Total boxes processed to NFC UTB N Upper bound quantity of boxes processed to NFC TB C Total boxes processed to FCOJ UTB C Upper bound quantit y of boxes processed to FCOJ UEC N Cost of extraction per box for NFC UEC C Cost of extraction per box for FCOJ USC N Storage cost per pound solid for NFC USC C Storage cost per pound solid for FCOJ PSS N Volume of NFC in storage UPSS N Upper bound volume of NFC in storage PSS C Volume of FCOJ in storage UPSS C Upper bound volume of FCOJ in storage OUT N NFC output removed from the processing plant weekly OUT C FCOJ output removed from the processing plant weekly LTSS t Lower bound average total soluble solids NFC% Percentage of fruit processed to NFC LRATIO t Lower bound Brix to acid ratio
199 Table 6 1. Continued Variable Definition URATIO t Upper bound Brix to acid ratio LAVGRAT Lower bound average Brix to acid ratio UAVGRAT Upper bound average B rix to acid ratio
200 Table 6 2. Price and cost estimates for fruit inputs, processing, and storages NFC FCOJ Unit Harvest Cost (a) hand (pick, roadside ) $2.0 0 per box $2.0 0 per box mechanical (pick, roadside ) hand cost $0.50 per box ha nd cost $0.50 per box Quality Standards (b) minimum soluble solids 9% 9% minimum B/A ratio 12/1 12/1 maximum B/A ratio 24/1 24/1 minimum average B/A ratio 12.5/1 12.5/1 maximum average B/A ratio 19.5/1 19.5/1 Delivered in fruit price (DIP) (c) $1.50 per PS $1.50 per PS Average FOB juice price (d) Low DIP + $0.30 per PS $0.85 per PS Average DIP + $0.30 per PS $1.20 per PS High DIP + $0.30 per PS $2.07 per PS Average Processing Margin (c,d,e) jui ce margin $0.35 per PS $0.15 per PS byproducts margin $0.05 per PS $0.05 per PS Unit Processing Cost (e,f,g) Extraction $0.02per PS $0.02per PS Pasteurization $0.02 per PS Evaporation $0.03per PS Unit Storage Price from ICE (h) fixed fee $0.05 per PS $0.04 per PS weekly fee $0.03 per PS $0.005 per PS Unit Storage Cost (f) weekly cost $0.0025 per PS $0.0004 per PS additional capacity $2.00 per PS NA Extraction capacity (c) 2008 weekly capacit y 3,471,700 boxes 3,169,800 boxes 2003 08 max weekly capacity 3,471,700 boxes 6,207,400 boxes Storage capacity (c) total capacity in PS 345 533 680 PS 831,200,000 PS Total capacity in gallons 313,295,566 at 11.8 degree Brix 200,000,000 at 42.0 degree Brix Avg. on hand Aug 1st (c) 275,106,023 PS 537,547,049 PS Weekly product output (c) average weekly output 10,290,000 PS per week 9,559,000 PS per week the Mechanical Harvesting of Florida Citrus Extension Program at the Highlands County Extension Service, February 4, 2004. (b) Florida Statutes. 2006. Title 949." (c) Florida Citrus Commission. 2009. Florida Citrus Processors Association 2007 2008 Season Processor Reports Statistical Survey. (d) Florida Citrus Mutual, Statistical Division, 1998 2008 (e) Personal communication with anonymous manager (1) of a Florida orange processing facility, July 2008 (f) Personal communication with anonymous manager (2) of a Florida orange processing facility, July 2008 (g) Personal communication with anonymous manager (10) of a Florida orange processing facility, July 2008 (h) Intercontinental Exchange. FCOJ A Futures. (cited 28 June 2008). Available from http://www.theice.com.
201 Table 6 3. Bearing orange tree acreage in Florida for 2004 and estimated for 2008 VarRegAge Aggregate number 2004 acreage 2008 acreage P ercent of all acreage V10I1G1 1 1,884 1,553 0.33% V10I1G2 2 1,703 1,404 0.30% V10I1G3 3 4,159 3,429 0.74% V10I1G4 4 7,075 5,833 1.26% V10I1G5 5 4,973 4,100 0.88% V10I2G1 6 654 539 0.12% V10I2G2 7 1,139 939 0.20% V10I2G3 8 8,055 6,641 1.43% V10I2G4 9 10,783 8,891 1.92% V10I2G5 10 1,087 896 0.19% V10I3G1 11 4,557 3,757 0.81% V10I3G2 12 3,731 3,076 0.66% V10I3G3 13 19,798 16,324 3.52% V10I3G4 14 22,612 18,644 4.02% V10I3G5 15 8,277 6,824 1.47% V10I4G1 16 5,005 4,127 0.89% V10I4G2 17 4,160 3,43 0 0.74% V10I4G3 18 13,612 11,223 2.42% V10I4G4 19 28,911 23,837 5.14% V10I4G5 20 13,495 11,127 2.40% V10I5G1 21 5,923 4,884 1.05% V10I5G2 22 4,541 3,744 0.81% V10I5G3 23 14,715 12,133 2.61% V10I5G4 24 24,250 19,994 4.31% V10I5G5 25 4,798 3,956 0.85 % V20I1G1 26 881 726 0.16% V20I1G2 27 717 591 0.13% V20I1G3 28 1,965 1,620 0.35% V20I1G4 29 2,095 1,727 0.37% V20I1G5 30 3,456 2,849 0.61% V20I2G1 31 61 50 0.01% V20I2G2 32 31 26 0.01% V20I2G3 33 98 81 0.02% V20I2G4 34 185 153 0.03% V20I2G5 35 22 3 184 0.04% V20I3G1 36 791 652 0.14% V20I3G2 37 352 290 0.06% V20I3G3 38 2,034 1,677 0.36% V20I3G4 39 2,234 1,842 0.40% V20I3G5 40 3,319 2,737 0.59% V20I4G1 41 785 647 0.14% V20I4G2 42 849 700 0.15% V20I4G3 43 1,446 1,192 0.26% V20I4G4 44 2,223 1, 833 0.40% V20I4G5 45 6,533 5,386 1.16% V20I5G1 46 1,976 1,629 0.35% V20I5G2 47 1,317 1,086 0.23% V20I5G3 48 4,643 3,828 0.83% V20I5G4 49 4,860 4,007 0.86% V20I5G5 50 2,365 1,950 0.42% V30I1G1 51 3,741 3,084 0.66% V30I1G2 52 2,879 2,374 0.51%
2 02 Table 6 3. Continued VarRegAge Aaggregate number 2004 acreage 2008 acreage Percent of all acreage V30I1G3 5 3 8,216 6,774 1.46% V30I1G4 54 11,389 9,390 2.02% V30I1G5 55 7,784 6,418 1.38% V30I2G1 56 514 424 0.09% V30I2G2 57 856 706 0.15% V30I2G3 58 1,952 1 ,609 0.35% V30I2G4 59 2,404 1,982 0.43% V30I2G5 60 771 636 0.14% V30I3G1 61 10,103 8,330 1.80% V30I3G2 62 8,134 6,707 1.45% V30I3G3 63 27,808 22,928 4.94% V30I3G4 64 24,411 20,127 4.34% V30I3G5 65 14,676 12,100 2.61% V30I4G1 66 9,705 8,002 1.72% V 30I4G2 67 6,564 5,412 1.17% V30I4G3 68 15,643 12,898 2.78% V30I4G4 69 18,101 14,924 3.22% V30I4G5 70 12,795 10,550 2.27% V30I5G1 71 10,569 8,714 1.88% V30I5G2 72 11,004 9,073 1.96% V30I5G3 73 40,659 33,524 7.23% V30I5G4 74 38,207 31,502 6.79% V30I5 G5 75 8,534 7,036 1.52% State total 562,755 463,994 season fruit (V20), late fruit (V30), Indian River (I1), north (I2), central (I3), west (I4), south(I5), age 3 5 (G1), age 6 8 (G2), age 9 13 (G3), age 14 23 (G4), and age 24+ (G5). U.S. Department of Agriculture. 2005. Commercial Citrus Inventory 2004. NASS, Florida Agricultural Statistics Service, April. U.S. Department of Agriculture. 2009. Commercial Citrus Inv entory 2008. NASS, Florida Agricultural Statistics Service, April
203 Table 6 4. Total boxes of oranges processed in Florida during the 2007 08 season. Week ending date Week Boxes 8/4/2007 1 0 8/11/2007 2 0 8/18/2007 3 0 8/25/2007 4 0 9/1/2007 5 0 9/8/2007 6 0 9/15/2007 7 0 9/22/2007 8 0 9/29/2007 9 0 10/6/2007 10 16449 10/13/2007 11 29088 10/20/2007 12 83253 10/27/2007 13 78207 11/3/2007 14 80637 11/10/2007 15 92122 11/17/2007 16 759929 11/24/2007 17 816244 12/1/2007 18 3062081 12/8/ 2007 19 4785841 12/15/2007 20 5494251 12/22/2007 21 5560460 12/29/2007 22 3993938 1/5/2008 23 5709819 1/12/2008 24 6219614 1/19/2008 25 6032620 1/26/2008 26 6227574 2/2/2008 27 5691585 2/9/2008 28 6118708 2/16/2008 29 5473572 2/23/2008 30 539130 7 3/1/2008 31 4474450 3/8/2008 32 4204067 3/15/2008 33 4289728 3/22/2008 34 4309590 3/29/2008 35 5922359 4/5/2008 36 5968391 4/12/2008 37 6324150 4/19/2008 38 6186566 4/26/2008 39 6288387 5/3/2008 40 5882095 5/10/2008 41 6209194 5/17/2008 42 61 51071 5/24/2008 43 5414020 5/31/2008 44 5134056 6/7/2008 45 5181433 6/14/2008 46 4219628 6/21/2008 47 3616428 6/28/2008 48 3012016
204 Table 6 4. Continued Week ending date Week Boxes 7/5/2008 49 1022447 7/12/2008 50 473048 7/19/2008 51 0 7/26/2008 52 0 Florida Citrus Commission. 2009. Florida Citrus Processors Association 2007 2008 Season Processor Reports Statistical Survey. Table 6 5. Summary results for all constrained industry production models. Model CIPM1 CIPM2 CIPM3 CIPM4 Total retur ns $1,777,637,205 $1,851,519,436 $1,897,685,870 $1,981,335,515 Grower returns $1,338,329,624 $1,413,519,334 $1,495,484,295 $1,564,963,183 Processor returns $439,307,581 $438,000,102 $402,201,575 $416,372,333 Total boxes 165,999,586 165,999,536 186,253,8 00 186,090,700 Boxes of NFC 86,896,842 87,196,682 87,244,800 92,972,800 Boxes of FCOJ 79,102,744 78,802,854 99,006,600 93,115,900 Total pound solids 1,110,787,043 1,111,113,373 1,242,223,700 1,232,501,000 Pound solids NFC 573,853,950 573,856,370 574,63 6,000 605,506,000 Pound solids FCOJ 536,933,093 537,257,003 667,587,700 626,995,000 Average degree Brix 12.83 12.80 12.83 12.81 Maximum degree Brix 16.06 16.06 14.62 14.50 Minimum degree Brix 11.09 11.07 10.57 10.57 Average Brix/Acid r atio 15.93 15.91 16.22 16.28 Maximum ratio 19.89 19.89 18.73 19.77 Minimum ratio 13.38 13.48 11.73 11.73 Acres harvested 428,793 426,047 463 994 463 994 Percent of acreage harvested 92.4% 91.8% 100% 100% Maximum weekly boxes mechanically harvested 316,210 6,007,990 316,210 6,007,990 Sets of mechanical harvesters required, 9 172 9 172 Note: assume one set of mechanical harvesters can harvest an average of 35,000 boxes per wee k.
205 Table 6 6. CIPM1, Processing results for constrained industry production model with 9 5% hand harvesting Week Total boxes Total box marginal value Total boxes of NFC Total boxes of FCOJ Pound solids of NFC Pound solids of FCOJ 1 0 4.24 0 0 0 0 2 0 4.52 0 0 0 0 3 0 4.79 0 0 0 0 4 0 5.05 0 0 0 0 5 0 5.30 0 0 0 0 6 0 5.59 0 0 0 0 7 0 5 .86 0 0 0 0 8 0 6.11 0 0 0 0 9 0 6.36 0 0 0 0 10 16,449 6.59 0 16,449 0 79,473 11 29,088 6.84 0 29,088 0 148,050 12 83,253 6.93 83,253 0 467,140 0 13 78,207 7.17 78,207 0 459,740 0 14 80,637 7.35 0 80,637 0 369,770 15 92,122 7.47 92,122 0 485,670 0 16 759,930 7.73 0 759,930 0 3,795,700 17 816,240 7.84 0 816,240 0 3,916,300 18 3,062,100 7.89 0 3,062,100 0 12,936,000 19 4,785,800 8.09 1,616,000 3,169,800 9,027,900 18,955,000 20 5,494, 2 00 8.22 2,324,400 3,169,800 13,280,000 19,177,000 21 5,560,4 0 0 8.34 2,390,600 3,169,800 13,996,000 19,490,000 22 3,993,91 0 8.44 824,110 3,169,800 4,880,500 19,999,000 23 5,709,800 8.54 2,540,000 3,169,800 15,444,000 19,657,000 24 6,219,600 8.63 3,049,800 3,169,800 18,712,000 20,808,000 25 6,032,600 8.72 2,862,80 0 3,169,800 17,855,000 20,556,000 26 6,227,600 8.81 3,057,800 3,169,800 19,143,000 21,308,000 27 5,691,600 8.89 2,521,800 3,169,800 15,723,000 21,884,000 28 6,118,700 9.02 3,471,700 2,647,000 21,835,000 17,656,000 29 5,473,5 00 9.04 2,303,700 3,169,800 14,598,000 22,214,000 30 5,391,300 9.10 2,221,500 3,169,800 14,118,000 22,532,000 31 4,474,400 9.18 3,471,700 1,002,700 22,579,000 6,958,000 32 4,204,0 00 9.20 1,034,200 3,169,800 6,688,500 23,336,000 33 4,289,700 9.23 1,119,900 3,169,800 7,236,700 22,6 40,000 34 4,309,600 9.27 2,193,700 2,115,900 14,219,000 14,990,000 35 5,922,3 00 9.35 3,471,700 2,450,600 22,960,000 17,568,000 36 5,968,400 9.45 3,471,700 2,496,700 22,992,000 18,385,000 37 6,324,1 00 9.50 3,471,700 2,852,400 23,201,000 20,802,000 38 6 ,186,5 00 9.57 3,471,700 2,714,800 23,724,000 20,253,000 39 6,288,400 9.58 3,471,700 2,816,700 24,114,000 20,218,000 40 5,882,100 9.68 3,471,700 2,410,400 24,357,000 18,237,000 41 6,209,200 9.73 3,471,700 2,737,500 24,763,000 20,956,000 42 6,151,0 00 9.7 4 3,471,700 2,679,300 24,772,000 20,171,000 43 5,414,000 9.77 3,471,700 1,942,300 24,929,000 14,786,000 44 5,134,0 00 9.74 3,471,700 1,662,300 24,652,000 12,174,000 45 5,181,400 9.81 3,471,700 1,709,700 24,425,000 13,209,000
206 Table 6 6. Continued Week T otal boxes Total box marginal value Total boxes of NFC Total boxes of FCOJ Pound solids of NFC Pound solids of FCOJ 46 4,219,600 9.78 3,471,700 747,900 24,367,000 5,676,100 47 3,616,400 9.76 3,471,700 144,700 23,976,000 1,092,700 48 3,012,000 9.92 3,012 ,000 0 20,365,000 0 49 1,022,400 10.84 1,022,400 0 6,633,100 0 50 473,050 10.71 473,050 0 2,875,700 0 51 0 10.70 0 0 0 0 52 0 10.69 0 0 0 0 Sum 165,999 586 86,896,842 79,102,744 573,853,950 536,933,093
207 Table 6 7. CIPM2, Processing results for con strained industry production model with 95% mechanical harvesting. Week Total b oxes Total box marginal v alue Total b oxes of NFC Total b oxes of FCOJ Pound solid s of NFC Pound solid s of FCOJ 1 0 4.69 0 0 0 0 2 0 4.97 0 0 0 0 3 0 5.24 0 0 0 0 4 0 5.50 0 0 0 0 5 0 5.75 0 0 0 0 6 0 6.04 0 0 0 0 7 0 6.31 0 0 0 0 8 0 6.56 0 0 0 0 9 0 6.81 0 0 0 0 10 16,449 7.04 0 16,449 0 79,473 11 29,088 7.29 0 29,088 0 148,050 12 83,253 7.48 0 83,253 0 412,650 13 78,207 7.68 0 78,207 0 388,860 14 80,637 7.80 0 80,6 37 0 369,770 15 92,122 7.99 92,122 0 485,670 0 16 759,930 8.18 0 759,930 0 3,795,700 17 816,240 8.29 0 816,240 0 3,916,300 18 3,062,100 8.34 0 3,062,100 0 12,936,000 19 4,785,800 8.61 1,616,000 3,169,800 9,027,900 18,955,000 20 5,494,2 00 8.75 2,324,4 00 3,169,800 13,280,000 19,177,000 21 5,560,4 00 8.87 2,390,600 3,169,800 13,996,000 19,490,000 22 3,993,91 0 8.96 824,110 3,169,800 4,880,500 19,999,000 23 5,709,800 9.06 3,395,900 2,313,900 20,649,000 14,349,000 24 6,219,600 9.16 3,049,800 3,169,800 18 ,712,000 20,808,000 25 6,032,600 9.24 3,471,700 2,560,900 21,653,000 16,607,000 26 6,227,600 9.34 3,057,800 3,169,800 19,143,000 21,308,000 27 5,691,600 9.42 2,521,800 3,169,800 15,723,000 21,884,000 28 6,118,700 9.47 3,471,700 2,647,000 21,835,000 17, 656,000 29 5,473,5 00 9.57 3,471,700 2,001,800 22,000,000 14,029,000 30 5,391,300 9.63 2,221,500 3,169,800 14,118,000 22,532,000 31 4,474,400 9.63 3,471,700 1,002,700 22,579,000 6,958,000 32 4,204,0 00 9.73 1,034,200 3,169,800 6,688,500 23,336,000 33 4, 289,7 00 9.76 1,119,900 3,169,800 7,236,700 22,640,000 34 4,309,56 0 9.72 3,471,700 837,860 22,504,000 5,935,500 35 5,922,3 00 9.80 3,471,700 2,450,600 22,960,000 17,568,000 36 5,968,400 9.90 3,471,700 2,496,700 22,992,000 18,385,000 37 6,324,1 00 9.95 3,4 71,700 2,852,400 23,201,000 20,802,000 38 6,186,5 00 10.02 3,471,700 2,714,800 23,724,000 20,253,000 39 6,288,400 10.03 3,471,700 2,816,700 24,114,000 20,218,000 40 5,882,100 10.13 3,471,700 2,410,400 24,357,000 18,237,000 41 6,209,200 10.18 3,471,700 2 ,737,500 24,763,000 20,956,000 42 6,151,0 00 10.19 3,471,700 2,679,300 24,772,000 20,171,000 43 5,414,000 10.22 3,471,700 1,942,300 24,929,000 14,786,000 44 5,134,0 00 10.19 3,471,700 1,662,300 24,652,000 12,174,000 45 5,181,400 10.26 3,471,700 1,709,700 24,425,000 13,209,000
208 Table 6 7. Continued Week Total b oxes Total box marginal v alue Total b oxes of NFC Total b oxes of FCOJ Pound solid s of NFC Pound solid s of FCOJ 46 4,219,600 10.23 3,471,700 747,900 24,367,000 5,676,100 47 3,616,400 10.21 446,600 3 ,169,800 3,084,300 23,937,000 48 3,011,990 10.19 2,587,600 424,390 17,496,000 3,174,600 49 1,022,400 11.29 1,022,400 0 6,633,100 0 50 473,050 11.16 473,050 0 2,875,700 0 51 0 11.15 0 0 69,816 0 52 0 11.14 0 0 11,357 0 Sum 165,999,53 6 87,196,682 78,8 02,854 573,937,543 537,257,003
209 Table 6 8 Harvesting results for CIPM1 and CIPM2, constrained industry production model s with 95% hand harvesting and 95% mechanical harvesting with h arve st date in weeks from August 1 and portion harvested weekly in pa renthesis Grove aggregate Optimal Week Harvest Week, with 95% hand harvesting Weeks, with 95% mechanical harvesting 1 21 33 33 2 21 33 33 3 21 33 33 4 21 33 33 5 21 33 33 6 25 23 23 7 25 23(.14) 24(.86) 24 8 25 24(.08) 25(.92) 23(.0 2) 24(.06) 25(.92) 9 25 24 24 10 25 24 24 11 25 10(.02) 22(.98) 23 12 25 23 19 13 25 20(.80) 22(.07) 23(.13) 19(.42) 22(.58) 14 25 12(.01) 13(.01) 14(.01) 16(.06) 17(.07) 18(.25) 19(.13) 21(.46) 12(.01) 13(.01) 14(.01) 15(.01) 16(.06) 17(.07) 18(.25) 19(.09) 21(.46) 23(.04) 15 25 15(.02) 19(.52) 22(.46) 20(.91) 23(.09) 16 29 32 29 17 29 32 32 18 29 29(.79) 31(.22) 29(.54) 31(.46) 19 29 26(.35) 27(.39) 31(.24) 32(.01) 26(.35) 28(.06) 30(.38) 31(.16) 32(.05) 20 29 28(.10) 29(.22) 30(.68 ) 27(.70) 29(.30) 21 22 25 22 22 26(.90) 23 22 25(.18) 25(.56) 26(.19) 24 22 25 22 25 26 24 34 34 27 24 34 34 28 24 34 34 29 24 34 34 30 24 34 34 31 21 26 26 32 21 26 26 33 21 26 26 34 21 26 26 35 21 26 26 36 23 23 23 37 23 23 23 38 23 23 23 39 23 23 23 40 23 23 23 41 29 28 28 42 29 28 28 43 29 28 28 44 29 28 28
210 Table 6 8. Continued Grove aggregate Optimal Week Harvest Week, with 95% hand harvesting Weeks, with 95% mechanical harvesting 45 29 28 28 46 28 32 32 47 28 32 32 48 28 34 34 49 28 32(.49) 33(.46) 34(.05) 32(.49) 33(.46) 34(.05) 50 28 32 32 51 35 48 48 52 35 47 48 53 35 47(.21) 48(.79) 47 54 35 47 48 55 35 48 47(.81) 48(.19) 56 38 50 49 57 38 50 4 9 58 38 49 50 59 38 49(.86) 50(.14) 49 60 38 48(.39) 50(.61) 48(.39) 49(.42) 50(.19) 61 36 43 41 62 36 42 42 63 36 40(.88) 43(.12) 40(.88) 42(.12) 64 36 39(.33) 41(.67) 42(.47) 43(.53) 65 36 41(.07) 42(.59) 43(.34) 39(.35) 41(.65) 66 34 45 45(.35) 47(.65) 67 34 43(.22) 47(.78) 45 68 34 45 45 69 34 44(.94) 46(.06) 44(.94) 47(.06) 70 34 43(.13) 45(.12) 46(.75) 43(.18) 46(.81) 71 36 36 37 72 36 36 34(.30) 36(.70) 73 36 34(.11) 35(.59) 36(.29) 36(.46) 37(.54) 74 36 34(.01) 37(.52) 38(.19) 39(.28) 34(.05) 35(.48) 38(.19) 39(.28) 75 36 38 38
211 Table 6 9. Estimated weekly volume in storage for CIPM1 and CIPM2, NFC reported as pound solids and FCOJ reported as pound solids and gallons at 42 degree Brix. Week CIPM1 NFC PS CIPM1 FCO J PS CIPM1 FCOJ gal at 42 degree Brix CIPM2 NFC PS CIPM2 FCOJ PS CIPM2 FCOJ gal at 42 degree Brix 1 264,816,023 527,988,049 126,767,839 264,816,023 527,988,049 126,767,839 2 254,526,023 518,427,049 124,472,281 254,526,023 518,427,049 124,472,281 3 244,2 36,023 508,867,049 122,176,963 244,236,023 508,867,049 122,176,963 4 233,946,023 499,307,049 119,881,644 233,946,023 499,307,049 119,881,644 5 223,656,023 489,757,049 117,588,727 223,656,023 489,757,049 117,588,727 6 213,366,023 480,197,049 115,293,409 213,366,023 480,197,049 115,293,409 7 203,076,023 470,637,049 112,998,091 203,076,023 470,637,049 112,998,091 8 192,786,023 461,077,049 110,702,773 192,786,023 461,077,049 110,702,773 9 182,496,023 451,517,049 108,407,455 182,496,023 451,517,049 108,407 ,455 10 172,206,023 442,037,049 106,131,344 172,206,023 442,037,049 106,131,344 11 161,906,023 432,647,049 103,876,842 161,906,023 432,647,049 103,876,842 12 152,106,023 423,047,049 101,571,921 151,606,023 423,447,049 101,667,959 13 142,306,023 413,547 ,049 99,291,008 141,306,023 414,347,049 99,483,085 14 132,006,023 404,347,049 97,082,125 131,006,023 405,147,049 97,274,201 15 122,206,023 394,747,049 94,777,203 121,206,023 395,547,049 94,969,279 16 111,906,023 388,947,049 93,384,646 110,906,023 389,84 7,049 93,600,732 17 101,606,023 383,347,049 92,040,108 100,706,023 384,147,049 92,232,185 18 91,306,023 386,747,049 92,856,434 90,406,023 387,547,049 93,048,511 19 90,006,023 396,147,049 95,113,337 89,106,023 396,947,049 95,305,414 20 93,006,023 405,74 7,049 97,418,259 92,106,023 406,547,049 97,610,336 21 96,706,023 415,647,049 99,795,210 95,806,023 416,447,049 99,987,287 22 91,306,023 426,147,049 102,316,218 90,406,023 426,947,049 102,508,295 23 96,506,023 436,247,049 104,741,188 100,706,023 431,747, 049 103,660,756 24 104,906,023 447,467,049 107,435,066 109,206,023 442,957,049 106,352,233 25 112,506,023 458,457,049 110,073,721 120,506,023 450,007,049 108,044,910 26 121,306,023 470,207,049 112,894,850 129,406,023 461,757,049 110,866,038 27 126,706, 023 482,537,049 115,855,234 134,806,023 474,077,049 113,824,021 28 138,306,023 490,627,049 117,797,611 146,406,023 482,177,049 115,768,799 29 142,606,023 503,287,049 120,837,227 158,106,023 486,647,049 116,842,029 30 146,406,023 516,257,049 123,951,272 161,906,023 499,617,049 119,956,074 31 158,706,023 513,657,049 123,327,023 174,206,023 497,017,049 119,331,824 32 155,106,023 527,437,049 126,635,546 170,606,023 510,797,049 122,640,348 33 152,106,023 540,515,649 129,775,666 167,506,023 523,877,049 125, 780,804 34 156,006,023 545,946,149 131,079,508 179,756,023 520,247,049 124,909,255 35 168,706,023 553,955,049 133,002,413 192,426,023 528,260,049 126,833,145 36 181,366,023 562,781,049 135,121,500 205,126,023 537,085,949 128,952,209 37 194,276,023 574, 024,049 137,820,900 218,046,023 548,329,049 131,651,632 38 207,716,023 584,718,049 140,388,487 231,476,023 559,023,049 134,219,219 39 221,536,023 595,377,049 142,947,671 245,296,023 569,682,049 136,778,403 40 235,606,023 604,055,049 145,031,224 259,366, 023 578,360,049 138,861,957 41 250,076,023 615,452,049 147,767,599 273,839,023 589,757,049 141,598,331 42 264,556,023 626,064,049 150,315,498 288,321,023 600,370,049 144,146,470 43 279,198,023 631,291,049 151,570,480 302,959,023 605,596,049 145,401,212 44 293,560,023 633,906,049 152,198,331 317,322,023 608,211,049 146,029,063 45 307,695,023 637,557,049 153,074,922 331,457,023 611,862,049 146,905,654
212 Table 6 9. Continued Week CIPM1 NFC PS CIPM1 FCOJ PS CIPM1 FCOJ gal at 42 degree Brix CIPM2 NFC PS CIP M2 FCOJ PS CIPM2 FCOJ gal at 42 degree Brix 46 321,772,023 633,674,049 152,142,629 345,534,023 607,979,049 145,973,361 47 335,459,023 625,207,049 150,109,736 338,328,023 622,356,049 149,425,222 48 345,534,023 615,648,049 147,814,658 345,534,023 615,972, 049 147,892,449 49 341,877,023 606,089,049 145,519,580 341,877,023 606,413,049 145,597,371 50 334,462,023 596,530,049 143,224,502 334,462,023 596,854,049 143,302,293 51 324,172,023 586,971,049 140,929,424 324,172,023 587,295,049 141,007,215 52 313,882, 023 577,412,049 138,634,345 313,882,023 577,736,049 138,712,137
213 Table 6 10. CIPM3, p rocessing results for constrained industry production model with 95% hand harvesting. Week Total b oxes Total box marginal v alue Total b oxes of NFC Total b oxes of FCOJ P ound solid s of NFC Pound solid s of FCOJ 1 0 0.00 0 0 0 0 2 0 0.00 0 0 0 0 3 0 0.00 0 0 0 0 4 0 0.00 0 0 0 0 5 0 0.00 0 0 0 0 6 0 0.00 0 0 0 0 7 0 0.00 0 0 0 0 8 0 0.00 0 0 0 0 9 0 0.00 0 0 0 0 10 0 0.00 0 0 0 0 11 0 0.00 0 0 0 0 12 0 0.00 0 0 0 0 13 0 0.00 0 0 0 0 14 0 0.00 0 0 0 0 15 1,695,000 0.00 0 1,695,000 0 8,546,700 16 3,169,800 0.00 0 3,169,800 0 15,833,000 17 3,169,800 0.00 0 3,169,800 0 15,209,000 18 3,169,800 0.00 0 3,169,800 0 13,392,000 19 3,169,800 0.00 0 3,169,800 0 18,955, 000 20 3,169,800 0.00 0 3,169,800 0 19,177,000 21 6,324,200 0.46 3,154,300 3,169,800 18,468,000 19,490,000 22 6,324,200 0.52 3,154,300 3,169,800 18,680,000 19,999,000 23 6,324,200 0.57 3,154,300 3,169,800 19,179,000 19,657,000 24 6,324,200 0.63 3,154, 300 3,169,800 19,354,000 20,808,000 25 6,324,200 0.68 3,154,300 3,169,800 19,673,000 20,556,000 26 6,324,100 0.69 3,154,300 3,169,800 19,748,000 21,308,000 27 6,324,200 0.69 3,154,300 3,169,800 19,667,000 21,884,000 28 6,324,200 0.67 3,154,300 3,169,80 0 19,839,000 21,144,000 29 6,324,200 0.63 3,154,300 3,169,800 19,988,000 22,214,000 30 6,324,200 0.64 3,154,300 3,169,800 20,046,000 22,532,000 31 6,324,200 0.65 3,154,300 3,169,800 20,515,000 21,996,000 32 6,324,200 0.64 3,154,300 3,169,800 20,399,000 23,336,000 33 6,324,200 0.61 3,154,300 3,169,800 20,383,000 22,640,000 34 6,324,200 0.57 3,154,300 3,169,800 20,446,000 22,455,000 35 6,324,200 0.53 3,154,300 3,169,800 20,861,000 22,723,000 36 6,324,200 0.54 3,154,300 3,169,800 20,890,000 23,342,000 37 6,324,200 0.54 3,154,300 3,169,800 21,080,000 23,117,000 38 6,324,200 0.54 3,154,300 3,169,800 21,555,000 23,647,000 39 6,324,200 0.53 3,154,300 3,169,800 21,909,000 22,753,000 40 6,324,200 0.51 3,154,300 3,169,800 22,130,000 23,983,000 41 6,324,20 0 0.48 3,154,300 3,169,800 22,499,000 24,266,000 42 6,324,200 0.47 3,471,700 2,852,400 24,772,000 21,474,000
214 Table 6 10. Continued Week Total b oxes Total box marginal v alue Total b oxes of NFC Total b oxes of FCOJ Pound solid s of NFC Pound solid s of FCOJ 43 6,324,200 0.39 3,471,700 2,852,400 24,929,000 21,714,000 44 6,324,200 0.24 3,471,700 2,852,400 24,652,000 20,889,000 45 6,324,200 0.13 3,154,300 3,169,800 22,192,000 24,491,000 46 6,006,900 0.00 2,837,100 3,169,800 19,912,000 24,057,000 47 1,490,00 0 0.00 1,490,000 0 10,290,000 0 48 1,521,900 0.00 1,521,900 0 10,290,000 0 49 1,586,100 0.00 1,586,100 0 10,290,000 0 50 0 0.00 0 0 0 0 51 0 0.00 0 0 0 0 52 0 0.00 0 0 0 0 Sum 186,253,800 87,244,800 99,006,600 574,636,000 667,587,700
215 Table 6 11. CIPM4, p rocessing results for constrained industry production model with 95% mechanical harvesting. Week Total b oxes Total box marginal v alue Total b oxes of NFC Total b oxes of FCOJ Pound solid s of NFC Pound solid s of FCOJ 1 0 0.00 0 0 0 0 2 0 0.00 0 0 0 0 3 0 0.00 0 0 0 0 4 0 0.00 0 0 0 0 5 0 0.00 0 0 0 0 6 0 0.00 0 0 0 0 7 0 0.00 0 0 0 0 8 0 0.00 0 0 0 0 9 0 0.00 0 0 0 0 10 0 0.00 0 0 0 0 11 0 0.00 0 0 0 0 12 0 0.00 0 0 0 0 13 0 0.00 0 0 0 0 14 0 0.00 0 0 0 0 15 0 0.00 0 0 0 0 16 3,169,800 0.00 0 3,169,800 0 15,833,000 17 3,169,800 0.00 0 3,169,800 0 15,209,000 18 5,712,700 0.00 2,542,900 3,169,800 13,779,000 13,392,000 19 3,169,800 0.00 0 3,169,800 0 18,955,000 20 3,169,800 0.00 0 3,169,800 0 19,177,000 21 6,324,200 0.29 3,154,300 3,1 69,800 18,468,000 19,490,000 22 6,324,200 0.34 3,154,300 3,169,800 18,680,000 19,999,000 23 6,324,200 0.39 3,154,300 3,169,800 19,179,000 19,657,000 24 6,324,200 0.45 3,154,300 3,169,800 19,354,000 20,808,000 25 6,324,200 0.50 3,154,300 3,169,800 19,67 3,000 20,556,000 26 6,324,200 0.51 3,154,300 3,169,800 19,748,000 21,308,000 27 6,324,200 0.50 3,154,300 3,169,800 19,667,000 21,884,000 28 6,324,200 0.48 3,154,300 3,169,800 19,839,000 21,144,000 29 6,324,100 0.49 3,154,300 3,169,800 19,988,000 22,214 ,000 30 6,324,100 0.50 3,154,300 3,169,800 20,046,000 22,532,000 31 6,324,200 0.52 3,154,300 3,169,800 20,515,000 21,996,000 32 6,324,100 0.49 3,154,300 3,169,800 20,399,000 23,336,000 33 6,324,200 0.46 3,154,300 3,169,800 20,383,000 22,640,000 34 6,3 24,200 0.42 3,154,300 3,169,800 20,446,000 22,455,000 35 6,324,200 0.45 3,154,300 3,169,800 20,861,000 22,723,000 36 6,324,200 0.45 3,154,300 3,169,800 20,890,000 23,342,000 37 6,324,200 0.45 3,154,300 3,169,800 21,080,000 23,117,000 38 6,324,200 0.45 3,154,300 3,169,800 21,555,000 23,647,000 39 6,324,200 0.44 3,154,300 3,169,800 21,909,000 22,753,000 40 6,324,200 0.41 3,154,300 3,169,800 22,130,000 23,983,000 41 6,324,200 0.38 3,154,300 3,169,800 22,499,000 24,266,000 42 6,324,200 0.32 3,154,300 3, 169,800 22,507,000 23,864,000
216 Table 6 11. Continued Week Total b oxes Total box marginal v alue Total b oxes of NFC Total b oxes of FCOJ Pound solid s of NFC Pound solid s of FCOJ 43 6,324,200 0.23 3,471,700 2,852,400 24,929,000 21,714,000 44 6,324,100 0.07 3,471,700 2,852,400 24,652,000 20,889,000 45 4,711,900 0.00 2,885,400 1,826,500 20,300,000 14,112,000 46 1,466,100 0.00 1,466,100 0 10,290,000 0 47 1,490,000 0.00 1,490,000 0 10,290,000 0 48 0 0.00 0 0 0 0 49 3,172,300 0.00 3,172,300 0 20,580,000 0 5 0 1,692,700 0.00 1,692,700 0 10,290,000 0 51 1,692,700 0.00 1,692,700 0 10,290,000 0 52 1,692,700 0.00 1,692,700 0 10,290,000 0 Sum 186,090,700 92,972,800 93,115,900 605,506,000 626,995,000
217 Table 6 12 Harvesting results for CIPM3 and CIPM4, constr ained industry production model s with 95% hand harvesting and 95% mechanical harvesting with h arve st date in weeks from August 1 and portion harvested weekly in parenthesis Grove aggregate Optimal Week Harvest Week, with 95% hand harvesting Weeks, with 95% mechanical harvesting 1 21 15 16 2 21 15 16 3 21 16(.82) 17(.18) 16 4 21 15(.57) 16(.43) 16(.21) 17(.79) 5 21 16 16 6 25 28 27 7 25 27 27 8 25 27(.49) 28(.46) 29(.05) 27 9 25 28 27(.54) 28(.46) 10 25 27 27 11 25 27 24 12 25 24 25(.30) 28(.7 0) 13 25 25(.92) 26(.08) 23(.47) 24(.53) 14 25 23(.39) 26(.48) 27(.13) 23(.25) 26(.52) 28(.23) 15 25 23(.10) 24(.90) 25 16 29 35 30 17 29 33(.80) 35(.20) 33 18 29 31 31 19 29 32(.15) 33(.40) 34(.45) 32(.16) 33(.39) 34(.45) 20 29 30(.26) 31(.21) 32( .53) 30(.28) 31(.21) 32(.51) 21 22 17 17(.33) 19(.68) 22 22 17 17 23 22 17(.22) 18(.19) 19(.59) 18(.52) 22(.48) 24 22 18(.17) 20(.25) 21(.52) 22(.06) 18(.23) 20(.25) 21(.52) 25 22 22 19(.99) 22(.01) 26 24 27 24 27 24 27 24 28 24 27 24 29 24 27 24 30 24 27 24 31 21 20 20 32 21 20 20 33 21 20 20 34 21 20 20 35 21 20 20 36 23 22 22 37 23 22 22 38 23 22 22 39 23 22 22 40 23 22(.52) 23(.49) 22(.94) 23(.06) 41 29 30 30 42 29 30 30 43 29 30 30 44 29 30 30 45 29 29(.26) 30(.74) 29(.46) 30(. 55)
218 Table 6 12. Continued Grove aggregate Optimal Week Harvest Week, with 95% hand harvesting Weeks, with 95% mechanical harvesting 46 28 29 28 47 28 29 28 48 28 29 29 49 28 29 28(.09) 29(.91) 50 28 29 29 51 35 39 37(.50) 38(.35) 39(.15) 52 35 39 38 53 35 38 38 54 35 38 38 55 35 38(.91) 39(.09) 38 56 38 49 52 57 38 48(.36) 49(.64) 52 58 38 49 52 59 38 49 52 60 38 49 52 61 36 42 42 62 36 42 42 63 36 42(.02) 43(.98) 42(.01) 44(.99) 64 36 42(.23) 44(.77) 41(.73) 42(.27) 65 36 42(.24) 45( .76) 42(.20) 43(.80) 66 34 37 36(.82) 37(.18) 67 34 35(.60) 38(.40) 35 68 34 36 35 69 34 35(.63) 36(.37) 35(.12) 36(.88) 70 34 35(.13) 37(.87) 37 71 36 41 40(.98) 50(.02) 72 36 41(.76) 45(.24) 40 73 36 39(.58) 41(.41) 48(.02) 39(.65) 40(.35) 74 36 40(.53) 46(.35) 48(.12) 45(.11) 46(.13) 47(.13) 49(.29) 50(.16) 51(.16) 52(.01) 75 36 46(.58) 47(.42) 41(.04) 45(.96)
219 Table 6 13. Estimated weekly volume in storage for CIPM3 and CIPM4, NFC reported as pound solids and FCOJ reported as pound solids a nd gallons at 42 degree Brix. Week CIPM3 NFC PS CIPM3 FCOJ PS CIPM3 FCOJ gal at 42 degree Brix CIPM4 NFC PS CIPM4 FCOJ PS CIPM4 FCOJ gal at 42 degree Brix 1 264,816,023 527,988,049 126,767,839 264,816,023 527,988,049 126,767,839 2 254,526,023 518,427,049 124,472,281 254,526,023 518,427,049 124,472,281 3 244,236,023 508,867,049 122,176,963 244,236,023 508,867,049 122,176,963 4 233,946,023 499,307,049 119,881,644 233,946,023 499,307,049 119,881,644 5 223,656,023 489,757,049 117,588,727 223,656,023 489,75 7,049 117,588,727 6 213,366,023 480,197,049 115,293,409 213,366,023 480,197,049 115,293,409 7 203,076,023 470,637,049 112,998,091 203,076,023 470,637,049 112,998,091 8 192,786,023 461,077,049 110,702,773 192,786,023 461,077,049 110,702,773 9 182,496,02 3 451,517,049 108,407,455 182,496,023 451,517,049 108,407,455 10 172,206,023 441,957,049 106,112,137 172,206,023 441,957,049 106,112,137 11 161,906,023 432,447,049 103,828,823 161,906,023 432,447,049 103,828,823 12 151,606,023 422,847,049 101,523,901 15 1,606,023 422,847,049 101,523,901 13 141,306,023 413,247,049 99,218,979 141,306,023 413,247,049 99,218,979 14 131,006,023 403,747,049 96,938,067 131,006,023 403,747,049 96,938,067 15 120,706,023 402,747,049 96,697,971 120,706,023 394,147,049 94,633,145 16 110,506,023 408,947,049 98,186,566 110,506,023 400,447,049 96,145,750 17 100,206,023 414,647,049 99,555,114 100,206,023 406,047,049 97,490,288 18 89,906,023 418,447,049 100,467,479 103,706,023 409,947,049 98,426,662 19 79,606,023 427,847,049 102,724 ,382 93,406,023 419,347,049 100,683,565 20 69,306,023 437,447,049 105,029,303 83,106,023 428,947,049 102,988,487 21 77,506,023 447,407,049 107,420,660 91,306,023 438,867,049 105,370,240 22 85,906,023 457,847,049 109,927,263 99,606,023 449,307,049 107,87 6,842 23 94,806,023 467,947,049 112,352,233 108,506,023 459,397,049 110,299,412 24 103,806,023 479,197,049 115,053,313 117,606,023 470,647,049 113,000,492 25 113,206,023 490,197,049 117,694,370 127,006,023 481,647,049 115,641,548 26 122,706,023 501,947 ,049 120,515,498 136,406,023 493,397,049 118,462,677 27 132,006,023 514,267,049 123,473,481 145,806,023 505,717,049 121,420,660 28 141,606,023 525,857,049 126,256,194 155,406,023 517,307,049 124,203,373 29 151,306,023 538,507,379 129,293,488 165,106,023 529,961,049 127,241,548 30 161,006,023 551,480,049 132,408,175 174,806,023 542,933,449 130,356,170 31 171,306,023 563,917,049 135,394,249 185,056,023 555,370,049 133,342,149 32 181,386,023 577,694,049 138,702,053 195,166,023 569,147,049 136,649,952 33 191,476,023 590,775,049 141,842,749 205,256,023 582,228,049 139,790,648 34 201,636,023 603,671,049 144,939,027 215,416,023 595,125,049 142,887,167 35 212,206,023 616,836,049 148,099,892 225,986,023 608,289,049 146,047,791 36 222,806,023 630,619,049 151 ,409,135 236,586,023 622,072,049 149,357,035 37 233,596,023 644,177,049 154,664,358 247,376,023 635,630,049 152,612,257 38 244,856,023 658,267,049 158,047,311 258,636,023 649,717,049 155,994,490 39 256,476,023 671,457,049 161,214,177 270,256,023 662,907 ,049 159,161,356 40 268,318,023 685,887,049 164,678,763 282,096,623 677,337,049 162,625,942 41 280,526,223 700,587,049 168,208,175 294,305,023 692,047,049 166,157,755 42 295,008,023 712,507,049 171,070,120 306,523,023 706,347,049 169,591,128 43 309,647 ,023 724,657,049 173,987,287 321,161,023 718,507,049 172,510,696 44 324,009,023 735,987,049 176,707,575 335,523,023 729,837,049 175,230,984 45 335,911,023 750,927,049 180,294,610 345,534,023 734,387,049 176,323,421
220 Table 6 13. Continued Week CIPM3 NFC PS CIPM3 FCOJ PS CIPM3 FCOJ gal at 42 degree Brix CIPM4 NFC PS CIPM4 FCOJ PS CIPM4 FCOJ gal at 42 degree Brix 46 345,534,023 765,417,049 183,773,601 345,534,023 724,827,049 174,028,103 47 345,534,023 755,857,049 181,478,283 345,534,023 715,267,049 171,73 2,785 48 345,534,023 746,307,049 179,185,366 335,244,023 705,707,049 169,437,467 49 345,534,023 736,747,049 176,890,048 345,534,023 696,147,049 167,142,149 50 335,244,023 727,187,049 174,594,730 345,534,023 686,597,049 164,849,231 51 324,954,023 717,62 7,049 172,299,412 345,534,023 677,037,049 162,553,913 52 314,664,023 708,067,049 170,004,093 345,534,023 667,477,049 160,258,595
221 CHAPTER 7 CONCLUSIONS AND FUTU RE WORK Results suggest the Florida citrus industry as a whole could gain financially from the adoption of mechanical harvesters. Financial gains were shown to come from the reduced direct cost of harvest, as well as the increase in boxes captured and the higher average pound solids per box. Under current industry operations, however, all fina ncial gains are shown to accrue only to growers with processors experiencing either no change or losses in profits due to mechanical harvesting. These conclusions satisfy the Kaldor Hicks criterion that it is theoretically possible for mechanical harvest er adoption to make growers better off and to compensate processors such that they are made no wo rse off The potential for gains to the industry is a necessary but not a sufficient condition for Pareto improvements, in which one party is made better off and no other party is made worse off. This chapter further explains the modeling resul ts, detailing their implications to the mechanic al harvesting adoption decision and suggesting potential market mechanisms tha t could facilitate industry wide a doption The chapter concludes with suggestions for future research endeavors ex panding upon this analysis. Similarities between CIPM1 and the annual crop value estimates reported by FASS and FCPA add validity to model estimates. The model is shown to be represen tative of a typical season, and is not indented to predict output for a specific season. Results from CIPM1 estimate combined industry returns of $1.8 billion. While there is no similar publically available estimate of combined industry returns, cro p val ues of $1.5 billion in 06 07 and $1.2 billion in 07 08 were reported by FASS ( USDA, FASS, 2009) FASS crop values represent only the on tree value of the fruit and do not include returns to processors. While e stimates of processor returns are not availab le total processor returns of $300 million to $600 million would result in total industry returns of a similar magnitude to CIPM1 estimates Model estimates of boxes of fruit processed
222 to NFC and FCOJ match the actual volumes reported by FCPA for the 200 7 08 season when constraints on total box processing capacity were set to approximate the 2007 08 season ( FDOC, FCPA, 2009). These similarities d emonstrat e that both the model and the market maximize returns by processing similar quantities of each produc t. For ex ample, the model sends 86,897,000 boxes to NFC, and FCPA reported 85,111,000 boxes processed Similarly, t otal pound solids estimated to NFC are 573,854,000 and reported pound solids values to NFC were 552,253,000. Predicted quality characteris tics and actual FCPA juice qualities produced are also shown to be similar. The average soluble solid measurement for the modeled season is 12.8 degree Brix and the FCPA reported average for the 2006 07 season was 12.6 degree Brix Average r atio is estim ated at 15.9 and the actual average for 2006 07 was about 16.5. The predicted volumes of NFC and FCOJ in storage each week also approximated the average volumes reported by the FCPA from 2000 2006. Implications of Mechanical Harvesting Comparing CIPM1 t o CIPM2 represent s industry wide adoption of mech anical harvesting, holding weekly schedul ing volumes constant The most notable change in results between hand harvesting and mechanical harvesting models is the estimated $75 million (5.6%) in crease in gro wer returns. N o other aspect of industry operation changed considerably, including the number of boxe s processed to NFC and FCOJ, pound solids captured, volume of juice in storage, acreage harvested, and dates of harvest. Because grower production costs do not necessarily change due to mechanical harvester adoption, the assumed $0.50 per box cost saving results in a $75 million direct increase in grower profits The equivalent percentage increase in profits should be much larger t han the 5.6% increase in grower revenue Additional revenues are equivalent to an increase in the on tree price of about $160 per acre for each of the estimated 464,000 acres in production. For example, assuming on tree revenue of about $3,000 per acre
223 and profits of about $500 per acre (Bayles, 2008), an additional $160 per acre would increase revenue by 5% and profits by 32%. Conversely, processor returns are estimated to slightly decrease, by about $1 million ( 0.3%). These results suggest that the positi ve financial incent ives for grower support of mechanical harvesting are not necessarily available for the processing in dustry. The potential for financial losses, or at best no financial gains, suggest that processors should not favor the adoption of mechanical harvesters u nder the current system of operation. Model results estimate that the number of sets of mechanical h arvesters required to achieve 95% adoption will increase from nine in CIPM1 to 172 in CIPM2. T hese 163 additional sets of harvesters could increase total industry returns by an estimated $75 million, or about $46 0,000 in additional annual returns per set. The most commonly used Oxbo mechanical harvesters cost approximately $1.3 million per set. These additional returns would accrue to industry players, d efined for this analysis as only growers and processers not to third party harvesters Harvest equipment is assumed to be owned by these third party harvesting companies. In this analysis, p ayments to harvesting companies are measured as per box cost, w ith none of the savings from mechanical harvesting collected by harvesters. Actual gains accruing to harvesting companies would be determined by the market dynamics and competitive nature that develops in the harvester industry. T he estimated i ncrease in industry returns represents the maximum amount the industry should be willing to pay to subsidize mechanical harvester adoption The potential to incre ase returns by an average of $46 0,000 for each additional set of harvesters in operation should be ince ntive for the industry to actively push for additional investment in mechanization. Model results suggest that t hese additional investments have not been pursued aggressively by
224 the industry as a whole because the industry is comprised of many actor s, not all of who m stand to gain from potential cost saving and increased returns Comparin g CIPM1 and CIPM4 simulate s a change from current industry operation to operation with industry wide adoption of mechanical harvesters and a single weekly harvesting and processing upper bound Compa risons between CIPM1 and CIPM4 produce the greatest increase in indust ry returns, $203 million (11.5%, or about $440 per acre ) but this shift could require major changes in industry operation, some of which may not be financ ially feasible. Large gains are estimated for growers, $227 million (16.9%), and large losses are estimated for processors, $27 million ( 5.2 %). Grower gains are due to the cost savings from mechanical harvesting and from capturing an additional 122 mil lion (11%) poun d solids of juice. E xtra pound solids are captured from harvesting an additional 35,000 acres not harvested in CIPM1 and from harvesting more fruit closer to the biological optimum for pound solid production. Possible Industry Changes to Ex pedite Mechanical Harvester Adoption Estimates show it is financially possible for the Florida citrus industry as a whole to be made better off but operational changes are necessary to assure that no party is made worse off. Such ope rational changes must allow the processor to receive some benefits from mechanization Successful adoption of the mechanical tomato harvester has been credited to t he systems approach taken by the California tomato industry (Schmitz and Seckler 1970). A crucial element of t omato harvester adoption came in the form of processor subsidization by allowing changes in processing techniques and lowering pro cessing standards. Similar changes could lead to successful mechanization of the Florida citrus harvest if both growers and p rocessors are willing to accept the changes. To realize Pareto improvements, growers may need to relinquish some of the ir financial gains from mechanical harvesting and processors m ay need to accept additional costs. Such changes would move the market fr om the theoretical Kaldor Hicks
225 compensation test to an actual market mechanism for pay ing compensation to the worse off party. Operational changes and the ability to compensate for losses due to mechanical harvesting cou ld be achieved through contracting Contracts used in input markets are typically designed to introduce predic tability, share risks, and motivate performance (Hueth et al., 1999). Contracts currently used for procuring fruit in the Florida orange juice industry integrate these three principles for fruit purchase but do not address the harvest. The grower is left solely responsible for managing the harvest aside from selection of the harvest date Incorporating the harvest into contracts could allow for more predictability, r educe exposure to some risks, and improve performance. The drawback of such contracts would be that the processor would t ake on additional costs and could be subject to additional risks involving the harvest The pot ential payoff to the processor would c ome from collecting a share of any gains from mechanical harvesting. Growers would have to forfeit some of their gains from mechanical harvesting but would remain as well off or be made better off than before adoption. In addition to providing incentives for the use of mechanical harvesting, contracting changes could push both parties to optimize the timing of harvest to capture the greatest, highes t quality yields possible. Similar to the Florida citrus industry, t he California processed tomato industry uses individual contracts between grower s and processor s S ome tomato growers also use a third party negotiator, t he California Tomato Growe rs Association, to determine incentive structure s and resolve disp utes (Hueth and Ligon, 2002). Florida g rowers ch oosing to adopt mechanical harvesting could u se a similar tactic and engage a third party negotiator to help mediate the transition. An outside negotiator could allow enough growers to capture gains from mechanical harvesting that they would be able to co llectively offset any losses to processors. If processors,
226 along with growers, experience financial gains from the negotiations and operational changes, then the results would provide a Pareto improvement. For example, if processors require additional ca pital investments to handle mechanically harvested fruit efficiently, a third party group representing adoptin g growers may be able to negotiate and arrange for payouts to offset all added costs A mechanical harvesting process ing fee could be established as well This additional per unit fee for processing mechanically harvested fruit could be used not only to cover additional processing costs, but also to award processors some of the gains from mechanical harvesting. Such a fee would be designed to enco urage processors to fully support mechanical ha rvesting while preserving financial g ains to growers. With mechanical harvesting, e stimated per box gains to growers ranged from $0.45 to $1.22 per box and the estimated per box losses to processors ranged fr om $0.01 to $0.20 per box. A mutually agreeable mechanically harvested fruit processing fee would allow growers to keep some of these cost savings and more than cover additional processor costs. For example, an extra $0.20 per box, or about $100 per trai ler, may be enough to cover the extra processing costs and keep processors from being made worse off. Processors however, may require a larger portion of gains to be made no worse off. For example, a Pareto improvement could be reached by setting the pr ocessing fee below the lowest estimated gains to the grower and above the highest estimated additional cost s to the processor At $0.40 per bo x, or about $200 per trailer extra processing costs should be more than covered and growers should still retain a large portion of the gains from mechanization. The difficulty of such a fee is negotiating the fee itself Under this approach growers have an incentive to understate their gains and processors have an incentive to overstate their losses.
227 A hold up could be occurring between growers and processors, in which growers are concerned that after making financial investments to transition to mechanical harvesting, their potential gains could be captured by processors. A hold up problem could explain why me chanical harvester adoption has not occurred eve n though it may be optimal for the industry P rocessors may be able to gain additional market power after adoption from knowing that growers have already incurred the sun k costs of harvester adoption and are now saving on the cost of harvest. Pr ocessors may be able to use this information to negotiate lower fruit prices and capture some of the additional grower profits. Long term contracting before adoption could alleviate this problem if both parties agree d to a division of cost saving s Another potential problem slowing the adoption of mechanical harvesting is the like lihood that short term gains could lead to lower fruit and juice prices in the long run. Price decreases could come from two sources, bot h of which should increase the annual supply of orange juice. First, if mechanical harvesting allow s for more optimal harvest and the captur e of additional pound solids, then supply sh ould increase. As predicted by economic theory, a supply increase cet eris paribus, will push market prices down and potentially decrease industry returns. In this case, the hand harvest labor market could actually be acting as a supply control, limiting the quantity of fruit that is able to be supplied during the harvest p eriod and keeping prices higher. The second cause of lower prices from increasing supplies co uld come from additional growe rs entering the market due t o increases in profits. E conomic theory predicts that if growers are experiencing economic profits, the n additional suppliers will enter the market, new groves will be planted and prices will eventually be pushed down. Extra profits generated by mechanical harvesting could be eliminated and suppliers c ould be left producing more fruit at lower juice prices Supp ly increases and subsequent drops in market pr ices are common with advance s in
228 agricultural technology. Citrus d isease s and residential development continue to counter this increase in supply, increasing fruit and juice prices. Estimates of price changes and their effect on total returns to industry players could reduce some of the uncertainty of supply changes, potentially expediting the adoption of mechanical harvesters. Such price changes could be included in future simulations to estimate impa cts to industry players. The difficulty in finding an agree able market based solution is that all parties must not only be made better off but also believ Each play er is a resourceful, evaluative maximizer and will adapt to any changes in an attempt to maximize their own utility (Jensen and Meckling, 1994) It is not possible to account for all potential changes in behavior so any new market mechanism should account for these changes and attempt to align grower and proce ssor incentives as much as possible. Additionally growers have traditionally paid for the harvest and therefore may believe they are entitled to any cost savings afforded through m echa nical harvester adoption. At the same time, p rocessors know if the switch to mechanical harvesting is profitabl e for the grower, then the potential exists to capture some of these new profits, especially if mechanical harvester adoption is pushing additional costs on the processor. The difficulty arises in determining th e true magnitude of the total cost saving s to the grower, the total additional expenses to the processor, and how much of the savings should be transferred from the grower to the processor Future Research Processing cost estimates used for this analysis do not fully incorporate the economies of scale experienced in fruit processing. Developing cost functions estimating per unit costs as a function of input volume could enhance the accuracy of model estimates. Further research is needed detailing the cos t of proces sing, attempting to incorporate a per unit process ing cost as a function of processing intensity. This cost estimate should include the changes in production
229 costs due to running a plant at varying throughput levels and scale economies associat ed with operating facilities of different sizes. Similar to processing costs, per unit harvesting costs are dependent on the boxes per acre a grove is expected to yield. Groves yielding more boxes per acre are less costly to harvest per box. Hand harvest and mechanical harvesting prices ar e typically quoted by the third party harvester as dollars per box, with quoted prices accounting for the estimated box yield per acre. While this analysis assumes a single cost per box to harvest by hand and a single c ost per box to harvest mechanically, functions estimating harvest cost would better replicate the actual cost s of harvest. The date of harvest also impacts the quoted harvest price, due to seasonal fluctuations in the demand for harvest services and the s upply of harvest labor. Functions could be estimated for mechanical and hand harvesting in which cost variations reflect changes in box yield and date of harvest. Replacing the assumed fixed unit harvest costs with functional cost estimates of boxes per acre could improve overall model estimate s. This analy sis could be used to develop a decision aid tool for growers and grove managers to determine the economically optimal date of harvest and estimate the losses from less than optimal harvest. Specific grove characteristics such as geographic region, fruit variety, tree age, and acreage could be combined with estimated maturity curves to predict the opt imal harvest date. Expected fruit prices and harvesting costs could be added to generate profit estima tes. Growers could customize the decision tool by adding data specific to their current market and crop conditions and predictions. Similar tools may currently be available to some of the largest growing and processing operations and smaller industry pl ayers could benefit from this additional information.
230 APPENDIX A IN GROVE FRUIT MATURITY MODELING variety, region, and age, including early fruit (V10), mid season fr uit (V20), late fruit (V30), Indian River (I1), north (I2), central (I3), west (I4), south(I5), age 3 5 (G1), age 6 8 (G2), age 9 13 (G3), age 14 23 (G4), and age 24+ (G5). The t value is shown in parentheses below each parameter estimate, with *, **, and *** representing statistical significance at the 10%, 5% and 1% levels, respectively. Table A 1. Drop, r egression parameter results. VarRegAge 0 1 2 F Value R 2 V10I1G1 G5 0 0.000622 0.000315*** 410.66*** 0.74 (0.68) (6.31) V10I2G1 G5 0 0.001465 0.000132*** 161.63*** 0.53 (1.66) (2.75) V10I3G1 G5 0 0.001656** 0.000116*** 262.03*** 0.63 (2.39) (3.10) V10I4G1 G5 0 0.004381*** 5.78E 05 441.12*** 0.74 (5.64) (1.39) V10I5G1 G5 0 0.004979*** 0.000103** 531.29*** 0.77 (5.63) (2.16) V20I1G1 G5 0 0.001509* 0.00022*** 323.68*** 0.68 (1.68) (4.49) V20I2G1 G5 0 0.000864 0.000166* 36.19*** 0.41 (0.50) (1.71) V20I3G1 G5 0 0.004505** 0.000117 80.80*** 0.37 (2.09) (1.00) V20I4G1 G5 0 0.003177*** 0.000107* 192.73*** 0.57 (2.97) (1.83) V20I5G1 G5 0 0.005041*** 4.65E 05 317.73*** 0.67 (5.11) (0.88) V30I1G1 G5 0 0.004395*** 5.18E 05*** 1163.36*** 0.82 (8.97) (2.99) V30I2G1 G5 0 0.002584*** 5.01E 05*** 471.80*** 0.68 (4.96) (2.68) V30I3G1 G5 0 0.003214* ** 3.64E 05*** 1108.15*** 0.82 (8.84) (2.82) V30I4G1 G5 0 0.002689*** 0.000072*** 1054.31*** 0.81 (6.51) (4.92) V30I5G1 G5 0 0.003765*** 3.89E 05*** 1441.49*** 0.85 (10.30) (3.00)
231 Table A 2. Fruit weight regression parameter results. VarRegAge 0 1 2 F Value R 2 V10I1G1 G5 0.26351*** 0.0071*** 0.00016*** 23.27*** 0.08 (22.44) (3.98) ( 2.65) V10I2G1 G5 0.27043*** 0.00726*** 0.00012 24.17*** 0.16 (16.17) (2.83) ( 1.35) V10I3G1 G5 0.23905*** 0.00751*** 0.00011*** 185.95*** 0.22 (35.20) (7.24) ( 3.12) V10I4G1 G5 0.28245*** 0.00369*** 3.5E 05 96.34*** 0.08 (48.57) (4.24) ( 1.23) V10I5G1 G5 0.27676*** 0.00731*** 0.00015*** 90.19*** 0.10 (42.46) (7.29) ( 4.51) V20I1G1 G5 0.28369*** 0.00816*** 0.00018*** 27.99** 0.08 (22.41) (4.38) ( 3.01) V20I2G1 G5 0.22755*** 0.01695* 0.00046 2.48 0.13 (3.82) (1.95) ( 1.68) V20I3G1 G5 0.24627*** 0.01012*** 0.00018*** 60.78*** 0.18 (17.92) (5.00) ( 2.78) V20I4G1 G5 0.28628*** 0.00432*** 3E 05 74.76*** 0.11 ( 32.56) (3.35) ( 0.72) V20I5G1 G5 0.29294*** 0.00642*** 9.9E 05* 56.39*** 0.14 (28.44) (4.10) ( 1.89) V30I1G1 G5 0.26706*** 0.00832*** 0.00012*** 197.13*** 0.16 (31.29) (10.89) ( 7.49) V30I2G1 G5 0.2671*** 0.00663*** 8.6E 05*** 35.89*** 0.16 (15.04) (4.18) ( 2.71) V30I3G1 G5 0.23828*** 0.01117*** 0.00015*** 672.58*** 0.30 (36.67) (18.74) ( 12.36) V30I4G1 G5 0.23926*** 0.01007*** 0.00014*** 375.17*** 0.21 (33.58) (15.64) ( 11.00) V30I5G1 G5 0.28027*** 0.00837*** 0.0001*** 478 .95*** 0.21 (44.33) (14.70) ( 9.04)
232 Table A 3. Pound solids per box regression parameter results. VarRegAge 0 1 2 F Value R 2 V10I1G1 G5 3.67885*** 0.18899*** 0.00234*** 354.83 0.58 (27.07) (9.16) ( 3.38) V10I2G1 G5 3.3504*** 0.24104*** 0.00406*** 261.73 0.68 (20.56) (9.64) ( 4.81) V10I3G1 G5 3.40018*** 0.23467*** 0.0043*** 1020.94 0.61 (45.71) (20.66) ( 11.29) V10I4G1 G5 3.51056*** 0.20934*** 0.0031*** 1987.96 0.63 (61.13) (24.34) ( 10.95) V10I5G1 G5 3.57341*** 0.19252** 0.00284*** 1072.87 0.58 (52.03) (18.22) ( 7.99) V20I1G1 G5 3.33952*** 0.21271*** 0.00188*** 945.01 0.75 (28.67) (12.41) ( 3.39) V20I2G1 G5 3.56492*** 0.22782*** 0.00299 39.99 0.80 (7.03) (3.07) ( 1.29) V20I3G1 G5 3.15698*** 0.26879*** 0.00484*** 557.75 0.68 (26.13) (15.09) ( 8.37) V20I4G1 G5 3.24385*** 0.25543*** 0.00388*** 1486.51 0.72 (40.15) (21.56) ( 10.18) V20I5G1 G5 3.2709*** 0.22523*** 0.00317*** 931.29 0.73 (34.41) (15.58) ( 6.56) V30I1G1 G5 2.575*** 0.23734*** 0.00227*** 3073.06 0.75 (27.05) (27.83) ( 13.22) V30I2G1 G5 2.71659*** 0.21867*** 0.00209*** 694.58 0.79 (14.51) (13.07) ( 6.22) V30I3G1 G5 2.60011*** 0.2268*** 0.00242*** 5127.11 0.77 (40.45) (38.48) ( 19.83) V30I4G1 G5 2.57186*** 0.2294 5*** 0.00235*** 4721.92 0.77 (36.94) (36.47) ( 18.36) V30I5G1 G5 2.27708*** 0.24728*** 0.00282*** 5174.72 0.74 (35.85) (43.23) ( 24.22)
233 Table A 4. Brix to acid ratio regression parameter results. VarRegAge 0 1 2 F Value R 2 V10I1G1 G5 2. 04416 *** 0.90933 *** 0.01256 *** 975.5*** 0.79 (5.53) (16.20) ( 6.68) V10I2G1 G5 2.19455 *** 0.76152 *** 0.00869 *** 459.3*** 0.79 (4.40) (9.95) ( 3.36) V10I3G1 G5 2.05747 *** 0.93937 *** 0.01305 *** 1932.2*** 0.75 (7.60) (22.72) ( 9.41) V10I4G1 G5 1.84707 *** 0.90426 *** 0.01282 *** 4043.0*** 0.78 (10.30) (33.67) ( 14.51) V10I5G1 G5 2.16908 *** 1.008 *** 0.01616 *** 2312.7*** 0.74 (9.42) (28.44) ( 13.56) V20I1G1 G5 1.71531 *** 0.75402 *** 0.00809 *** 1422.7*** 0.82 (5.52) (16.49) ( 5.48) V20I2G1 G5 3.38047 0.5025 0.00426 18.3*** 0.63 (1.65) (1.68) ( 0.46) V20I3G1 G5 2.03266 *** 0.66019 *** 0.00578 *** 915.7*** 0.78 (5.62) (12.39) ( 3.34) V20I4G1 G5 1.61851 *** 0.75378 *** 0.0088 *** 2137.3*** 0.79 (6.86) (21.80) ( 7.91) V20I5G 1 G5 1.90741 *** 0.83352 *** 0.00896 *** 1118.0*** 0.76 (5.13) (14.76) ( 4.75) V30I1G1 G5 1.50367 *** 0.21833 *** 0.00321 *** 9104.9*** 0.90 (9.29) (15.06) ( 10.99) V30I2G1 G5 1.17044 *** 0.23589 *** 0.00227 *** 1567.0*** 0.89 (3.19) (7.19) ( 3.44) V3 0I3G1 G5 1.3471 *** 0.24148 *** 0.00244 *** 12584.0*** 0.89 (10.48) (20.49) ( 10.02) V30I4G1 G5 1.35242 *** 0.23847 *** 0.00239 *** 10845.5*** 0.89 (9.81) (19.15) ( 9.45) V30I5G1 G5 1.52132 *** 0.23227 *** 0.00298 *** 15721.9*** 0.90 (12.70) (21.54) ( 13.5 9)
234 Table A 5. Total soluble solids regression parameter results. VarRegAge 0 1 2 F Value R 2 V10I1G1 G5 9.35901*** 0.03898* 0.00354*** 402.59*** 0.61 (62.09) (1.70) (4.62) V10I2G1 G5 8.85942*** 0.07854** 0.00224** 187.64*** 0.60 (40.51) (2.34) (1.98) V10I3G1 G5 8.8994*** 0.08625*** 0.00148*** 940.32*** 0.59 (100.24 ) (6.36) (3.25) V10I4G1 G5 9.02999*** 0.04982*** 0.00314*** 2232.51*** 0.66 (137.77) (5.08) (9.71) V10I5G1 G5 8.97897*** 0.04441*** 0.00289*** 1028.86*** 0.56 (110.56) (3.55) (6.88) V20I1G1 G5 8.93735*** 0.01854 0.00562*** 1017.74*** 0.76 (6 4.82) (0.91) (8.58) V20I2G1 G5 8.82732*** 0.11249 0.0023 45.93*** 0.82 (13.66) (1.19) (0.78) V20I3G1 G5 8.48336*** 0.10769*** 0.002*** 872.7*** 0.77 (66.39) (5.72) (3.27) V20I4G1 G5 8.57234*** 0.0953*** 0.00247*** 1741.98*** 0.75 (94.18) (7. 14) (5.75) V20I5G1 G5 8.54873*** 0.07228 *** 0.00297*** 852.1*** 0.71 (72.08) (4.01) (4.93) V30I1G1 G5 6.83712*** 0.22297*** 0.00122*** 4102.96*** 0.80 (65.05) (23.68) ( 6.43) V30I2G1 G5 6.75499*** 0.22335*** 0.00143*** 828.99*** 0.81 (30.2 6) (11.19) ( 3.58) V30I3G1 G5 6.72208*** 0.22236*** 0.00147*** 6591.99*** 0.81 (88.42) (31.90) ( 10.18) V30I4G1 G5 6.88003*** 0.20304 *** 0.00098 *** 6951.24*** 0.83 (90.93) (29.70) ( 7.07) V30I5G1 G5 6.37326*** 0.23603*** 0.0017 *** 6784.46** 0.79 (85.21) (35.04) ( 12.45)
235 Table A 6. Average pieces per tree (PPT), average tree density (DEN), and bearing acreage (A) point estimate parameter results. VarRegAge PPT (pieces) DEN (trees per acre) A (acres) V10I1G1 221.5256 141.189 1884 V1 0I1G2 393.666 147.0346 1703 V10I1G3 567.709 134.0466 4159 V10I1G4 951.2 121.583 7075 V10I1G5 874.552 102.8554 4973 V10I2G1 122.775 134.5566 654 V10I2G2 700.868 129.4118 1139 V10I2G3 842.152 133.1595 8055 V10I2G4 1177.824 130.8448 10783 V10I2G5 1195 .724 96.41214 1087 V10I3G1 445.055 123.5462 4557 V10I3G2 647.376 129.2147 3731 V10I3G3 956.174 133.6903 19798 V10I3G4 1481.842 125.9995 22612 V10I3G5 2022.019 99.71004 8277 V10I4G1 328.962 122.0979 5005 V10I4G2 610.603 125.3846 4160 V10I4G3 1020.68 8 134.352 13612 V10I4G4 1477.448 124.2572 28911 V10I4G5 1758.163 100.9781 13495 V10I5G1 276.515 142.1239 5923 V10I5G2 579.463 153.7106 4541 V10I5G3 869.371 155.6779 14715 V10I5G4 1206.587 147.4309 24250 V10I5G5 1344.638 122.1342 4798 V20I1G1 156.72 4 146.1975 881 V20I1G2 311.991 136.5411 717 V20I1G3 396.859 132.2646 1965 V20I1G4 601.207 126.6348 2095 V20I1G5 859.551 108.1887 3456 V20I2G1 186.905 118.0328 61 V20I2G2 144.34 148.3871 31 V20I2G3 498.945 126.5306 98 V20I2G4 620.648 113.5135 185 V 20I2G5 1061.272 88.78924 223 V20I3G1 322.832 114.9178 791 V20I3G2 692.1456 120.4545 352 V20I3G3 901.659 128.9577 2034 V20I3G4 1379.18 123.9033 2234 V20I3G5 1748.944 100.0301 3319 V20I4G1 179.837 130.1911 785 V20I4G2 361.425 129.0931 849 V20I4G3 703 .119 142.047 1446
236 Table A 6. Continued V20I4G4 1228.295 124.6964 2223 V20I4G5 1608.678 108.0667 6533 V20I5G1 196.365 139.8279 1976 V20I5G2 458.384 138.4966 1317 V20I5G3 766.422 158.712 4643 V20I5G4 1008.691 141.3992 4860 V20I5G5 1255.577 125.8774 2 365 V30I1G1 143.994 138.626 3741 V30I1G2 265.184 136.1237 2879 V30I1G3 419.782 156.9377 8216 V30I1G4 552.498 134.0241 11389 V30I1G5 687.565 102.2739 7784 V30I2G1 342.144 143.3852 514 V30I2G2 388.772 144.743 856 V30I2G3 591.038 142.5717 1952 V30I2G 4 804.03 136.0233 2404 V30I2G5 1165.037 84.5655 771 V30I3G1 232.605 126.794 10103 V30I3G2 441.493 133.9439 8134 V30I3G3 586.752 133.8068 27808 V30I3G4 853.983 125.1321 24411 V30I3G5 1355.169 102.8414 14676 V30I4G1 280.003 129.5106 9705 V30I4G2 552. 316 136.1517 6564 V30I4G3 645.037 138.1832 15643 V30I4G4 839.195 131.6889 18101 V30I4G5 1177.399 108.0735 12795 V30I5G1 183.786 138.5183 10569 V30I5G2 376.62 151.9266 11004 V30I5G3 485.045 164.1113 40659 V30I5G4 644.204 151.2184 38207 V30I5G5 917.7 98 127.6658 8534
237 Table A 7. Boxes per a cre regression parameter results. VarRegAge int t t^2 t^3 t^4 V10I1G1 91.575 2.4104 0.0855 7.432E 04 1.736E 05 V10I1G2 169.474 4.4608 0.1582 1.375E 03 3.213E 05 V10I1G3 222.811 5.8647 0.2080 1.808E 03 4. 224E 05 V10I1G4 338.610 8.9127 0.3161 2.748E 03 6.420E 05 V10I1G5 263.371 6.9323 0.2459 2.137E 03 4.993E 05 V10I2G1 49.639 1.2599 0.0301 1.445E 04 2.846E 06 V10I2G2 272.535 6.9172 0.1650 7.932E 04 1.563E 05 V10I2G3 336.957 8.5523 0.2040 9.80 7E 04 1.932E 05 V10I2G4 463.073 11.7532 0.2804 1.348E 03 2.655E 05 V10I2G5 346.398 8.7919 0.2097 1.008E 03 1.986E 05 V10I3G1 146.046 4.3464 0.0908 4.225E 04 7.688E 06 V10I3G2 222.185 6.6123 0.1382 6.428E 04 1.170E 05 V10I3G3 339.534 10.1046 0 .2111 9.822E 04 1.787E 05 V10I3G4 495.926 14.7589 0.3084 1.435E 03 2.611E 05 V10I3G5 535.513 15.9371 0.3330 1.549E 03 2.819E 05 V10I4G1 126.053 1.0945 0.0302 2.650E 05 9.062E 07 V10I4G2 240.271 2.0863 0.0575 5.051E 05 1.727E 06 V10I4G3 430.36 4 3.7369 0.1030 9.047E 05 3.094E 06 V10I4G4 576.146 5.0028 0.1379 1.211E 04 4.142E 06 V10I4G5 557.167 4.8380 0.1334 1.171E 04 4.005E 06 V10I5G1 120.850 2.5903 0.0948 2.855E 06 6.834E 06 V10I5G2 273.899 5.8708 0.2149 6.471E 06 1.549E 05 V10I5G3 416.191 8.9207 0.3265 9.833E 06 2.354E 05 V10I5G4 547.026 11.7250 0.4291 1.292E 05 3.094E 05 V10I5G5 505.014 10.8245 0.3961 1.193E 05 2.856E 05 V20I1G1 72.223 1.9684 0.0652 3.871E 04 1.015E 05 V20I1G2 134.279 3.6597 0.1212 7.198E 04 1.888E 05 V20I1G3 165.456 4.5094 0.1493 8.869E 04 2.326E 05 V20I1G4 239.982 6.5406 0.2166 1.286E 03 3.373E 05 V20I1G5 293.126 7.9891 0.2646 1.571E 03 4.121E 05 V20I2G1 55.777 4.1066 0.1252 5.927E 04 1.865E 05 V20I2G2 54.152 3.9870 0.1216 5.754E 04 1.81 1E 05 V20I2G3 159.618 11.7520 0.3583 1.696E 03 5.337E 05 V20I2G4 178.126 13.1146 0.3998 1.893E 03 5.956E 05 V20I2G5 238.244 17.5408 0.5348 2.532E 03 7.966E 05 V20I3G1 101.516 3.7143 0.1060 1.477E 04 8.803E 06 V20I3G2 228.134 8.3470 0.2383 3. 320E 04 1.978E 05 V20I3G3 318.170 11.6413 0.3323 4.630E 04 2.759E 05 V20I3G4 467.598 17.1086 0.4883 6.804E 04 4.055E 05 V20I3G5 478.714 17.5153 0.4999 6.966E 04 4.151E 05 V20I4G1 74.475 0.8873 0.0192 9.522E 05 8.247E 07 V20I4G2 148.412 1.7681 0.0384 1.898E 04 1.644E 06 V20I4G3 317.694 3.7849 0.0821 4.062E 04 3.518E 06 V20I4G4 487.197 5.8042 0.1259 6.229E 04 5.395E 06
238 Table A 7. Continued V20I4G5 552.980 6.5879 0.1429 7.070E 04 6.124E 06 V20I5G1 89.371 1.5081 0.0442 6.122E 05 1.405 E 06 V20I5G2 206.635 3.4869 0.1023 1.416E 04 3.248E 06 V20I5G3 395.926 6.6812 0.1960 2.712E 04 6.224E 06 V20I5G4 464.239 7.8339 0.2298 3.180E 04 7.298E 06 V20I5G5 514.431 8.6809 0.2547 3.524E 04 8.087E 06 V30I1G1 59.232 1.5850 0.0368 1.683E 05 1. 325E 06 V30I1G2 107.114 2.8663 0.0665 3.043E 05 2.396E 06 V30I1G3 195.487 5.2310 0.1213 5.554E 05 4.373E 06 V30I1G4 219.725 5.8796 0.1363 6.243E 05 4.915E 06 V30I1G5 208.663 5.5836 0.1295 5.929E 05 4.667E 06 V30I2G1 145.594 3.2378 0.0637 5.945E 05 2.358E 06 V30I2G2 167.003 3.7138 0.0731 6.819E 05 2.705E 06 V30I2G3 250.081 5.5613 0.1094 1.021E 04 4.050E 06 V30I2G4 324.576 7.2180 0.1420 1.325E 04 5.256E 06 V30I2G5 292.391 6.5022 0.1279 1.194E 04 4.735E 06 V30I3G1 78.084 3.4094 0.0645 2.708E 05 1.815E 06 V30I3G2 156.564 6.8361 0.1293 5.429E 05 3.640E 06 V30I3G3 207.863 9.0760 0.1717 7.208E 05 4.833E 06 V30I3G4 282.920 12.3532 0.2336 9.811E 05 6.578E 06 V30I3G5 368.983 16.1110 0.3047 1.280E 04 8.579E 06 V30I4G1 96.404 3.7983 0. 0759 1.360E 04 4.180E 06 V30I4G2 199.912 7.8764 0.1574 2.821E 04 8.669E 06 V30I4G3 236.956 9.3359 0.1866 3.344E 04 1.028E 05 V30I4G4 293.792 11.5752 0.2313 4.146E 04 1.274E 05 V30I4G5 338.275 13.3279 0.2664 4.773E 04 1.467E 05 V30I5G1 79.278 2.0691 0.0416 1.935E 05 1.152E 06 V30I5G2 178.185 4.6505 0.0935 4.350E 05 2.588E 06 V30I5G3 247.887 6.4697 0.1301 6.052E 05 3.601E 06 V30I5G4 303.363 7.9176 0.1592 7.406E 05 4.407E 06 V30I5G5 364.885 9.5233 0.1915 8.908E 05 5.300E 06
239 APPENDIX B SAS PROGRAMMING CODE Estimated equations for fruit drop DATA FASSALLD; INFILE 'C: \ Documents and Settings \ jsearcy \ Desktop \ A1 FASS DROP.txt' expandtabs; INPUT YR DAY WEEK AREA $ AGE $ TYPE $ DROP; WEEKSQ = WEEK WEEK; PERDROP = DROP / 100 ; RUN ; DATA FAS SD10 ; SET FASSALLD; IF TYPE = "10"; RUN ; DATA FASSD10A1; SET FASSD10; IF AREA = "1"; RUN ; DATA FASSD10A2; SET FASSD10; IF AREA = "2"; RUN ; DATA FASSD10A3; SET FASSD10; IF AREA = "3"; RUN ; DATA FASSD10A4; SET FASSD10; IF AREA = "4"; RUN ; DATA FASSD10A5; SE T FASSD10; IF AREA = "5"; RUN ; DATA FASSD20; SET FASSALLD; IF TYPE = "20"; RUN ; DATA FASSD20A1; SET FASSD20; IF AREA = "1"; RUN ; DATA FASSD20A2; SET FASSD20; IF AREA = "2"; RUN ; DATA FASSD20A3; SET FASSD20; IF AREA = "3"; RUN ; DATA FASSD20A4; SET FASSD20; IF AREA = "4"; RUN ; DATA FASSD20A5; SET FASSD20; IF AREA = "5"; RUN ; DATA FASSD30; SET FASSALLD; IF TYPE = "30"; RUN ; DATA FASSD30A1; SET FASSD30; IF AREA = "1"; RUN ; DATA FASSD30A2; SET FASSD30; IF AREA = "2";
240 RUN ; DATA FASSD30A3; SET FASSD30; IF AREA = 3"; RUN ; DATA FASSD30A4; SET FASSD30; IF AREA = "4"; RUN ; DATA FASSD30A5; SET FASSD30; IF AREA = "5"; RUN ; DATA DROP10A1; SET FASSD10A1; PROC REG ; DR10A1: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP10A2; SET FASSD10A2; PROC REG ; DR10A2: MODEL PERDROP =WEEK WEEKSQ / noint; RUN ; DATA DROP10A3; SET FASSD10A3; PROC REG ; DR10A3: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP10A4; SET FASSD10A4; PROC REG ; DR10A4: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP10A5; SET FASSD10A5; PROC REG ; DR10A5: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP20A1; SET FASSD20A1; PROC REG ; DR20A1: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP20A2; SET FASSD20A2; PROC REG ; DR20A2: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP20A3; SET FASSD20A3; PROC REG ; DR20A 3: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP20A4; SET FASSD20A4; PROC REG ; DR20A4: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP20A5; SET FASSD20A5; PROC REG ; DR20A5: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP30A1; SET FASSD30A1; PROC RE G ; DR30A1: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP30A2; SET FASSD30A2; PROC REG ;
241 DR30A2: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP30A3; SET FASSD30A3; PROC REG ; DR30A3: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP30A4; SET FASSD30A4; PROC REG ; DR30A4: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; DATA DROP30A5; SET FASSD30A5; PROC REG ; DR30A5: MODEL PERDROP=WEEK WEEKSQ / noint; RUN ; __ Es timated equations for fruit weight, ratio, pound solids, and total soluble solids DATA FASSALL; INFILE 'C: \ Documents and Settings \ jsearcy \ Desktop \ A1 FASS ALL.txt' expandtabs; INPUT YR DAY WEEK AREA $ AGE $ YRPLANT TYPE $ ID $ ACID TSS RATIO JUCPER PS FRUWT JUCWT; WEEKSQ = WEEK WEEK; FRUPCWT = FRUWT / 15 ; RUN ; DATA FASS10; SET FASSALL; IF TYPE = "10"; RU N ; DATA FASS10A1; SET FASS10; IF AREA = "1"; RUN ; DATA FASS10A2; SET FASS10; IF AREA = "2"; RUN ; DATA FASS10A3; SET FASS10; IF AREA = "3"; RUN ; DATA FASS10A4; SET FASS10; IF AREA = "4"; RUN ; DATA FASS10A5; SET FASS10; IF AREA = "5"; RUN ; DATA FASS20; SET F ASSALL; IF TYPE = "20"; RUN ; DATA FASS20A1; SET FASS20; IF AREA = "1"; RUN ; DATA FASS20A2; SET FASS20; IF AREA = "2"; RUN ; DATA FASS20A3; SET FASS20; IF AREA = "3"; RUN ;
242 DATA FASS20A4; SET FASS20; IF AREA = "4"; RUN ; DATA FASS20A5; SET FASS20; IF AREA = "5 "; RUN ; DATA FASS30; SET FASSALL; IF TYPE = "30"; RUN ; DATA FASS30A1; SET FASS30; IF AREA = "1"; RUN ; DATA FASS30A2; SET FASS30; IF AREA = "2"; RUN ; DATA FASS30A3; SET FASS30; IF AREA = "3"; RUN ; DATA FASS30A4; SET FASS30; IF AREA = "4"; RUN ; DATA FASS30A5 ; SET FASS30; IF AREA = "5"; RUN ; DATA EST10A1; SET FASS10A1; PROC REG ; FTWT10A1: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST10A2; SET FASS10A2; PROC REG ; FTWT10A2: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST10A3; SET FASS10A3; PROC REG ; FTWT10A3: MODEL FRUPCWT= WEEK WEEKSQ; RUN ; DATA EST10A4; SET FASS10A4; PROC REG ; FTWT10A4: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST10A5; SET FASS10A5; PROC REG ; FTWT10A5: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST20A1; SET FASS20A1; PROC REG ; FTWT20A1: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST20A2; SET FASS20A2; PROC REG ; FTWT20A2: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST20A3; SET FASS20A3; PROC REG ; FTWT20A3: MODEL FRUPCWT=WEEK WEEKSQ; RUN ;
243 DATA EST20A4; SET FASS20A4; PROC REG ; FTWT20A4: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST20A5; SET FASS20A5; PROC REG ; FTWT20A5: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST30A1; SET FASS30A1; PROC REG ; FTWT30A1: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST30A2; SET FASS30A2; PROC REG ; FTWT30A2: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST30A3; SET FASS30A3; PR OC REG ; FTWT30A3: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST30A4; SET FASS30A4; PROC REG ; FTWT30A4: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; DATA EST30A5; SET FASS30A5; PROC REG ; FTWT30A5: MODEL FRUPCWT=WEEK WEEKSQ; RUN ; __ DATA EST10A1; SET FASS10A1; PROC REG ; RA10A 1: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST10A2; SET FASS10A2; PROC REG ; RA10A2: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST10A3; SET FASS10A3; PROC REG ; RA10A3: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST10A4; SET FASS10A4; PROC REG ; RA10A4: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST10A5; SET FASS10A5; PROC REG ; RA10A5: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST20A1; SET FASS20A1; PROC REG ; RA20A1: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST20A2; SET FASS20A2; PROC REG ; RA20A2: MODEL RATIO=WEEK WEEKSQ;
244 RUN ; DATA EST20A3; SET FASS2 0A3; PROC REG ; RA20A3: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST20A4; SET FASS20A4; PROC REG ; RA20A4: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST20A5; SET FASS20A5; PROC REG ; RA20A5: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST30A1; SET FASS30A1; PROC REG ; RA30A1: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST30A2; SET FASS30A2; PROC REG ; RA30A2: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST30A3; SET FASS30A3; PROC REG ; RA30A3: MODEL RATIO=WEEK WEEKSQ; RUN ; DATA EST30A4; SET FASS30A4; PROC REG ; RA30A4: MODEL RATIO=WEEK WEEKSQ; RUN ; DAT A EST30A5; SET FASS30A5; PROC REG ; RA30A5: MODEL RATIO=WEEK WEEKSQ; RUN ; _____ DATA EST10A1; SET FASS10A1; PROC REG ; PS10A1: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST10A2; SET FASS10A2; PROC REG ; PS10A2: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST10A3; SET FASS10A3; PRO C REG ; PS10A3: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST10A4; SET FASS10A4; PROC REG ; PS10A4: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST10A5; SET FASS10A5; PROC REG ; PS10A5: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST20A1; SET FASS20A1; PROC REG ;
245 PS20A1: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST20A2; SET FASS20A2; PROC REG ; PS20A2: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST20A3; SET FASS20A3; PROC REG ; PS20A3: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST20A4; SET FASS20A4; PROC REG ; PS20A4: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST20A5; SET FASS20A5; PROC REG ; PS20A5: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST30A1; SET FASS30A1; PROC REG ; PS30A1: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST30A2; SET FASS30A2; PROC REG ; PS30A2: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST30A3; SET FASS30A3; PROC REG ; PS30A3: MODEL PS=WEEK WEEKSQ; R UN ; DATA EST30A4; SET FASS30A4; PROC REG ; PS30A4: MODEL PS=WEEK WEEKSQ; RUN ; DATA EST30A5; SET FASS30A5; PROC REG ; PS30A5: MODEL PS=WEEK WEEKSQ; RUN ; __ DATA EST10A1; SET FASS10A1; PROC REG ; TSS10A1: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST10A2; SET FASS10A2; PROC REG ; TSS10A2: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST10A3; SET FASS10A3; PROC REG ; TSS10A3: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST10A4; SET FASS10A4; PROC REG ; TSS10A4: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST10A5; SET FASS10A5;
246 PROC REG ; TSS10A5: MODEL TSS=WE EK WEEKSQ; RUN ; DATA EST20A1; SET FASS20A1; PROC REG ; TSS20A1: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST20A2; SET FASS20A2; PROC REG ; TSS20A2: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST20A3; SET FASS20A3; PROC REG ; TSS20A3: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST20A4; S ET FASS20A4; PROC REG ; TSS20A4: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST20A5; SET FASS20A5; PROC REG ; TSS20A5: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST30A1; SET FASS30A1; PROC REG ; TSS30A1: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST30A2; SET FASS30A2; PROC REG ; TSS30A2: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST30A3; SET FASS30A3; PROC REG ; TSS30A3: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST30A4; SET FASS30A4; PROC REG ; TSS30A4: MODEL TSS=WEEK WEEKSQ; RUN ; DATA EST30A5; SET FASS30A5; PROC REG ; TSS30A5: MODEL TSS=WEEK WEEKSQ; RUN ;
247 APPENDIX C BIOLOGICAL FRUIT AND DROP DATA Figure C l. Pound solids per box for early fruit sampled in the Indian River region. Figure C 2. Acid percentage for early fruit sampled in the Indian River region.
248 Figure C 3. Brix to acid ratio for early fruit sampled in the Indian River region. Figure C 4. Total soluble solid percentage for early fruit sampled in the Indian River region.
249 Figure C 5. Fruit weight for 15 piece sample of early fruit in the Indian River region. Figure C 6 Juice weight for 15 piece sample of early fruit in the Indian River region.
250 Figure C 7. Pound solids per box for early fruit sampled in the N orth region. Figure C 8. Acid percentage for early fruit sampled in the N orth region.
251 Figure C 9. Brix to acid ratio for early fruit sampled in the N orth region. Figure C 10. Total soluble solid percentage for early fruit sampled in the N orth region.
252 Figure C 11. Fruit weight for 15 piec e sample of early fruit in the N orth region. Figure C 12. Juice weight for 15 piec e sample of early fruit in the N orth region.
253 Figure C 13. Pound solids per box for early fruit sampled in the C entral region. Figure C 14. Acid percentage for early fruit sampled in the C entral region.
254 Figure C 15. Brix to acid ratio for early fruit sampled in the C entral region. Figure C 16. Total soluble solid percentage for early fruit sampled in the C entral region.
255 Figure C 17. Fruit weight for 15 piec e sample of early fruit in the C entral region. Figure C 18. Juice weight for 15 piec e sample of early fruit in the C entral region.
256 Figure C 19. Pound solids per box for early fruit sampled in the W est region. Figure C 20. Acid percentage for early fruit sampled in the W est region.
257 F igure C 21. Brix to acid ratio for early fruit sampled in the W est region. Figure C 22. Total soluble solid percentage for early fruit sampled in the W est region.
258 Figure C 23. Fruit weight for 15 piec e sample of early fruit in the W est region. Figure C 24. Juice weight for 15 piec e sample of early fruit in the W est region.
259 Figure C 25. Pound solids per box for early fruit sampled in the S outh region. Figure C 26. Acid percentage for early fruit sampled in the S outh region.
260 Figu re C 27. Brix to acid ratio for early fruit sampled in the S outh region. Figure C 28. Total soluble solid percentage for early fruit sampled in the S outh region.
261 Figure C 29. Fruit weight for 15 piec e sample of early fruit in the S outh region. Figure C 30. Juice weight for 15 piec e sample of early fruit in the S outh region.
262 Figure C 3l. Pound solids per box for mid season fruit sampled in the Indian River region. Figure C 32. Acid percentage for mid season fruit sampled in the Indi an River region.
263 Figure C 33. Brix to acid ratio for mid season fruit sampled in the Indian River region. Figure C 34. Total soluble solid percentage for mid season fruit sampled in the Indian River region.
264 Figure C 35. Fruit weight for 15 p iece sample of mid season fruit in the Indian River region. Figure C 36. Juice weight for 15 piece sample of mid season fruit in the Indian River region.
265 Figure C 37. Pound solids per box for mid season fruit sampled in the north region. Figur e C 38. Acid percentage for mid season fruit sampled in the north region.
266 Figure C 39. Brix to acid ratio for mid season fruit sampled in the north region. Figure C 40. Total soluble solid percentage for mid season fruit sampled in the north reg ion.
267 Figure C 41. Fruit weight for 15 piece sample of mid season fruit in the north region. Figure C 42. Juice weight for 15 piece sample of mid season fruit in the north region.
268 Figure C 43. Pound solids per box for mid season fruit sampled in the C entral region. Figure C 44. Acid percentage for m id season fruit sampled in the C entral region.
269 Figure C 45. Brix to acid ratio for m id season fruit sampled in the C entral region. Figure C 46. Total soluble solid percentage for mid season fruit sampled in the C entral region.
270 Figure C 47. Fruit weight for 15 piece sam ple of mid season fruit in the C entral region. Figure C 48. Juice weight for 15 piece sam ple of mid season fruit in the C entral region.
271 Figure C 49. Pound solids per box for m id season fruit sampled in the W est region. Figure C 50. Acid percentage for m id season fruit sampled in the W est region.
272 Figure C 51. Brix to acid ratio for m id season fruit sampled in the W est region. Figure C 52. Total soluble solid percentage for m id season fruit sampled in the W est region.
273 Figure C 53. Fruit weight for 15 piece sam ple of mid season fruit in the W est region. Figure C 54. Juice weight for 15 piece sam ple of mid season fruit in the W est region.
274 Figure C 55. Pound solids per box for m id season fruit sampled in the S outh region. Figure C 56. Acid percentage for m id season fruit sampled in the S outh region.
275 Figure C 57. Brix to acid ratio for m id season fruit sampled in the S outh reg ion. Figure C 58. Total soluble solid percentage for m id season fruit sampled in the S outh region.
276 Figure C 59. Fruit weight for 15 piece sample of mid season fruit in the S outh region. Figure C 60. Juice weight for 15 piece sample of mid se ason fruit in the S outh region.
277 Figure C 6l. Pound solids per box for late season fruit sampled in the Indian River region. Figure C 62. Acid percentage for late season fruit sampled in the Indian River region.
278 Figure C 63. Brix to acid rati o for late season fruit sampled in the Indian River region. Figure C 64. Total soluble solid percentage for late season fruit sampled in the Indian River region.
279 Figure C 65. Fruit weight for 15 piece sample of late season fruit in the Indian Riv er region. Figure C 66. Juice weight for 15 piece sample of late season fruit in the Indian River region.
280 Figure C 67. Pound solids per box for early fruit sampled in the N orth region. Figure C 68. Acid percentage for early fruit sampled in the N orth region.
281 Figure C 69. Brix to acid ratio for early fruit sampled in the N orth region. Figure C 70. Total soluble solid percentage for early fruit sampled in the N orth region
282 Figure C 71. Fruit weight for 15 piece sample of late sea s on fruit in the N orth region. Figure C 72. Juice weight for 15 piece samp le of late season fruit in the N orth region.
283 Figure C 73. Pound solids per box for la te season fruit sampled in the C entral region. Figure C 74. Acid percentage for la t e season fruit sampled in the C entral region.
284 Figure C 75. Brix to acid ratio for la te season fruit sampled in the C entral region. Figure C 76. Total soluble solid percentage for la te season fruit sampled in the C entral region.
285 Figure C 77. Fruit weight for 15 piece samp le of late season fruit in the C entral region. Figure C 78. Juice weight for 15 piece samp le of late season fruit in the C entral region.
286 Figure C 79. Pound solids per box for la te season fruit sampled in the W est re gion. Figure C 80. Acid percentage for la te season fruit sampled in the W est region.
28 7 Figure C 81. Brix to acid ratio for la te season fruit sampled in the W est region. Figure C 82. Total soluble solid percentage for late season fruit sampled i n the W est region.
288 Figure C 83. Fruit weight for 15 piece samp le of late season fruit in the W est region. Figure C 84. Juice weight for 15 piece samp le of late season fruit in the W est region.
289 Figure C 85. Pound solids per box for late seas o n fruit sampled in the S outh region. Figure C 86. Acid percentage for la te season fruit sampled in the S outh region.
290 Figure C 87. Brix to acid ratio for la te season fruit sampled in the S outh region. Figure C 88. Total soluble solid percent age for la te season fruit sampled in the S outh region.
291 Figure C 89. Fruit weight for 15 piece samp le of late season fruit in the S outh region. Figure C 90. Juice weight for 15 piece samp le of late season fruit in the S outh region.
292 Figu re C 91 Percentage fruit d rop for early f ruit in the Indian River region Figure C 92. Percentage fruit d rop for early fruit in the N orth region.
293 Figure C 93. Percentage fruit d rop for early fruit in the C entral region. Figure C 94. Percentage fruit d rop for early fruit in the W est region.
294 Figure C 95. Percentage fruit d rop for early fruit in the S outh region. Figure C 96. Percentage fruit d rop for mid season f ruit in the Indian River region.
295 Figure C 97. Percentage fruit d rop for m id sea son fruit in the N orth region. Figure C 98. Percentage fruit d rop for mid season fruit in the C entral region.
296 Figure C 99. Percentage fruit d rop for mid season fruit in the W est region. Figure C 100. Percentage fruit d rop for mid season f ruit i n th e S outh region.
297 Figure C 101. Percentage fruit d rop for late f ruit in the Indian River region. Figure C 102. Percentage fruit d rop for late fruit in the N orth region.
298 Figure C 103. Percentage fruit d rop for late fruit in the C entral region. Figure C 104. Percentage fruit d rop for late fruit in the W est region.
299 F igure C 105. Percentage fruit d rop for late fruit in the S outh region.
300 APPENDIX D GAMS PROGRAMMING COD E Industry Model GAMS Code Solver MINOS Option iterlim=1000000; iteration limit increased to 1,000,000 SET I grove aggregate / 1*75 / T week / 1*52/; ------------------------------------------------------*I enumerates grove aggregate assume 75 grove aggregate s of fruit to be harvested, grove aggregate r epresent each fruit variety (E,M,L) in each region (1,2,3,4,5) and *each age group (age1,age2,age3,age4,age5)/ *I = 1=Type10Area1Age1, 2= Type10Area1Age2, 3= Type10Area1Age3, 4= Type10Area1Age4, *5=Type10Area1Age5, 6= Type10Area2Age1, 7= Type10Area2Age2, 71=Type30Area5Age1, 72= Type10Area5Age2, 73= Type10Area5Age3, 74= Type10Area5Age4, *75=Type30Area5Age5 *T is the period harvested, week 1 represents August 1st -------------------------------------------------------PARAMETERS PSPBint(I) interce pt parameter for Pound solids per Box Function /1 3.6788500 2 3.6788500 74 2.2770800 75 2.2770800/ PSPBt(I) t parameter for Pound solids per Box Function /1 0.1889900 2 0.1889900 74 0.2472800 75 0.2472800 / PSPBt2(I) t^2 parameter for Pound solids per Box Function /1 0.0023400 2 0.0023400 74 0.0028200 75 0.0028200 / ; PARAMETER PSPB(I,T) pounds solid per box; PSPB(I,T) = PSPBint(I) + (PSPBt(I) ord(T) ) + (PSPBt2(I) ord(T)* ord(T)); DISPLAY PSPB; estimated function for pounds solids per box over time PARAMETERS TSSint(I) intercept parameter for total soluble solids Function /1 9.35901 2 9.35901 73 6.37326
301 74 6.3732 6 75 6.37326/ TSSt(I) t parameter for total soluble solids Function /1 0.03898 2 0.03898 74 0.23603 75 0.23603 / TSSt2(I) t^2 parameter for total soluble solids Function /1 0.00354 2 0.00354 74 0.0017 75 0.0017 / ; PARAMETER TSS(I,T) total soluble solids; TSS(I,T) = TSSint(I) + (TSSt(I) ord(T)) + (TSSt2(I) ord(T)* ord(T)); estimated function for total soluble solids over time PARAMETERS RATint(I) intercept parameter for Ratio Function /1 2.04416 2 2.04416 74 1.52132 75 1.52132/ RATt(I) t parameter for Ratio Function /1 0.90933 2 0.90933 74 0.23227 75 0.23227 / RATt2(I) t^2 parameter for Ratio Func tion /1 0.01256 2 0.01256 74 0.00298 75 0.00298/ ; PARAMETER RAT(I,T) Brix to acid Ratio; RAT(I,T) = RATint(I) + (RATt(I) ord(T)) + (RATt2(I) ord(T)* ord(T)); estimated function for Brix to acid ratio over time Parameters BPAint(I) Intercept Parameter for Boxes per Acre Function /1 79.6816844276 2 141.5997802295 74 302.9524179087 75 431.6165737123 / BPAt(I) t Parameter for Boxes per Acre Function
302 /1 2.0973453608 2 3.7271255532 74 7.9068641139 75 11.2649161911 / BPAt2(I) t^2 Parameter for Boxes per Acre Function /1 0.0743872237 2 0.1321911629 74 0.1589757508 75 0.2264928907/ BPAt3(I) t^3 Parameter for Box es per Acre Function /1 0.0006466607 2 0.0011491600 74 0.0000739578 75 0.0001053678 / BPAt4(I) t^4 Parameter for Boxes per Acre Function /1 0.0000151069 2 0.0000268460 74 0.0000044008 75 0.00 00062698 / ; PARAMETER BPA(I,T) boxes per acre; BPA(I,T)=BPAint(I)+(BPAt(I)*ord(T))+(BPAt2(I)*ord(T)*ord(T))+ (BPAt3(I)*ord(T)*ord(T)*ord(T))+ (BPAt4(I)*ord(T)*ord(T)*ord(T)*ord(T)); *display BPA; PARAMETER SOLA(I,T) Pound solids per a cre; SOLA(I,T) = PSPB(I,T) BPA(I,T); PARAMETER ACRES(I) bearing acres of each variety in each region /1 1884 2 1703 74 38207 75 8534 / ; ------------------------------------------*From 2004 FASS tree census, Acrage of bearing trees for each type and region ------------------------------------------PARAMETER LB(T) lower bound boxes of fruit required to be delivered each period /1 0 2 0 51 0 52 0/ UB(T) upper bound allo wable boxes of fruit delivered to plant /1 0.00
303 2 0.00 51 0.00 52 0.00 /; ------------------------------------------------------------------------------*UB represents the volume of boxes processed during the 2007 08 seas on This allows for a total of 166,000,423 boxes, weekly max of 6,324,150 boxes max fcoj 3,169,824 max nfc 3,471,732 Lower bound of 50,000 boxes assumes 100 trailers per week must be delivered to the processor from weeks 14 through 48 Upper bound of 10,000,000 boxes allows for 20,000 trailers of fruit to be delivered each week, assuming 500 boxes is equivalent to one trailer load ------------------------------------------------------------------------------PARAMETER ESTPSPBN(T) estim ated pound solids per box of NFC for each time period / 1 5.647075 2 5.647075 51 6.079115 52 6.079115 / *from 2008 processors report 4A ESTPSPBC(T) estimated pound solids per box of FCOJ for each time period / 1 4 .831479 2 4.831479 51 6.250951 52 6.250951 /; *from 2008 processors report 4A POSITIVE VARIABLE X(I,T) harvest of block i in period t; VARIABLE Z Industry Returns ; POSITIVE VARIABLE TB(T) total weekly boxes; POSITIVE VARIABLE TBN(T) total weekly boxes of NFC; POSITIVE VARIABLE TBC(T) total weekly boxes of FCOJ; POSITIVE VARIABLE PSN(T) weekly PS of NFC; POSITIVE VARIABLE PSC(T) weekly PS of FCOJ; *VARIABLE *AVGPSPB(T) average PS per b ox for the week; VARIABLE PSSN(T) weekly PS of NFC in storage; VARIABLE PSSC(T) weekly PS of FCOJ in storage; SCALAR DIP delivered in price per pound solid
304 /1.50/ UPC grove production price per acre /850.00/ h percentage of crop harvested by hand /.95/ /.05/ UHCHH unit harvest cost of hand harvest per box /2.00/ UHCMH unit harvest cost of mechanical harvest per box /1.50/ PMN processing margin per pound solid of NFC /0.35/ PMC processing margin per pound solid of FCOJ /0.15/ UECN unit cost of extraction per box of NFC /0.04/ UECC unit cost of extraction per box of FCOJ /0.05/ USCN weekly cost of storage per pound s olid of NFC /0.0025/ USCC weekly cost of storage per pound solid of FCOJ /0.0004/ TBNCAP total weekly NFC extraction capacity in boxes /3471732/ *2008 weekly Max boxes extracted to nfc from processors report TBCCAP total w eekly FOJC extraction capacity in boxes / 3169824/ *2008 weekly Max boxes extracted to fcoj from processors report PSSNCAP total NFC storage capacity in PS /70427657 / PSSNMIN minimum PS volume required in NFC in storage / 2751 06023/ Max capacity of 315,108,967 GALLONS on hand, @ 11.8 Brix is equal to (315,108,967 1.1029) = 347,533,680 PS *2008 processors report *August 1st average PS @ 11.8 Brix on hand is 275,106,023 *difference is 70427657 PS PSSCCAP total FCOJ sto rage capacity in PS /293652951/ PSSCMIN minimum PS volume required in FCOJ in storage / 537547049/ *This model does not concentrate, therefore the estimated 200,000,000 gallons *of concentrate storage at 42.0 Brix is equvalrnt to about *831,2 00,000 PS = (200,000,000 4.156) *Average gallons in storage on Aug 1st 129,342,408 at 42 Brix *(129,342,408 4.156) = 537,547,049 *difference is 293,652,951 PS TSSLB total soluble solid lower bound /6.5/ RATLB ratio lower bound /9/ /0/ RATUB ratio upper bound /24/ AVGRATLB season average ratio lower bound /12.5/
305 AVGRATUB season average ratio upper bound /19.5/ OUTN total weekly NFC output volume in PS /10290000/ OUTC tot al weekly FOJC output volume in PS /9559000/ ; *NFC output of 10 million gallons at 11.8 brix is equivalent to 10,290,000 PS *FCOJ output of 2.3 million gallons at 42.0 brix is equivalent to 9,559,000 PS *2008 procssors report output, adjusted for n ew inputs to storage EQUATIONS OBJ total industry returns HARVEST(I) harvest of block i DELLB(T) minimum weekly required delivery DELUB(T) maximum weekly allowable delivery BOX(T) total boxes havested in time t BOXSU M(T) total box volume less than NFC and FCOJ boxes TBNMAX(T) total boxes processed to NFC in time t TBCMAX(T) total boxes processed to FCOJ in time t TBNUB(T) maximum weekly boxes to NFC TBCUB(T) maximum weekly boxes to FCOJ PSSNUB(T) maximum NFC storage capacity PSSCUB(T) maximun FCOJ storage capacity PSSNLB(T) minimum NFC in storage PSSCLB(T) minimun FCOJ in storage ONHANDN(T) total NFC pound solids in storage at time t ONHANDC(T) total FCO J pound solids in storage at time t *SOLSOD(I,T) minimum total soluble solid *RATIOLB(I,T) minimum allowed ratio *RATIOUB(I,T) maximum allowed ratio AVRATLB minimum allowed average ratio AVRATUB maximum allowed average ratio PSPN( T) total PS processed to NFC in time t PSPC(T) total PS processed to FCOJ in time t *AVPSPB(T) average PS per box harvested at time t; OBJ.. Z =E= SUM((I,T),DIP SOLA(I,T) ACRES(I)* X(I,T)) SUM((I,T),UPC ACR ES(I)* X(I,T)) SUM(T,UHCHH (h TB(T))) SUM(T,UHCMH ((1 h) TB(T))) + SUM(T,PMN PSN(T)) SUM(T,UECN TBN(T)) SUM(T,USCN PSSN(T)) + SUM(T,PMC PSC(T)) SUM(T,UECC TBC(T)) SUM(T,USCC PSSC(T)) (( PSSNMIN) USCN 52) (( PSSCMIN) USCC 52) ; HARVEST(I).. SUM(T,X(I,T)) =E= 1 ; DELLB(T).. SUM(I,X(I,T) BPA(I,T) ACRES(I)) =G= LB(T) ; DELUB(T).. SUM(I,X(I,T) BPA(I,T) ACRES(I)) =L= UB(T) ; BOX(T).. SUM(I,X(I,T) BPA(I,T) ACRES(I)) =E= TB(T) ; BOXSUM(T).. TB(T) =E= TBN(T) + TBC(T); TBNMAX(T).. TBN(T) =L= TB(T) ;
306 TBCMAX(T).. TBC(T) =L= TB(T) ; TBNUB(T ).. TBN(T) =L= TBNCAP ; TBCUB(T).. TBC(T) =L= TBCCAP ; PSSNUB(T).. PSSN(T) =L= PSSNCAP; PSSCUB(T).. PSSC(T) =L= PSSCCAP; PSSNLB(T).. PSSN(T) =G= PSSNMIN; PSSCLB(T).. PSSC(T) =G= PSSCMIN; ONHANDN(T).. PSSN(T) =E= PSSN(T 1) + (PSN(T) OUTN); ONHANDC(T).. PSSC(T) =E= PSSC(T 1) + (PSC(T) OUTC); *SOLSOD(I,T).. TSS(I,T) =G= TSSLB ; *RATIOLB(I,T).. RATLB =L= RATTAB(I,T) ; *RATIOUB(I,T).. RATUB =G= RATTAB(I,T) ; AVRATLB.. (SUM((I,T),X(I,T) SOLA(I,T) ACRES(I) RAT(I,T))) =G= AVGRATLB (SUM((I,T),X(I,T) SOLA(I,T) ACRES(I))); AVRATUB.. (SUM((I,T),X(I,T) SOLA(I,T) ACRES(I) RAT(I,T))) =L= AVGRATUB (SUM((I,T),X(I,T) SOLA(I,T) ACRES(I))); *AVPSPB(T).. SUM(I,X(I,T) SOLA(I,T) ACRES(I)) =E= (AVGPSPB(T) SUM(I,X(I,T) SOLA(I,T) ACRES(I))); PSPN(T).. PSN(T) =E= ESTPSPBN(T) TBN(T) ; PSPC(T).. PSC(T) =E= ESTPSPBC(T) TBC(T) ; MODEL INDRET /OBJ, HARVEST, DELLB, DELUB, BOX, BOXSUM, TBNMAX, TBCMAX, TBNUB, TBCU B, PSSNUB, PSSCUB, PSSNLB, PSSCLB, ONHANDN, ONHANDC, PSPN, PSPC, *SOLSOD, RATIOLB, RATIOUB, AVRATLB, AVRATUB *AVPSPB/; SOLVE INDRET USING LP MAXIMIZING Z;
307 LIST OF REFERENCES Proceedings from the Florida State Horticultural Society. 119:199 202. mixed International Journal of Production Economics 49: 85 90. Allais, M. 1943. la Recherche d'une discipline conomique Premiere Partie L'Economie Pure. (In Quest of an Economic Discipline Part One Pure Economics ) 1 st ed. Paris: Atelie rs Industria. Bayles, T. 2008 rove Herald Tribune, December 16. Management Science 15:215 27. Horticulture Soc. Annual Meeting, FL, June. Bixby, A., B. Downs and M. Self. 2006. A scheduling and capable to promise application for Swift and Company. Interfaces 36: 131 144 Blanco, H. and F.M. Roka. 2009. Cost/Benefit Analysis of Abscission Registration for Citrus Mechanical Harvesting eeting, Atlanta GA, February. c. Int. Soc. Citriculture. 3:1074 9. Borroto, C.G., V.M. Lopez, A. Gonzalez, and L. Pyla. 1981 71. Brandt J.A. and B.C. French 1983 rvesting and the California Tomato American Journal of Agricultural Economic s 65:265 72. American Soc. Agr. Bio. Engineers Paper 021108 :1 18. HortTechnology 15:69 72. Buker, R.S., J.P. Syvertsen, J.K. Burns, F.M. Roka, W.M. Miller, M. Salyani, and G.K. Brown. HS961, July.
308 Burns, J.K., F.M. Roka, K. Li, L. Pozo, and season Valencia Orange Mechanical Harvesting with an Abscission Agent and Low HortScience 41:660 3. Caixeta Journal of the Operational Research Society 57:637 42. Cembali, T R.J. Folwell, R.G. Huffaker, J.J. McCluskey, and P.R. Wandschneider. 2007. Agricultural Systems 92:266 94. of the Florida State Horticulture Soc. 103:251 5. of the Optimal Number, Size, and Location of Orange Chern W.S. and R.E. Just. 1978 Econometric Analysis of Supply Response and Demand for Processing Tomatoes in California of California, Berkley. Citrus Mechanical Harvesting, University of Florida, IFAS. (cited 28 May 2009). Available from http://citrusmh.ifas.ufl.edu Davies, F.S. and L.G. Albrigo. 1994. Citrus. Wallingford, UK: CAB International. Econometrica 48:1451 61 Durham, C.A., R.J. Sexton and J.H. Song. 1995. Transportation and Marketing Efficiency in the Calif ornia Pr 343, University of California, Berkley. Durham, C.A., R.J. Sexton and J.H. Song. 1996 Transportation Efficiency in the California Processing American Journal of Agricultural Economics 78:115 25. Enerfab Process Systems Group. Citrus Fact Sheet 1 (cited 15 June 2009). Available from http://www.enerfab.com Fairchild, G.F. 1974 imated Production Levels and Harvest Labor Requirements and Costs Citrus Report, June. Fernandes, W. B. 2003 s of the World Processed Orange I ndustry sertation, University of Florida
309 Ferrer, J.C., A. M. Cawley, S. Maturana, S. Toloza, and J. Vera. International Journal of Production Economics 112:985 99. Florida Departmen t of Citrus 2005. Florida Citrus Processors Association 2003 2004 Season Processor Reports Statistical Survey. Florida Department of Citrus. 2007. Florida Citrus Processors Association 2005 2006 Season Processor Reports Statistical Survey. Florida Dep artment of Citrus. 2008a. Citrus Reference Book. Economic and Market Research Department. Lakeland, FL. Florida Department of Citrus. 2008 b Florida Citrus Processors Association 2006 2007 Season Processor Reports Statistical Survey. Florida Departme nt of Citrus 2009. Florida Citrus Processors Association 2007 2008 Season Processor Reports Statistical Survey. 601: The Florida Citrus Code of 1949." Fonsec a, G. New School for Social Research Department of Economics, NewYork, NY. The Paretian System (cited 4 April 2007). Available from http://www.newschool.edu Futch, S.H. and F.M. Roka. 2005 a Gainesville, FL: IFAS Communications, HS239. 4p. Futch, S.H. and F.M. Roka. 2005 b Gainesville, FL: IFAS Communications, HS238. 3p. Goodrich, R. and M. A. Brown. 2001. pean Markets for NFC: Supply and Demand 2, Food a nd Resource Econ. Dep. Un iversity of Florida acceptance of Oranges, Gra American Soc. for Horticultural Science. 49:107 15. USDA Tech. Bul. No. 753. 89p. Harding, P.L. and W.E Horticultural Soc. 54: 52 6. Hicks, J.R. 1939 The Economic Journal 49:696 712.
310 Hillier, F.S. and G. J. Lieberman. 1974. Operations Research 2nd ed. San Francisco CA : Holden Day, Inc. Hydrocarbon Processing 52:71 4 Hopper, R.J. and R.J. Folwell. Acta Hort 479:477 51 Huang S.Y. and R.J. Sexton 1996 Competitive Market: Application to Mechanical Harvesting of P rocessing Tomatoes in American Journal of Agricultural Economics 78:558 71. Hueth, B. and E. Ligon. European R eview of Agricultural Economics 29:237 35. Hueth, B., E. Ligon, S. Wolf, and S. Wu Review of Agricultural Economics 21:374 89. Foo d Processing: Principles and Applications Ames, Iowa: Blackwell Publishing, 361 80. Hyman, B. R. F.M. Roka, and J. Burns. 2005. Evaluating Citrus Mechanical Harvesting Systems University of Florida, IFAS. Intercontinental Exchange. FCOJ A Futures (c ited 28 June 2008). Available from http://www.theice.com Jackson, L.K. 1991. Citrus Growing in Florida 3 rd ed. Gainesville, FL: The University of Florida Press. Jensen M. C. and W.H. Meckling. 1994 The Natur e of Man. The Journal of Applied Corporate Finance 7:4 19. Jhunjhunwala B. 1974 Hicks Southern Economic Journal 40:493 6. Fruit Maturity, Ripening, and Quality Relationships nt. Symp. On the Effect of Pre and Post Harvest Functions on Storage of Fruit. Acta Hort. 485. Kaldor, N. 1939 The Economic Journal 49:549 52. the Florida State Horticultural Society. 97: 40 4. Kreiner, H.W. 1994 Operations Research 42:987 97.
311 Masini, G. hemical Engineering P an A merican C ollaboration Working Paper. IFAS, University of Florida McC arl B.A. and T.H. Spreen. 2007. Applied Mathematical Programming Using Algebraic Systems online book available from http://agecon2.tamu.edu/people/faculty /mccarl bruce/mccspr/t hebook.pd f and Inventory in a Production Management Science 25:799 807. Morris, R.A., C. Erick, and M. Estes. 2009. The Incidence of Greening and Canker Infection in Florida Citrus Groves from September 2007 through August 2008 Univ. of Florid a, EDIS Doc FE 823. Economic Evaluation of Citrus Greening Management and Control Strategies Univ. of Florida, EDIS Doc FE 712 esting of Florida Citrus Extension Program at the Highlands County Extension Service, February 4, 2004. Muraro, R.P. 2006. Estimated Average Picking, Roadsiding and Hauling Charges for Florida Citrus, 2005 06 Season University of Florida, IFAS, Citru s Research and Education Center, Lake Alfred Florida. Muraro, R.P., T. H. Spreen, and M. Pozzan. 2003. Comparative Costs of Growing Citrus in Florida and Sao Paulo (Brazil) for the 2000 01 Season. Univ. of Florida, EDIS Doc FE 364. Murtag h, B.A, M. A. Saunders and P. E. Gill GAMS Solver Manual: MINOS. (cited June 2008). Available from http://www.gams.com Neff, E. 2006 Citrus Industry 87:26 7. Oxbo Interna tional Corporation. (cited 23 December 2009). Available from http://www.oxbocorp.com Palacios, J., R.D. Emerson, and J.J. Vansickle. 2006 Consumer Boycotts and Farm Labor Welfare: A Multi Level Market Analysis Amer Agri Econ Assoc. Annual Meeting, Long Beach, CA July. Pareto, V. 1906. Manuale di Economia Politica Societa Editrice Libraria, Milano, Italy. Translated into English by A.S. Schwier as Manual of Political Economy Macmillan, New York, 1971
312 g Paper, Food and Resource Economics Dept. IFAS, University of Florida. Polopolus, L.C., R.D. Emerson, N. Chunkasut, and R.Chung. 1996. The Florida Citrus Harvest: PrevailingWages, Labor Practices, and Implications. Final Report to the Florida Department of Labor and Employment Security, Division of Labor, Employment and Training, Tallahassee, FL. Ramerez, J.M. 1977. Influence of Date of Harvest on Yields and Leaf Carbohydrates of Citrus PhD Dissertation, University of Florida. Rasmussen, W.D. 1968. Technology and Culture 9:531 43. Roka, F.M. and D. Cook. 1998. Farmworkers in Southwest Florida: Final Report to the Southwest Florida Regional Planning Council S outhwest Florida Regional Planning Council Fort Myers, FL. R oka F.M. and B Hyman 2004. Evaluating Performance of Citrus Mechanical Harvesting Systems 2003/04 Season Report to the Citrus Harvesting Research Advisory Council Lakeland, FL Universit y of Florida, IFAS Southwest Research and Education Center Immokalee FL. Roka, F.M. and R. Emerson. 1999 "Piece Rates, Hourly Wages and Daily Farm Worker Income. Citrus and Vegetable Magazine April:10 2 Rok a, F.M., G.K. Brown, and R.P. Mu rar o. 200 0. ISC Measuring Performance of Citrus Harvesting Systems. December : 280 4 Mechanical Harvesting Field Day & Workshop. UF/IFAS SWFREC, Immokalee, FL. Roka, F.M. and R. 2. of Citrus Mechan 9. Rogers, E.M. 1983. Diffusion of Innovations 3rd ed. New York, NY: The Free Press, Macmillan Co. Hydrocarbon Processing 52:75 8.
313 Rouse, A.H. and Society. 66:268 72. Florida Grower 100: 24 5. Univ. of Florida, EDIS Doc. HR 219. Schmitz, A. and D. Seckler. 1970 the Tomato Harves American Journal of Agricultural Economics 52:569 77. 60. Sinclair, W.B. 1961. The Orange: Its Biochemistry and Physiology. University of California Div. Agr. Sci. Berkeley. HortScience 45:1297 1300. Te tra Pak Processing Systems AB. 1998. The Orange Book. Pyramid Communication s, Lund, Sweden. Thompson, G.L. and S. Thore 1992. Computational economics: Economic modeling with optimization software The Scientific Press, San Francisco, California. Thor Journal of the Operations Research Society of America 1:33 8. U.S. Department of Agriculture. 2005. Commercial Citrus Inventory 2004 NASS, Florida Agricultural Statistics Service, Tallahasse e April. U.S. Department of Agriculture. 2006. Unadjusted Maturity Tests: Average of Regular Bloom Fruit from Sample Groves, 2005 2006 Season for Late Fruit. NASS, Florida Agricultural Statistics Service, April. U.S. Department of Agriculture. 200 7. Citrus Summary 2005 2006 Florida Agricultural Statistics Service, Tallahassee March. U.S. Department of Agriculture. 2008a. Vegetables and Melons: Tomatoes Briefing Room Economic Research Service, Washington DC. U.S. Department of Agriculture 2008 b. World Markets and Trade: Orange Juice. F oreign A griculture S ervice, Washington DC, April. U.S. Department of Agriculture. 2009a. Citrus Summary 2007 200 8 Florida Agricultural Statistics Service, Tallahassee March.
314 U.S. Department of Agri culture. 2009 b Commercial Citrus Inventory 200 8 Florida Agricultural Statistics Service, Tallahassee April. U.S. International Trade Commission. 2006. Certain Orange Juice f rom Brazil Publication 3838. Washington D.C. Ward, R. W and R. L. Kilmer 1989. The Citrus Industry: A Domestic and International Economic Perspective Iowa State University Press Ames, Iowa. Wardowski, W.W., J. Whigham, W. Grierson, and J. Soule. 1995. Quality tests for Florida Citrus University of Florida Cooperative Extension Service, SP99, Gainesville, FL. Walters, L.M., R.D. Emerson, and N. Iwai. 2008. "Proposed Immigration Policy Reform and Farm Labor Market Outcomes." Selected Paper prepared for presentation at the Amer. Ag ri. Econ. Assoc. Annual Meetin g, Orlan do, FL, July acid Ratios of Oranges Utilized by Florida Citrus Processors for the Seasons 1958 59 through 1963 283. Wheaton, T.A 1 Alternat Futch, S.H. and W.J. K ender, e d s. Citru s Flowering and Fruiting Short C ourse Lake Alfred FL: IFAS, Citrus Research and Education Center 87 92. Whitney, Conference, Florida Section, ASME 41:33 59. Zekri, M 23. Zellner, J. Policy Center. Dept. of Agr. and Res. Econ. Research Report No. 17. The University of Connecticut.
315 BIOGRAPHICAL SKETCH Jacob Searcy was born at Alachua General Hospital in Gainesville, Florida He graduated from Vero Beach Elementary in 1986. Jacob enjoys longs walks with his beautiful wife, Jennison. He can often be spotted about the town with his hat wearing son, Isaac, and short, long dogs, Bender and Sugar Pea. Ja cob thinks it is great that you are interested in the m echanical harvesting of citrus and ask s that if you are eating a Florida orange while reading this paper, that you please be careful not to get any juice on the pages. If you are eating a California o range, you may need to get a glass of Florida orange juice to help wash it down.