APPLICATION OF LINEAR PROGRAMMING TO THE EVALUATION OF AGRICULTURAL FLOOD CONTROL , PROJECTS Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY DANIEL GLENN PIPER 1972 -.. .. . A. ...... ._ ,_ . , . J A . I 4| , ..4. I“... I...» I ..... .‘ WW ......‘.I,...,,_.“_., ‘,.,.._ m.“ ..‘..‘I.. 1. . I... . v _ I. I "I’J‘" hi“. “I 4. IV ... -'-‘4'-‘~'.‘.‘ .‘I‘. r... ‘,-'.‘.I:.'.' I ‘.x,'. .I‘. .l m .‘ I ,4. ."i, m ”7.: I... ' H I”. I .L ‘ . ....,:,........, .n..,. H. mm .... .w- ..w um I. .. I, I ww. 'Il. '. . I. I , I .4‘ .“',.. ”w ‘ .‘ ’_ .1”. .H. _I_.‘ _..__,‘ I ~_.... , ,, 3 m. ".‘r""“" ...- .I... H . I I. pan WNW...” , , ‘ ‘ LIBRARY University This is to certify that the thesis entitled Application of Linear Programming to the Evaluation of Agricultural Flood Control Projects presented by Daniel Glenn Piper has been accepted towards fulfillment of the requirements for Ph . D . Agricultural Economics degree in (Mo/flaw JM Major professor y 0-7 639 ABSTRACT APPLICATION OF LINEAR PROGRAMMING TO THE EVALUATION OF AGRICULTURAL FLOOD CONTROL PROJECTS By Daniel Glenn Piper This study investigates the feasibility of estimating agri- cultural flood control benefits using a modified river basin linear programming (LP) projection model. The regional LP basin planning model (RLP-BP) employed in this study was previously used to evaluate the potential for additional drainage, flood control, and irrigation development in the Wabash River Basin. The modification of the Wabash RLP-BP was undertaken in order to evaluate the anticipated effects of six existing and two proposed Corps of Engineers (COB) reservoirs. The resulting LP model, hereafter referred to as the regional linear programming project evaluation model (RLP-PE) was utilized to evaluate the anticipated effects of these projects on the Wabash Basin agricultural economy in 1980. The RLP-PE model has the capability of evaluating on-farm cost savings and land use changes which are anticipated to be induced by the Corps flood control projects. This model was formulated to provide estimates of the cost of producing a specific output from Daniel Glenn Piper the Wabash Basin in 1980 under conditions of present flood hazards and with flood protection afforded by the Corps' reservoirs. The difference between these two estimates provides a single measure of the efficiency gain which reflects both the direct damage reduction and the net enhancement effect to agricultural lands protected by the Corps' reservoirs. The analysis can be repeated for each target year in the river basin survey (e.g., 1980, 2000, and 2020), to ob— tain point estimates of the expected future benefits of the projects. By extrapolating the estimated agricultural production cost savings over the project life and discounting them back to a present value, an estimate is derived of the agricultural crop benefit component of the proposed flood control project. Compared to current Federal Agency procedures for estimating crop flood control project benefits, the RLP-PE model offers three main conceptual benefits. First, this model provides infermation about the effects of a proposed project from a national efficiency point of view. Inelastic demand for farm commodities is assumed; thus, the project's efficiency gains represent savings to the nation by meeting its food and fiber at less cost. Second, the RLP-PE pro- vides a means by which the n§£_enhancement of the project can be estimated. Improving project protected flood plain productivity as a result of flood protection will increase production there and, in the long run will be offset by loss of production elsewhere. The RLP-PE calculates these offsets and thus estimates the net enhance- ment effect. Third, fUture benefits are projected on the basis of separate estimates of changes in the demand for farm commodities and Daniel Glenn Piper changes in technology. This contrasts with the Federal Agency prac- tice of simply projecting future benefits on the basis of some assumed growth rate. An empirical test of the RLP-PE model was conducted. The effect of adding the Big Pine and Lafayette reservoirs to the exist- ing six Corps reservoirs in the Wabash Basin was evaluated. One year (1980) was selected to test the procedures for converting the RLP-BP model to a RLP-PE model. The conversion was successful and estimates were obtained of the efficiency gains and changes in land use patterns that would be expected to result from the projects. APPLICATION OF LINEAR PROGRAMMING TO THE EVALUATION OF AGRICULTURAL FLOOD CONTROL PROJECTS BY Daniel Glenn Piper A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1972 ‘v 61" ACKNOWLEDGMENTS The author would like to express sincere appreciation to Dr. A. Allan Schmid who served as chairman of both the guidance and thesis committees. His pertinent questions, suggestions, and con- structive criticisms have improved the manuscript significantly. Comments by other committee members were also helpful, that is, by Drs. Larry Connor, Lawrence Libby, and Gail Updegraff. The author would like to express appreciation to Dr. Roger W. Strohbehn who provided overall guidance to the initial phase of the study while Leader, North Central Resource Group, Economic Research Service (ERS). The special assistance of Robert F. Boxley, Jr., a co-worker during the early period, is also acknowledged. Carmen Sandretto, ERS leader in the Wabash River Basin study, was particularly helpful in modifying the basin planning model data and in interpreting the computer results. A debt of gratitude is due Dr. Melvin Cotner, Director of the Natural Resource Economics Division (NRED), ERS, and William A. Green, Assistant Director for Water Planning Activities, who pro- vided administrative and technical support as well as financial arrangements fer the study. ii Within the Corps of Engineers, the suggestions and comments of Nathan Back, formerly Chief, Institute for Water Resources, and his staff are appreciated. Russell Whistler and his staff in the Louisville District Corps office were helpful in providing the flood loss and project data. The author thanks John Hostetler for diverting the resources of his staff to allow completion of the final phase of the under— taking. Mrs. Priscilla PrOphet assisted in all computer prOgramming done at Michigan State University. Waldon Miller aided greatly in providing the author with good background information regarding the basis fbr the Wabash Basin data used in the study. The author thanks Miss Linda Farrell and Mrs. Joan Roby for their assistance in typing drafts of the manuscript. These people, as well as the author, are on the staff of the NRED, ERS, USDA at East Lansing. Finally, the encouragement and understanding of my wife, Judith Ann, and children Mark, Mary Esther, and Rosann are deeply appreciated. Without their support and sacrifice, this work would not have been completed. iii TABLE OF CONTENTS Page- LIST OF TABLES . . . . . . . . . . . . . . . . viii LIST OF FIGURES . . . . . . . . . . . . . . . . xii Chapter I. INTRODUCTION . . . . . . . . . . . . . . l The Problem . . . . . . . . . 3 Objectives and Sc0pe of the Study Objectives . 3 Research Background and Area of Study 5 General Characteristics, wabash River Basin 9 Selection of Study Area 9 II. ATTRIBUTES OF BENEFIT EVALUATION METHODOLOGY . . . . . . . . . . . . . . 12 Flood Plain Resource Productivity . . . . . . . 12 Viewpoint Taken . . . . . . . . . . . 13 Projection of Future Benefits . . . . . . . . l4 Adverse Effects of Proposed Projects . . . . . . 14 Benefit Projections and Alternative Models . . . . . . . . . . . 15 Objectives of Water Resource DevelOpment . . . . . . . . . . . . 21 Output Valuation Issues . . . . . . . . . . 23 Demand Conditions in U.S. Agriculture . . . . . 28 The RLP Model Approach . . . . . . . . . . 29 iv Chapter Page 111. OBJECTIVE OF THE RESEARCH INQUIRY . . . . . . . 32 Introduction . . . . . . . . . 32 Selection and Features of Linear PrOgramming . . . . . . . 33 Rationale for Using the Wabash RLP Model . . . . . . . . . . . . . . . 35 IV. FEATURES OF THE WABASH BASIN RLP MODEL . . . . . 39 Determination of Regional Commodity Demand Levels . . . . . . . . . 39 Sensitivity of the ERS Model's Assumptions . . . . . . . . 41 Sensitivity of Water Construction Agency Models . . . . . . . . . . . 42 Land Resource Availability . . . . . . . . . 43 Production Costs . . . . . . . . . . . . 4S Projected Technology . . . . . . . . . 47 Constraints Built into the Model . . . . . . . 48 V. MAJOR ISSUES AND PROBLEMS IN MODIFYING THE WABASH RLP MODEL FOR PROJECT EVALUATION . . . . . . . 51 Obtaining Land Resource Data fbr Project Flood Plain Areas . . . . . . . . . 52 The Conservation Needs Inventory . . . . . . . 52 Soil Survey Reports . . . . . . . . . . 55 Project Justification Studies . . . . . . . . S6 Estimating Flood Plain Crop and Pasture Yields . . . . . . . . 58 Comparison of Upstream and Downstream Flooding . . . . . . . . . 59 Partial Flood Control Protection . . . . . . . 64 VI. MODIFICATION OF WABASH RLP MODEL . . . . . . . 71 Introduction . . . . . 71 Modifications to the RLP Planning Model . . . . 72 Identification of Project-Affected Soil Groups . . . . . . 72 Yield Estimates for Project- -Affected Flood Plain Lands . . . . 74 Cost of Crop Production Estimates for Project~ Affected Flood Plain Lands . . . . . . . 80 Summary of Input Revisions . . . . . . . . 81 Chapter Page VII. EMPIRICAL INVESTIGATIONS . . . . . . . . . . 83 Introduction . . . . . . . . . 83 Low and High Demand Analysis . . . . . . . . 85 Efficiency Gains . . . . . . . . . . . . 87 Changes in Land Use . . . . . . . . . . . 93 Subregional Analysis . . . . . . . . . . . 99 Changes in Land Use . . . . . . . 99 Changes in Total Value of Production . . . . . 102 Effects on Resource Owners . . . . . . . . 104 Comparison of RLP Benefits with Corps Benefits . . . . . 111 Methods of Estimating Enhancement Benefits . . . 114 The Corps Approach . . . . . . . . . . . 114 The Wabash RLP Approach . . . . . . . . . 115 Future Flood Plain Growth . . . . . . . . . 117 The Corps Approach . . . . . . . . . . . 117 The RLP Approach . . . . . . . . . . . 121 RLP Estimated Growth in Benefits for Life of Project . . . . . . . 122 Resolution of Corps and RLP- PE Benefit Differences . . . . . . . . . . . . . 130 Land Use Differences . . . . . . . . . . 130 Crop Yield Differences . . . . 132 Differences in Value of Flood Plain Output . . . 133 Difference in Expected Damage Factors . . . . 134 Estimating the Net Effect of Differences . . . 135 Study Costs . . . . . . . . . . . . . . 141 VIII. SUMMARY, CONCLUSIONS, AND POLICY IMPLICATIONS . . . 144 Summary . . . . . . . . . . . . . . . 144 Limitations . . . . . . . . . . . . . . 151 Introduction . . . . . . . . . . . . . 151 Data Problems . . . . . . . . . . 152 Basic Decision-Making Unit . . . . . . . . 154 Inability to Identify Beneficiaries . . . . . 155 Relevant Uninvestigated Areas . . . . . . . 156 vi Chapter Page Application of Results . . . . . . . . . . 156 Introduction . . . . . . . . . 156 Substitute for Corps Analysis . . . . . . 157 Use of RLP as a Check on Corps Analysis . . . . 157 BIBLIOGRAPHY . . . . . . . . . . . . . . . . 161 APPENDICES Appendix A. General Characteristics, Wabash River Basin . . . . 166 B. Selection of Study Area . . . . . 179 C. Flood Control Plus Optimal Drainage Alternative . . 183 D. Wabash Basin Linear Programming Model . . . . . . 187 vii Table 10. 11. 12. LIST OF TABLES Cropland Withdrawal, Wabash River Basin, 1958—2020 (Index 1958 = 100) . . . . Projected Yield Increases Per Year, wabash River Basin Production Minimums for Selected Crops by Subarea, Wabash River Basin Chi-Square Test of Agreement Between Soil Survey and CNI for Selected Soil Types, Ten Indiana Counties . . . . . . . . . Composite Test of CNI Reliability for Estimating Flood Plain Acreage, Ten Indiana Counties Comparison of Average Annual Crop Losses Upstream vs. Downstream Areas, wabash River Basin . Comparison of Flood Damage Factors by Crop, Corps and Wabash Basin Planning Model Data, Wabash River Basin . . . . . . . . Indices of Flood Protection to Cr0p1and on the wabash River Flood Plain Variations in Crop Loss Estimates Due to Relative Change in Crop Valuation,Hypothetical Reach . Sample Unit Cr0p Damages as Percent of Total Cr0p Value Per Acre, Wabash River Flood Plain . Crop Acreage and Flood Protection Index, by Selected River Reaches, Wabash River Basin . Land Use, Production Cost, and Output Comparisons, Low vs. High Demand Levels, Wabash River Basin, 1980 Projected viii Page 45 48 50 S4 55 60 61 65 69 76 77 86 Table 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. Effect of Flood Control Reservoirs on Agricultural Production Costs, Wabash River Basin, 1980 Projected Demand and Cost Conditions . . On-Farm Agricultural Production Costs, Low Demand Conditions, Wabash River Basin, 1980 Projected . On-Farm Agricultural Production Costs, High Demand Conditions, Wabash River Basin, 1980 Projected . Major Land Use, No Development Situation, Wabash River Basin, 1980 Projected Adjustments in Basin Land Use Due to Corps Reservoirs, Wabash River Basin, 1980 Projected . Land Use Under Alternative Flood Control Condi- tions, Low Versus High Demand Levels, Wabash River Basin, 1980 Projected Agricultural Lands Idled by Introduction of Corps Reservoirs, Wabash River Basin, 1980 Projected . Flood Plain Land Use, Corps Reservoir Impact Area, wabash River Basin, 1980 Projected . Changes in Land Use Due to Flood Control Projects, Wabash River Basin, 1980 Regional Shifts in Total Values of Crop Production Induced by Corps Flood Projects, High Demand Conditions, Wabash River Basin, 1980 Comparative Farm Size, Wabash Flood Plain Survey with Comparisons . Cost and Return Coefficients, Flood Plain Soil Groups, Wabash River Flood Plain, Per Acre Basis, Subarea 5, 1980 . Cost and Return Analysis, Flood Plain Soil Groups, Wabash River Flood Plain, Subarea S, 1980 Impact of Flood Control Projects on Non-Project Farmers, Subarea 5, 1980 ix Page 88 91 92 94 96 97 98 99 101 103 106 108 109 111 Table 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. Estimation of Annual Flood Control Benefits Under Alternative Benefit Evaluation Procedures, Big Pine-Lafayette Project, 1980 Growth Indices for Selected Economic Indicators-- U.S. 1957-1968 . Projection of 1980 Agricultural Crop Benefit Proportion to 2069, Big Pine- Lafayette Project, wabash River Basin . . . . Computation of Average Annual Benefits for Big Pine-Lafayette Project, Low Demand Conditions, RLP-PE Data, 1969-2069 Computation of Average Benefits for Big Pine- Lafayette Project, High Demand Conditions, 1969— 2069 . Benefit Cost Evaluation of Agricultural Crop Benefits Under Alternative Benefit Formulations, Big Pine-Lafayette Project . Percentage Share of Corn and Soybeans, Corps and RLP Projections, Big Pine-Lafayette Flood Plain, 1980 . . . . Yield Comparisons, Corps and RLP-PE Yields, Big Pine-Lafayette Reservoirs, 1980 Projected Effect of Alternative Land Use, Yield, and Product Value Assumptions on Total Value of Flood Plain Output for 1980 Wabash River Below Big Pine- Lafayette Reservoirs . . . . Derivation of Differences Between Corps and RLP-PE Total Value of Flood Plain Output for 1980, wabash River Below Big Pine-Lafayette Reservoirs . . . Summary of Differences Between Corps and RLP-PE Total Value of Flood Plain Output for 1980, Wabash River Below Big Pine-Lafayette Reservoirs . . . . . . . Page 113 120 125 126 127 128 132 133 136 138 139 Table 39. 40. A-l. A-3. B-l. C-1. C-2. Summary of Differences Between RLP—PE and Corps Benefit Estimates for 1980, Big Pine-Lafayette Project . . . . . . Breakdown of Costs Associated with Estimating Project Benefits Using RLP Model, September, 1967-December, 1969 . . Breakdown of Costs Associated with a Future RLP Application Over One Year Period . Acreage of Principal Basin Cr0ps Grown Wabash River Flood Plain Characteristics Average Annual Flood Losses, Wabash River, 1960 and 1963 Values . Crop Acreage and Flood Protection Index, by Selected River Reaches, wabash River Basin Effect of Flood Control plus Optimal Drainage Alternatives on Agricultural Production Costs, Wabash River Basin, 1980 Adjustments in Land Use due to Big Pine-Lafayette Reservoirs under Flood Control plus Drainage Assumption, Wabash River Basin, 1980 xi Page 140 142 143 168 171 172 181 184 186 Figure A-2. A-3. LIST OF FIGURES Wabash River Basin and Economic Sub-areas Effects of Elasticity Conditions on Output Valuation . Wabash River Basin . Land Resource Regions and Major Land Resource Areas, Wabash River Basin . Physiographic Map, Wabash River Basin xii Page 11 25 176 177 178 CHAPTER I INTRODUCTION The Problem The Flood Control Act of 1936 set forth the criterion for evaluation of proposed public expenditures in water resource invest- ments. In this Act, the Specification was made that benefits must exceed costs for a flood control project ”to whomsoever they may accrue" in order for a project to have economic feasibility and thus to be eligible for authorization by Congress. The interpretation of this Act by the Army Corps of Engineers and other public agencies responsible for construction of flood control projects has been that a national economic efficiency analysis of each proposed project be conducted. In order to receive favorable consideration by Congress, the present value of the expected benefit stream must exceed the present value of the expected cost stream. If the net present value is positive, the project is considered efficient and it is assumed that national income will increase if the project is undertaken. Federal investment in flood protection and prevention has amounted to more than $7 billion since the Flood Control Act of 1936. The current Federal expenditures are approximately $500 million per year and the annual rate of investment is increasing.1 Agricultural flood control benefits have been and will continue to be important in the benefit-cost justification analysis by federal agencies. The assumption made for individual project evaluation is that flood protection afforded to the small impact is too small to have any appreciable effect on product prices. Although this assump- tion may be reasonable for any single project taken separately, it cannot be expected to be valid for the summation of all projects in a region or nation. Three important characteristics of U.S. agriculture seem to warrant a departure from federal agencies use of completely elastic demand (assumption of no price effect as supply increases because of the project). First, the various agricultural regions of the United States compete with one another in a common market characterized by very inelastic demands. Second, the various regions produce cr0ps in common, however, the regions differ in comparative advantage due to differences in soil, climate, and extent to which man-made produc- tivity investments have been made in altering natural production characteristics. Examples of the latter include drainage, irrigation, and flood control. Third, the U.S. agricultural economy is charac- terized by excess capacity. Acreage allotments and other restrictions 1U.S., Congress, House, Task Force on Federal Flood Control Policy, A Unified National Program for Managing Flood Losses, H. Doc. 465, 89th Cong., 2nd sess., 1966, p. 3. are necessary in order to prevent even larger price-depressing surpluses. Under these conditions, the issue arises as to whether the public investment in flood control can be justified relative to other uses of public funds. Will further development of flood plain lands have a positive benefit to society? One must also evaluate the dis- tributional consequences of flood control investment. Will the crea- tion of more highly productive flood plain land in one region create offsetting income losses in other parts of the region or in other regions? Clearly an empirical procedure for evaluating some of these questions is needed in order to resolve the above issues. Objectives and Scope of the Study Objectives The primary purpose of this investigation is to examine and test the feasibility of utilizing a regional linear programming (RLP) model of the agricultural sector, as developed for river basin plan- ning, as an analytical device to estimate agricultural benefits of specific flood control projects proposed by the Army Corps of Engi- neers. A secondary objective is to assess the distributional effects of flood control projects, identifying the gainers and losers, if any, from this public investment. The Corps of Engineers and the other federal agencies responsi- ble for evaluating water resource investments have employed economic analysis for years. This economic analysis, however, has been con- fined to the single objective of maximizing gross national product. This approach has been faulted by many economists because it fails to consider the distributional effects on other parts of the system, whether the system be the region or the nation as a whole. Bromley, Schmid, and Lord1 recently pointed out the fallacy of relying exclu- sively on the project-by-project analysis conducted by the Corps and other water resource agencies: The sum total of many small decisions which society makes may produce results which carry great significance for the quality of all of our lives. Often such "system effects" go unrecog- nized . . . For example, continued reservoir development for hydropower production, flood control, navigation, water supply, and . . . recreation may seem quite justifiable when examined on a piecemeal project-by-project basis.2 In their work, Bromley, Schmid, and Lord imply that a more systematic approach to evaluation of flood control investment is a worthwhile undertaking. This view is shared by Knetsch3 in a 1969 paper: Analysis--program planning, budgeting, and benefit-cost-is essentially an aid in determining efficient allocation and in- vestment of public funds. Comparison of alternative means of achieving given ends is an integral part of such analysis. . In many cases the principals for determining gains and losses have not been correct nor defined in as meaningful terms as might be possible.4 1Daniel w. Bromley, A. Allan Schmid, and William B. Lord, Public Water Resource Project Planninggand Evaluation: Impacts, Incidence, and Institutions (Madison, Wisc.: Center for Resource Policy Studies and Programs, 1971). 2Ibid., p. 2. 3Jack L. Knetsch, "Economic Analysis in Natural Resource Programs," The Analysis and Evaluation of Public Expenditures: The PPB System, Vol. III (Washington, D.C.: U.S. Government Printing Office, Joint Economic Committee, 1969). 41bid., p. 1096. The RLP model which will be evaluated in this application will offer the advantage of systematically evaluating the regional impacts of two flood control projects. In this study, the RLP model will not evaluate the effect of alternative public or private water resource investments. Nor will other possible ways of achieving the same regional agricultural output be evaluated. For example, it is possible to evaluate the effect of utilizing additional fertilizer in the Basin to achieve the same level of output at perhaps lower societal costs than in the case of flood control investment. Thus, the rather limited objective of this application will be to access the RLP as an alternative framework for agricultural flood control benefit estimation. The RLP possesses the advantage of being a systems analysis, with system being defined in the sense that it is 1) a portion of the universe around which is drawn an imaginary boundary for purpose of studying what is enclosed with this boundary; and 2) a medium which relates a cause to an effect, or an input to an output.1 A further objective of the study is that the theoretical and empirical advantages associated with the application of an RLP model for evaluating consequences of flood control be documented for the benefit of researchers applying other systems models to water resource planning in the future. Research Background and Area of Study This research was conducted under contract between the Corps of Engineers, U.S. Department of the Army (COE), and the Natural 1John A. Dracup, "Systems Analysis, A Substitute for Planning," Evaluation Processes in Water Resources Planning (Urbana, 111.: American Water Resources Association, 1970). Resource Economic Division (NRED), Economic Research Service (ERS), U.S. Department of Agriculture as a part of a larger study of alterna- tive procedures for the evaluation of agricultural flood control benefits.1 The Wabash River Basin was selected for empirical analysis for administrative convenience and because of considerations involved in other parts of the overall study. The Wabash RLP model was employed in the on-going Wabash River Basin Type II comprehensive study. The model is a relatively large minimum-cost linear programming analytical model, having a matrix that is about 2,000 rows by 10,000 columns. Since the available core storage at the Michigan State University was insufficient at the outset of the Wabash Type II study in 1966, the decision was made by NRED to utilize the McDonnell Corporation IBM 360-75 computer at St. Louis, Missouri. Three analytical techniques have been utilized by the NRED in conducting river basin studies. The first major study, the Texas River Basins completed in 1961, did not use a formal mathematical model but instead relied on trend analysis and crop budgeting techniques. The dominant analytical technique, the least cost linear programming model (LCLP) has been utilized by NRED in seven of the thirteen Type I 1Robert F. Boxley, Jr., "The Relationship Between Land Values and Flood Risk in the Wabash River Basin" (unpublished Ph.D. disserta- tion, Michigan State University, 1969). Institute for Water Resources, Corps of Engineers, Department of the Army, "The Relationship Between Land Values and Flood Risk in the Wabash River Basin," IWR Report 69-4, Alexandria, Virginia, December, 1969. framework studies.1 Input-output analysis was used in two Type I studies and simple trend analysis was used in the remaining four Type I studies. The use of simulation models in river basin analysis was also explored for NRED by Battelle Memorial Institute of Columbus, Ohio. The use of simulation was rejected by NRED due to the finding of estimated costs of $500,000 to deve10p a workable river basin model.2 Although input-output and trend analysis continue to be used by NRED in river basin projections, the LCLP remains the dominant analytical technique. Six of the sixteen Type II studies3 conducted by NRED utilized the LCLP. The author made a survey of twenty-nine NRED Type IV studies4 and found the LCLP being utilized in eighteen of these completed or ongoing efforts. 1According to USDA Soil Conservation Service Memorandum-9 (Rev. 2), Type I studies are defined as "comprehensive framework studies which will furnish a general appraisal of overall water and related land resource development needs and serve as a guide to further detailed planning within the Regions." 2John E. Hostetler, "Sensitivity Analysis of Selected Linear Programming Assumptions: A Study of the Agricultural Projections in River Basin Research” (unpublished Ph.D. dissertation, Michigan State University, 1970), pp. 5-6. 3According to SCS River Basin Memorandum-9 (Rev. 2), June 13, 1968, Type II surveys "are studies of a river basin or other area in greater detail then Type I studies. They define and evaluate projects in sufficient detail to comprise a basis for authorization or imple- mentation of those Federal and federally assisted projects to be initiated in the next 10 to 15 years." 4According to SCS River Basin Memorandum-9 (Rev. 2), Type IV studies "usually are State sponsored surveys of water and related land resources for all or part of a State or a river basin in which one or more Federal agencies cooperate with the State or each other." The nature of the LCLP utilized in NRED river basin studies has continued to evolve over time from relatively small-sized models with few activities to large scale, highly constrained, multiple activity models. This is due to the advent of improvements in both computer hardware (physical facilities) and software (computer pro- gramming techniques). A typical example of an early LCLP model, the Grand River Basin Type II study had a matrix of 212 rows by 554 vectors. This was about the maximum size matrix which could be run on the computer at Michigan State University in the mid-1960's. Late LCLP models such as the Upper Mississippi, Missouri, and Wabash studies had considerably larger matrices. The large LCLP models enabled the inclusion of more detailed soils information as well as the simultaneous consideration of more resource activities. At the same time, there also appeared two different approaches to basin pro- jections using the more complex LCLP models. The North Central Re- source Group in East Lansing continued in the pattern established in the Ohio River Basin Type I study. This approach minimized the number of constraints to economic efficiency in the agricultural economy in future target years. Hence, their projections were based primarily on the assumption that major shifts in water and related land use were possible, depending upon the comparative cost advantages of the various agricultural lands. By contrast, the Great Plains Group in Lincoln, Nebraska, carefully constrained their LCLP to allow only relatively small deviations from the base period situation (1964). They made the assumption that future agricultural production patterns will likely reflect relatively minor deviations from current patterns. General Characteristics, Wabash River Basin The Wabash River Basin comprises the southern two-thirds of Indiana, southeastern one-sixth of Illinois, and 319 square miles of two Ohio counties. The Wabash River is the second largest tributary of the Ohio River. The valley of the Wabash River is flat and wide, and its flood plain is a highly agriculturalized area. The result is that floods are frequent and destructive. The principal cause of floods on the river is excessive rainfall. The Wabash Basin has been surveyed by the federal government in 1932, 1944, and in the 1960's, the latest comprehensive study drawing to a close in June 1971. Six federally-financed Corps reservoirs in conjunction with over 145 named public and private levees provide a moderate level of protection against the frequent, fairly small floods. Although ten additional Corps reservoirs have been authorized, to date the Patoka reservoir is the only one currently under construction. (See Appendix A for a more complete discussion of Wabash Basin general characteris- tics.) Selection of Study Area In selecting the Wabash Basin projects to evaluate in the empirical application, four primary criteria were considered: (1) data availability, i.e., existence of recent usable hydrological and flood damage data; (2) project impact area sufficiently large; (3) signifi- cant project crop_loss reduction; and (4) land conversion possibilities. 10 After evaluating project evaluation data at the Louisville District Corps office, the author, in conjunction with Dr. Strohbehn and Dr. Boxley,selected the Big Pine and Lafayette reservoirs. These authorized but not yet constructed multiple-purpose reservoirs were found to best satisfy all four selection criteria. (See Appendix B for a more complete discussion of the selection of the study area.) Wabash River Basin and Economic Sub-areas .. l "(9 _ - Big Pine Reservoir - Cagles Mills Reservoir - Huntington Reservoir - Lafayette Reservoir - Mansfield Reservoir - Mississinewa Reservoir - Monroe Reservoir - Salamonie Reservoir Iofimor‘tw) CHAPTER II ATTRIBUTES OF BENEFIT EVALUATION METHODOLOGY In an earlier report of the ERS to the Corps regarding the flood control benefit evaluation study1 four principal attributes to consider in an ideal flood control benefit evaluation procedure were presented. The attributes included: (1) effect on the productivity of flood plain resources; (2) the viewpoint taken--resource owner's or society's; (3) projection of future benefits; and (4) adverse effects (if any) of the pr0posed projects. A discussion of the relevance of this list of attributes follows. Flood Plain Resource Productivity_ The productivity or returns to flood plain resources can be increased by any action that shifts the supply curve downward and/or to the right. Flood control may reduce the costs of inputs used in agricultural production, e.g., reduction in soil preparation and re- planting costs. Increased productivity of inputs used on flood plain lands may result by reducing or eliminating direct loss of agricultural 1Institute for Water Resources, Department of the Army, Corps of Engineers, IWR Report 71-4, July, 1971. 12 13 output, and by eliminating yield reductions due to delayed planting. Flood control may also contribute to more efficient utilization of flood plain land either in its present use by allowing more intensive production practices to be adopted (e.g., heavier fertilizer applica— tions) or by enabling land use shifts to higher value craps or to commercial, industrial, or residential uses. The first two effects-- reducing input costs and reducing yield losses--are considered direct damage reduction. Benefits arising from more efficient utilization or from shifts to higher value uses are considered enhancement benefits. Viegpoint Taken Flood control benefits can be evaluated from two different viewpoints. First, from the viewpoint of what the flood plain occu— pants would suffer due to flooding or what they would receive as flood protection benefits; and second, from the viewpoint of society as a whole. It should be pointed out that distinguishing between these two viewpoints is an important one. This is due to the fact that societal benefits may not necessarily equal the sum of benefits obtained by individual flood plain occupants. An individual may benefit greatly from a flood control project yet society as a whole may be no better off. This is particularly true when the demand for the additional output produced on a newly-protected flood plain is inelastic. In the case of most agricultural commodities, increased net returns to flood plain occupants may be offset either by equivalent reductions in pro- duction elsewhere or by increased costs of price support and production control programs. 14 Prpjection of Future Benefits In addition to determining estimated benefits under current conditions, an ideal benefit estimation procedure would evaluate whether benefits can be expected to increase or decrease over time. A sound analysis will require projection of future rates of flood plain development that are likely to occur in the absence of flood protection. Under existing agency procedures, the Corps of Engineers projects the prospective "normal" state of development without the project in order to provide a basis for modifying the basic current average annual damages to determine "prospective average annual damages over the life of the project."1 This procedure provides an estimate of future growth that is expected to occur in the absence of the project. This is in addition to enhancement benefits which are realized only with the project. In the Wabash Basin Corps Interim Reports, future growth was estimated over a 100-year period. Adverse Effects of Proposed Projects An ideal benefit evaluation framework should also consider the negative effects that may result from the installation of a flood control project. For example, the release of flood waters from a re- servoir may cause streams to have bankful conditions for prolonged 1U.S. Army, Corps of Engineers, "Survey Investigations and Reports: General Procedures," Engineering Manual EM 1120—2-101 (includes change 16), October 12, 1964, pp. 50—50b. (Mimeographed.) 15 periods, resulting in impaired drainage to adjoining flood plain land.1 Let us now turn to a brief discussion of some of the problems associated with the projection of future benefits and to a considera- tion of an alternative analytical framework for making projections. Benefit Projections and Alternative Models In his 1965 book, Ruttan1 provides not only a new methodology for making agricultural projections, but also a critique of current projections methodology. Ruttan correctly points out that most pro- jections of resource use,inc1uding those of USDA, are made on the basis of the quantity of inputs required to support some projected level of final output. Terming these as ”requirements approaches," Ruttan maintained that they are inflexible procedures which fail to consider 1The Patoka Reservoir, a partially-constructed Corps of Engi- neers multiple-purpose reservoir in southern Indiana will have this type of adverse effects. Due to a large amount of reservoir storage capacity and limited channel capacity because of a narrow and crooked channel downstream, it will take 90 days to deplete the flood waters of a lS-year recurrence interval storm if only bankful conditions are to be observed. The recreation benefits are ten times those from flood protection during the summer season. Since the trade—off in benefits favors recreation, a faster drawdown will be used by the Corps so that the seasonal (recreational) pool can be obtained as quickly as possible. The Corps in this case will buy more easement rights from flood plain land owners so that the scheme of reservoir operation can proceed in the way which will minimize drawdown time. (Based on a May 8, 1968, telephone conversation with Russell Whistler, Basin Planning Branch, Louisville District Corps office.) 2Vernon W. Ruttan, The Economic Demand for Irrigated Acreage: New Methodology and Some Preliminary Projections, 1954-80 (Baltimore: Johns Hopkins Press, 1965). 16 the capacity for resource and product substitution in the U.S. agri- cultural economy. In making his case, Ruttan states that The elasticity with which the regional distribution of farm out- put and factor input combinations respond to economic forces leads to considerable dissatisfaction with the requirements framework employed in most of the regional resource use projec- tions of the decade and a half . . . the use of the requirements framework, in its simplest form, involves the implicit assump- tion that resource combinations and consumption patterns are techn010gically, institutionally or psychologically determined and are inelastic with respect to changes in the prices of resource inputs relative to each other, consumption items relative to each other, or resource inputs relative to con- sumption items.1 Ruttan further states that the impact of altering the distri- bution of agricultural activity can be modified by public investment in one region but not in another region: The effects of expansion or contraction in demand or of changes in comparative advantage in production in one agri- cultural region are transmitted to other regions. Since the RLP is based on the ”requirements approach" frame- work, it too is subject to the criticisms of Ruttan. Let us now com- pare Ruttan's suggested methodological improvements using the Cobb- Douglas formulation with the RLP approach. In an effort to overcome the limitations of the requirements approach, Ruttan suggests that an ideal model consist of the following: (1) factor supply functions for all resources and other inputs, (2) transportation rates or cost functions for inputs and final prod- ucts, (3) the geographic location of inputs and markets, (4) production functions relating input levels to output levels for each product, and (5) demand functions for each product. 1Ibid., pp. 16-17. 2Ibid., p. 16. 17 Ruttan concedes that more sophisticated forms of the "require- ments" approach (e.g., linear programming) may overcome many of the limitations. On the other hand, he points out that their Almost insatiable appetite for data becomes a problem so that such applications have, in the past largely been confined to relatively few activities and to limited geographic areas. Empirical studies which have attempted to deal with a complex set of activities for major geographic areas have typically employed what might be termed "aggregate budgeting” rather than the more SOphisticated programming techniques. Ruttan also admits that his "elasticity" approach is "limited by an opposite deficiency [in that] models currently developed have not been able to absorb as much data as many analysts would like to incorporate into their models." Ruttan refers to the work of Zvi Griliches2 in saying that for these elasticity models Statistical models which include large numbers of variables tend to be limited by lack of significance of the elasticity coefficient; those which include few variables, by biased elasticity coefficients.3 (Italics mine.) Ruttan presented three elasticity models. His productivity model "is developed to permit a comparison of current resource produc- tivity and cost levels."4 His demand and equilibrium models are "developed to facilitate projection of future farm output growth and factor input levels."5 In distinguishing between his last two models, Ruttan notes that 1Ibid., pp. 18-19. 2Zvi Griliches, "Specification Bias in Estimates of Production Functions," Journal of Farm Economics, XXXIX (February, 1957), 8-20. 3Ibid., p. 19. 4lbid. SIbid. 18 The basic distinction lies in the determination of output growth in each region. In the demand model the regional output levels are determined from outside the system, while in the equilibrium model they are determined simultaneously along with factor input levels. Regarding usefulness in making projections, only Ruttan's last two models appear to be relevant in comparison with the RLP model. The Ruttan demand model contains five equations. It is made operational by the inclusion of a hypothetical supply function for irrigated land and of a procedure for determining regional production requirements when given a national production requirement. By de- riving a regional output requirement, the demand model can be solved to obtain the number of acres of irrigated land in each region which equates the annual marginal value productivity of irrigated land with a specified annual rate of return which should equal the annual cost of bringing an acre of irrigated land into production and producing a crap on it in each region. When this happens, interregional equilib- rium is attained. The demand model contains some serious deficiencies. It assumes, in common with the "requirements approach," that regional demand in a period is inelastic with respect to price. The second deficiency is that it embodies the concept of a perfectly elastic supply of homogeneous irrigation land. The third shortcoming is that it contains only a historical production function. Fourth, the model traces out a "demand curve" by setting the acreage of irrigated land at alternative levels and solving for the marginal value productivity of irrigated land consistent with this specified output level. This 1Ibid. 19 "demand curve" is unusual because changes in output in response to changes in the price of irrigated land are not permitted. This is not only unrealistic, but it seems that this is yet another requirement model with a pre-ordained inelastic demand for output. The equilibrium model differs from Ruttan's previous two in that the implicit assumption that current Operating expenses are a costless input is replaced by the explicit inclusion of a supply func- tion for that mix of inputs. In addition, regional output requirements are no longer derived from an exogenously determined national require- ment but are generated as an Optimum by the model itself. Hence, it is the only Ruttan model not employing the requirements approach. Un- fortunately it is also the one in which Ruttan has the least faith as a projection.1 Let us now summarize some of the deficiencies of the Ruttan models: 1. None of the models improve on the "requirements" assumption of inelastic regional demand functions. All demand functions, both national and regional, are assumed to be price inelastic. 2. The interregional competition problem is ignored (as does the RLP). 3. The production function is still static and historically based (as in the RLP).2 1lbid., p. 25. 2lbid., pp. 78-79. 20 4. The supply function for land is both perfectly elastic and unconstrained. Even if the available irrigation land is assumed to be completely homogeneous within a region, there will still be a finite and possibly limiting supply of it. In conclusion, Ruttan's study provides an excellent critique of previous attempts at projection in the water and land use field. His models represent an effort to overcome many of the objections found in linear programming. In the end, however, the author concludes that Ruttan's models do not compare favorably with mathematical models such as the RLP which have the capability of bringing more information to bear upon the problem of determining the projected need for water resource investment. Given that a new generation of computers and software have evolved since Ruttan's 1965 analysis, his objections to LP seem far less powerful than previously. From the standpoint of the planners a rational LP model, although complex, is probably easier to conceptualize than Ruttan's three models. The LP models have the further advantage of evaluating many resource development alternatives simultaneously, a feature notably lacking in Ruttan's models. In this regard, the advantages of the RLP model will be dis- cussed in the next chapter in the section entitled "Selection and Features of Linear Programming." 21 Objectives of Water Resource DevelOpment In ERS's report to the Corps,1 but not reported in the final Corps report, IWR 69-4,2 Boxley argues that "benefits derived from providing flood protection to agricultural lands are efficiency gains.” . 3 . . . ga1ns." Boxley reasons that the Green Book4 1mp11es that the eff1- ciency criteria is the underlying rationale for water resource invest- ments: The objective of economic analysis in planning river basin and watershed programs is to provide a guide for effective use of the required economic resources, such as land, labor, and materials, in producing goods and services to satisfy human wants by determining whether economic resources would be the case without the project.5 Boxley asserts that an efficiency gain is realized ”if, as the results of water resource deve10pment, either more goods or serv- ices are obtained with the same resources, or the same goods and services are obtained from fewer resources than would be required in 1U.S., Department of Agriculture, Economic Research Service, Analysis of Alternative Procedures for the Evaluation of Agricultural Flood Control Benefits, Vol. I, edited by R. F. Boxley, Jr. (Wash- ington, D.C.: Government Printing Office, August, 1969), 201 pp. 2Institute for Water Resources, IWR Report 69-4, December, 1969. 3U.S., Department of Agriculture, Flood Control Benefits, 4Subcommittee on Evaluation Standards, Report to the Inter- agency Committee on Water Resources, Pr0posed Practices for Economic Analysis of River Basin Projects (Washington, D.C.: Government Printing Office, May, 1958). SIbid., p. s. 22 the absence of deve10pment.”1 To this he added the third case that efficiency gains also result if: A greater level of output is obtained with fewer productive re- sources than would have been required in the absence of protection.2 As noted in the previous section on "attributes of benefit evaluation," two viewpoints from which to evaluate are indicated. First, the national viewpoint or the viewpoint of society as a whole. Second, the vieWpoint of a specific group directly or indirectly af- fected by a public water resource investment, for example, flood plain landowners. Boxley argues that "there seems little doubt that the appro- priate viewpoint should first be the national viewpoint [since] . this is apparent in the many charges to compute 'benefits to whomso- ever they may accrue' that can be found in many government documents, as this passage from the Green Book illustrates"3 A summation of project effects, beneficial or adverse, to whom- soever they may accrue, in terms of market values would approach full coverage from a public viewpoint if allowance could be made in the summation for all transferences, cancellations, and offsets; i.e., values that are realized by one individual or group at the expense of some other individual or group.4 Let us now turn to the effect of adopting efficiency gains in conjunction with taking the viewpoint of society as a whole (national viewpoint). 1U.S., Department of Agriculture, Flood Control Benefits, Vol. I, p. 6. 2Ibid., p. 8. 3Ibid., p. 7. Subcommittee on Evaluation Standards, River Basin Projects, May, 1958, p. 6. 23 Oupput Valuation Issues The key issue regarding output valuation turns around the question of whether flood control projects will displace other firms' outputs or whether output will increase. If the former, we need not concern ourselves with constructing a demand curve but only satisfy ourselves that the willingness of consumers to pay is at least as large as without project costs. The appropriate output valuation concept hinges on whether or not nonmarginal change is implied by public investment in the project. As discussed by Schmid, there is a vast difference between marginal changes and nonmarginal changes. Marginal change is concerned only with adjustments at the margin whereas "major large-scale water development projects . . . and whole programs of development involve a number of projects over a period of years of such a magnitude as to have a nonmarginal effect on supply and thus on prices."1 The three cases under which efficiency gains are obtained noted in the previous section related to three elasticity of demand condi- tions for the output changes resulting from flood control or other water resource investment. 1A. Allan Schmid, "Water Resource DeveIOpment: Public and Private Investment," in Opportunities for Regional Research on Water Resources Problems, ed. by D. T. Massey and G. D. Rose (Iowa City: University of Iowa, Agricultural Law Center, College of Law, 1968), p. 62. 24 Case Elasticity of Demand Condition1 a. Fewer productive resources 3. Completely inelastic demand required to produce same for agricultural products. level of agricultural output. (vertical demand curve). b. Same productive resources b. Completely elastic demand yield a greater level of for agricultural products output. (horizontal demand curve). c. Greater level of output is c. Downward $10ping demand obtained with fewer produc- curve somewhere between tive resources than would slope of cases (a) and have been required in the (b). absence of protection. The three cases regarding elasticity of demand conditions for the project output changes are displayed in Figure 2. Panel I represents the case of a completely inelastic (vertical) demand curve (DD). This case implies that consumers place no value on any production in excess of a given level of output. (8050) represents the aggregate supply curve for a given output prior to the project. Following the project installation, the supply curve shifts to (SO'SO) because of lowered production costs and output price falls to P1. Economic efficiency gains are measured by the area be- tween the old (SOSO) and new (SO'SO') supply curves. It should be noted that this analysis would apply in the case of nonmarginal change. The nonmarginal change could either be the result of many 1Elasticity of demand refers to the extent to which quantity changes in reSponse to some given price change, or vice versa, a relationship that is generally assumed negative in price analysis. For example, an elasticity of demand of -.5 means that for a 10 per- cent increase in price there will be a 5 percent decrease in quantity sold; conversely putting 5 percent more of the commodity on the market (increasing supply by 5 percent) will lower the price by 10 percent. 25 .cofiumnflm> psmuso :o mcofiufipcoo quUfiummHo mo muoommm .N onnwfim unmade Hench usmuso Hence 0 o .x x on H m o a muuohoua Houucou muoononm Honuaou poofim Houm< van ouomom poon nopm< was ouomom cofiuucnm zammsm Hmcofimom mnemonmm< convocsu xfimmsm Hmcowmom oummoumm< oo woumaom mcofluwpcou panama uflummflm op woumaom mnofluflpcou nausea uwummfloaH HH Honda H Honda 26 poscfiucou .N oenwfim C) O’ xufiucmso Ho ------- --- ‘——--—— . _ _ . . _ . . . . . . _ . _ _ _ _ illlltllll om muoonoum Houucou poon poum< was ouomom cofiuucsm xflmmsm Hmcoflwom oumwonwm< ou woumfiom maofiufipcou afiummaocH unnameom HHH aocmm 27 small projects in the nation or a large, single project as it affects a region of a nation. Panel 11 represents the case of a completely elastic demand curve. The demand curve is perfectly elastic, and implies that the market will accept any increase (or decrease) in supply at a constant price. The supply curve (8050) represents the supply curve prior to the public project. The supply curve (SO'SO') represents the supply curve following the project. The supply curve shifts rightward be- cause productivity of some suppliers has been improved by the project. All of the additional output is assumed to marketed without affecting the post project price levels. The Corps of Engineers, Soil Conservation Service, Bureau of Reclamation and other public agencies have used the assumption of elastic demand as displayed in Panel II. If the elastic demand ap- proach is granted, efficiency gains from the project are represented by the area between the old (5050) and new output supply curves (SO'SO'). Panel 111 represents the case of a downward-sloping demand. This situation is seen to be intermediate between the completely elastic and completely inelastic cases just discussed. The project effects in this case are represented by the rightward and downward shift of the supply curve from S to 5'. Calculation of the net gain to society is a matter of debate in the economic literature since it . . . . . 1 1nvolves the interpretatlon of consumer surplus whlch 15 area C. 1For a summary of the literature regarding consumer surplus and related issues, see John Martin Currie, John A. Murphy, and Andrew 28 The increased output generated by the project pushes the output price down affecting all producers. A necessary assumption associated with this approach is that marginal producers are unable to cover variable costs after the price falls and are, as a result, forced to drop out of production. In the RLP model of benefits, there is no consumer surplus since there is no change in total quantity produced. Thus, the role of the role of consumer surplus in welfare analysis doesn't arise because of the character of the assumptions made. The project output in the RLP model displaces other firms' output, so there is no change in consumer surplus since output is con- stant. There can only be resource savings (and some transfers among resource owners of different productivity). Demand Conditions in U.S.Agriculture The classical study of demand for U.S. agricultural commodities was made by Brandow in 1961.1 In his study, Brandow found the price elasticity at retail to be -.34 for all foods. A more recent econo- metric analysis by Egbert2 found that price elasticity for consumption Schmitz, "The Concept of Economic Surplus and its Use in Economic Analysis," The Economic Journal [the quarterly journal of the Royal Economic Society], LXXXI (December, 1971), 741-99. Also see Bromley, Schmid, and Lord, Public Water Resource Project, p, A-2. 1George Brandow, Interrelationships Among Demands for Farm Products and Implications for Control of Market Supply, Bulletin 680 (State College, Pa.: Pennsylvania State University, College of Agriculture, 1961). 2Alvin C. Egbert, "An Aggregate Model of Agriculture-Empirical Estimates and Some Policy Implications," American Journal of Agri- cultural Economics, LI (February, 1969), 71-86. 29 demand for the period 1964-66 is —.06. The interpretation of this coefficient is that a 1 percent decrease in retail flood prices will cause only a 1/16 percent increase in quantity demanded by consumers. The U.S. agricultural economy is also characterized by excess capacity. Acreage allotments and other restrictions are necessary in order to prevent even larger price depressing surpluses. In his analysis, Egbert found that even though output expansion is bridled by holding available crop- land out of production, average prices are projected to decline moderately through 1974 . . . The implication is: Even if cur- rent programs are continued, additional land will need to be withdrawn from production in order to keep prices of farm com- modities advancing at the same pace as other commodities, at least through the mid-seventies. Looking at it another way, this result indicates that surplus productive capacity in (U.S agriculture will continue to grow over the next several years. 1) From the Brandow and Egbert studies, it can be seen that the demand curve is quite inelastic. Thus, public investment in additional flood control projects will result in project-induced output which will be accepted by the economy, but at lower prices. The RLP Model Approach The RLP model is a cost savings model which assumes a fixed requirements assumption regarding output. The RLP assumes that total output is not affected by the project. Graphically, the outcomes anticipated using the RLP model are illustrated in Panel I of Figure 2. In order to use the RLP model as a benefit measurement model, we must assume that the without-project costs for the fixed output could in fact be covered by consumer willingness to pay. This issue 11bid., pp. 80-82. 30 was discussed by Bromley, Schmid and Lord1 and involves the case where the alternative cost approach to benefit evaluation is relevant. For this situation, Steiner is relevant: Where a viable alternative exists and will be activated in the absence of government action, its costs do substitute for bene- fits. If the list of services is the same, the benefits will be equal. In this case benefit measurement is totally un- necessary, and comparative costs provide necessary and sufficient conditions for choice. The RLP model can be used to estimate cost savings when an effective fixed output demand is assumed. The cost saving with the project is estimated. An estimate is made of the resources saved in producing the gigep_output compared with producing the given output under "without" project conditions. It is granted that the RLP assumptions do not conform to realism with respect to the U.S. agricultural economy. In fact, out- put has increased as a result of development projects and prices have been depressed as a consequence. The objective chosen here is not to predict future outputs and price, but instead to interpret the value of project effects as a gauge to project investments regardless of the eventual "real world” outcomes. It is a policy option to regard real gains only as cost savings on a fixed output even thoggh we have no way to manage basin resources as a single firm and hold output constant. Given the demand characteristics of American agriculture, this is a plausible policy 1Bromley, Schmid, and Lord, Public Water Resource Project, p. 17. 2Peter O. Steiner, "The Role of Alternative Cost in Project Design and Selection," in Water Research, ed. by A. V. Kneese and Stephen C. Smith (Baltimore, Md.: The Johns Hopkins Press, 1966). 31 option even if the structure of agriculture is not modified to actually achieve the cost minimization result. If this policy Option is chosen, we can design a model to implement it. The RLP has been tailored to meet the specifications of this policy Option. CHAPTER III OBJECTIVE OF THE RESEARCH INQUIRY Introduction The central purpose Of the research in this part is to examine and test the feasibility Of utilizing a regional linear programming (RLP) model of the agricultural sector, as developed for river basin planning, as an analytical device to estimate agricultural benefits Of specific flood control projects prOposed by the Corps Of Engineers. Regional linear programming models provide an analytical framework for evaluating economic need for and the consequences Of water resource investments in a specified region. Operational RLP planning models have required the assembly Of extensive information about projected demand for agricultural commodities and Of crop production data re- flecting both flood-prone and flood-free conditions. Cost and yield information Of both flood plain and upland soils in a region for selected years are projected. The model, with its informational base, thus provides a potential for estimating agricultural flood control benefits resulting from a prOposed project. The availability Of Operational regional planning models does not automatically assure their application to the task of estimating benefits from a specific flood control project(s). Several problems 32 33 must be resolved prior to adapting the planning model to a project evaluation assignment. Selection and Features of Linear Programming_ Linear programming offers a means by which a broad range of production possibilities for flood plain soils can be considered simultaneously with similar possibilities for upland soils. Production costs and yield responses for a variety of crOps on specified soil groups, under with and without flooding conditions, can be analyzed via computer to determine likely changes in land use as a result of providing flood protection. Associated with these land use changes are reductions in the costs of producing the necessary food and fiber in the region, which can be interpreted as a saving due to flood protec— tion. This system Offers the possibility of analyzing more detailed information regarding the agricultural effects of flood protection than is possible through the use of the "composite acre" approach that is used in conventional agency methods. The adOption of linear programming as a tool of analysis, however, should not be made without recognizing the underlying basic assumptions of this technique. The linear programming model, like any other model, is an abstraction of reality. According to Swanson, "The researcher who uses an LP model abstracts those features of a problem which are be- lieved to be most crucial and places them in a systematic framework. The LP model assumes that the production processes may be broken down 34 into elementary processes or activities tied tOgether by a set of linear relations."1 These activities, together with the Specified stock of avail- able resources, define the production possibilities of Opportunities. Numerical estimates of resource availability, production coefficients, and activity weights must then be obtained. Four postulates of linear programming have been listed by Dorfman--linearity, divisibility, additivity, and finiteness. (l) Assumption of Linearity—-demands that for each activity the ratios between the two inputs and the product are fixed and hence independent of the level at which the activity operates. Thus inputs are combined in technically fixed proportions.3 The production function that is represented by an LP model is assumed to be homogeneous in the first degree, that is, there are constant returns to scale in any one process. This implies that the same quantity of output is obtained from each given set of inputs, regardless of the number of input sets used. (2) Assumption of Divisibility-—given the process or activity, all non-negative levels of the process are considered as possi- bilities. Since activity levels are not forced to take integral values (and can thus assume fractional levels), neither are the resource requirements required to take integral values.4 1Earl R. Swanson, "Programming Optimal Farm Plans," Farm Size and Output Research, Southern Cooperative Series, Bulletin No. 56 (June, 1958), 47. 2Robert Dorfman, The Application Of Linear Programming to the Theory Of the Firm (Berkeley: University of California Press, 1951), Chapter IV. 3Swanson, "Programming Optimal Farm Plans," p. 47. 4Ibid. 35 This assumption should not cause any problem if the units of inputs and outputs can be defined in small quantities so that any rounding of fractional inputs or outputs in the final solution can be made without significantly altering the values of the numbers in the solution. (3) Assumption of Additivity--this implies that with the simul- taneous operation of two or more activities, the total product (TP) produced is equal to the sums of the products produced by the individual processes. The quantities of inputs required are sums of the requirements of each individual activity. (4) Assumption of Finiteness—-means that of all possible pro- cesses, only a few are considered as alternatives.1 Rationale for Using the Wabash RLP Model In the Wabash RLP planning model, a maximum of five resource management alternatives were considered for a given soil group--flood protection, flood protection plus drainage, drainage, irrigation, and existing resource condition. Within each management group only one process (input combination) was considered for the production of each of nine crops. In actuality, however, a broad range of input combina- tions could be considered for producing a given crop on a given soil group. The effect of the finiteness assumption may be reduced by in- creasing the number of alternative processes. As computational facili- ties become more adequate, finiteness is less of a problem. As indicated in Volume I, the efficiency benefits to society of providing flood protection to agricultural land could be estimated by " . . . either summing the individual benefits after netting-out all income transfers and cancellations, or by directly estimating the 1Ibid., p. 49. 36 shift in the aggregate supply curve."1 Regional linear programming models provide a direct estimate of the shift in the supply curve. A cost-minimizing LP model is constructed which Specifies the least-cost method of achieving a predetermined level of agricultural output from a region given the capability of the land and water resources in the region and the level of production technology utilized by farmers in the region. The model consiSts of: (l) a set of demands (point estimates) of commodities expected to be produced in the region; (2) an inventory of acres (aggregated into soil groups) within the region that have similar yield and cost-of-production characteristics; (3) crOp yields Obtained on each soil group; and (4) variable production costs associated with each crOp for each soil group. Yields and costs of the specified soil groups are derived to reflect average conditions as experienced by farmers in the region with these soil groups in their current state of development. A second set of yield and cost estimates are derived to reflect the productive capacity of the soil groups under average farm management conditions if the water problem is eliminated. In the case Of flood plain soils, the first set reflect yields and costs under existing flooding conditions and the second set reflect changes in flood risk after installation of flood control structures. These two sets Of data thus provide the basis for calculating "without develOpment" and "with development" solutions. Separate models can be constructed for selected time periods that incorporate the anticipated technical and economic 1U.S., Department of Agriculture, Economic Research Service, Analysis of Alternative Procedures for the Evaluation of Agricultural Flood Control Benefits, Vol. I (August, 1969), p. 31. 37 conditions for each reSpective selected time period. Target years of 1980, 2000, and 2020 have been used in the river basin planning studies. It should be pointed out to the reader that if the RLP pro- gram selects and uses a given soil group then a fixed combination of inputs is assumed used. The only difference is represented by a slight increase in variable inputs associated with the increased output, i.e., increased fertilizer and seed inputs necessary to sustain the post develOpment yield increase, as well as slightly increased harvesting costs. Comparisons of the total cost-of—production under "without" and "with" develOpment, when aggregated over the planning period, provides an estimate of the savings to society that would be realized as a result of the flood control project. This saving is equivalent to the efficiency benefit (including both crop damage reduction and net enhancement) that would accrue to the flood control project. Once the basic models are constructed for a region, they can be used to examine the effect of alternative sites or sizes of structures, and to analyze alternative configurations of a system of flood control structures. In addition to estimating the efficiency gains, the model can be used to indicate the land use changes that are likely to occur as a result of the flood control project. Land use changes are identi- fied on both flood plain areas and upland areas, as the proposed project alters the comparative advantage of different soil groups in different locations. The upland land use changes represent the offsets 38 that are expected to occur in the long run as a result of the project, in response to forces operating in the private market. By specifying the model to represent three points in time, e.g., 1980, 2000, and 2020, a dynamic perSpective can be obtained of the land use pattern as least cost output would respond to changes over time in commodity demand and technology of production under "with" and "without" project situations. An operational planning model has been develOped for the Wabash River Basin. This basin will be used for all subsequent analyses of this study to indicate the composition of the model, the type of adjustments that are necessary to convert a planning model to a project evaluation model, and to interpret the information from the solution output. CHAPTER IV FEATURES OF THE WABASH BASIN RLP MODEL Components of the Wabash RLP planning model are discussed briefly in this chapter to indicate the general format of the model and the type of data required for its implementation. This dis- cussion is not intended to be a critical review of the assumptions and rationale underlying the various components of the model. Determination of Regional Commodity Demand Levels The first step in the process of determining future agricul- tural demands on a river basin, such as the Wabash Basin, is to esti- mate national output of food and fiber for selected time periods. Estimates of the national demand levels for the major commodities pro- duced in the nation were provided by commodity specialists of the U.S. Department of Agriculture, based on domestic and foreign export de- mands. Estimated domestic demand levels are based on projections of the population for each of the three target years 1980, 2000, and 2020, and on projections of per capita consumption rates for the major commodities. The summation of domestic demands and projected export demand, by commodities, determines the expected total national food and fiber output. The identification of the Ohio Basin's expected 39 40 output was based on extrapolation of past regional trends in crop and livestock production. Adjustments were made to reflect the judgment of commodity specialists regarding probable shifts in production among the country's water resource regions. Domestic and export demands for livestock and livestock products were translated into requirements for feed and forage for each water resource region. The level of crop output for the Wabash Basin was derived in- directly from the national projections of output for the Ohio River Basin. This was done through a detailed evaluation of the Wabash Basin's historical contribution to the total Ohio Basin production. As indicated on page 30, a national viewpoint toward estimated flood control benefits is obtained by adopting a model which is based on inelastic demands for agricultural commodities. Given the highly inelastic nature of most crops that are likely to be grown on the flood plain after flood protection is provided, the assumption of fixed de- mands in the model is not unrealistic.1 An analysis of historic trends in the geographic location of the nation's output of food and fiber and judgment of commodity spe- cialists regarding future shifts among region, provide a basis for indicating probable future levels Of agricultural production in various regions. In order to provide an indication of "national efficiency" benefits from the prOposed flood control projects, a major assumption 1In this study, the analysis will be confined to a single point estimate of demands for the various commodities, i.e., a com- pletely inelastic demand. In actual practice, however, it may be desirable to use two or more point estimates of demand to reflect different points on the demand curve. Selection of relevant alternate points will depend on the nature of the supply curve and the probable shift in the supply curve associated with the proposed flood protection. 41 is necessary. The model assumes that the Nation's food and fiber re- quirements will be produced in the various regions at the projected levels, an analysis of resource use in a given region, with and without flood protection. This does not necessarily mean that estimates of land use adjustments resulting from provision of flood protection to a region will actually occur entirely within the study region. To the extent land shifted out Of production within the region has lower (higher) marginal unit costs of production then cropland that remains in production elsewhere in the Nation, the national efficiency gains will be under-estimated (over-estimated). If the adjustments occur outside the region, this implies an increae in the relative share Of agricultural commodities produced within the region. Sensitivity of the ERS Model's Assumptions In 1970, John Hostetler, field group leader for the North Central Resource Group, NRED, ERS, completed his thesis1 evaluating the sensitivity of the ERS river basin planning model to changes in assumptions. In his study, Hostetler tested five classes of assump- tions with respect to sensitivity. They were assumptions relating to: (1) livestock feeding relationships, (2) projected demand levels, (3) soil management practices, (4) minimum acreage constraints, and (5) level of crOp producing technology adoption. 1Hostetler, "Sensitivity Analysis of Selected Linear PrOgram- ming Assumptions." 42 This study found that of the five classes of assumptions tested, both the assumption concerning soil management practices and minimum acreage restricts were relatively insensitive. Hostetler noted that "of the five classes of assumptions tested, the assumption concern- ing soil management practices should be dropped from future river basin models."1 Regarding feeding relationships, ration composition was relatively unimportant but feeding efficiency (conversion ratio) was fairly sensitive. The demand level specification differences cause "production costs to be in error in the same direction by approximately the same degree."2 The crop producing technology_assumption was moderately sensitive to changes. In the Wabash Basin RLP model, the sensitivity of two of the five classes of assumptions are relevant. First, the minimum acreage_ constraints, which were used by ERS in an effort to better approximate anticipated land use changes. Second, the evaluation of commodity demand will be undertaken. Sensitivity of Water Construction Agency Models A recent thesis3 evaluated the sensitivity of the Soil Con- servation Service's project evaluation model. Since the SCS model employs the same conceptual and analytical framework as that employed by the Corps of Engineers, Vondruska's findings are applicable to a comparison of Corps' and the RLP models. 1Ibid., pp. 121-22. 2Ibid., p. 114. 3John Vondruska, "An Economic Analysis of Small Watershed Project Evaluation Procedures" (unpublished Ph.D. dissertation, Michigan State University, 1971). 43 In his study, Vondruska evaluated the sensitivity of price, yield, and hydrological variables with respect to their effect on net present values (NPV) and benefit-cost (B-C) ratios for selected Michi- gan SCS projects. The finding that commodity output values used were highly sensitive was fairly comparative to the Hostetler finding of crOp producing technology sensitivity. The latter relates to the effect of on-farm production cost coefficients as anticipated under alternative levels Of adOption by farmers, i.e., if most farmers adopt cost reducing technology, basinwide production costs fall significantly. By contrast, Vondruska evaluated the effect of varying commodity market price levels and found them to be very sensitive in influencing NPV and B-C ratios. Although the Vondruska investigation did not include an investi- gation of SCS project benefits under an alternative analytical frame- work as does the RLP, his approach is unique in that it provides the first critical in-depth review of the benefit-cost model common to the Corps and SCS. His findings will be kept in mind as we proceed to our empirical investigation. We will return to evaluate similarities and differences between the Vondruska study and this study in the final chapter. Land Resource Availability The basic units of the RLP model are groupings of soils with similar yield and cost-of—production characteristics. These groupings are derived from an examination of the Land Resource Area/Land Capability Unit (LRA/LCU) classification of soils in consultation with soil scientists. Specific acreages of soils in each LRA/LCU for a 44 given area are estimated from information in the USDA Conservation Needs Inventory (CNI).1 The agricultural land base, including crop- land and pasture, is the residual land area of the Wabash Basin after deducting urban, forest, and other land use needs. Urban land is pro- jected to expand over the period to reflect growth in population and the accompanying increased demands for land. The demand for urban land use was obtained from projections of expected additional land re- quirements for an increased pOpulation. Forest lands in the Wabash Basin are expected to remain relatively stable in the foreseeable future. Thus, the land area available for crOp and pasture land is expected to decline. Table 1 indicates estimates of the availability of crOpland for the period 1958 to 2020. Pasture land was permitted to transfer to cropland at a Specific cost per acre if its soil characteristics were suitable for crop production and the RLP model determined the transfer would be advantageous. In each projection year, a relatively small amount of pasture land was transferred to cropland in the Wabash Basin RLP planning model. 1U.S., Department of Agriculture, Conservation Needs Inventory [for States of Indiana, Illinois, and Ohio], 1958. 45 TABLE l.--Crop1and Withdrawal, Wabash River Basin, 1958-2020 (Index 1958 = 100) , 1958 : Index : Percent Subarea : CrOpland : : , Change 1980 : 2000 : 2020 : 1958-2020 1---l,000 acres--- --Percent-- 1 1,999 92.6 89.1 86.4 -l3.6 2 2,783 94.3 89.0 85.5 -14.5 3 1,378 96.1 95.8 95.4 - 4.6 4 2,733 97.6 96.8 95.9 - 4.1 5 1,409 97.2 96.6 96.4 — 3.6 6 : 3,891 96;3_ gfl;§_ ‘93LE - 6.8 Total : 14,234 95.7 93.4 91.8 - 8.2 Sources: (1) 1958 Conservation Needs Inventory for Ohio, Indiana, Illinois. (2) Wabash Basin Type II Comprehensive Survey data. Production Costs On-farm production costs associated with producing the nine major crops grown in the Wabash Basin were necessary inputs to the Wabash RLP planning model. The sum of the four main production cost categories--preharvest work, materials, plant nutrients, and harvest- ing costs--provides an estimate of input costs associated with each potential crop activity. Data pertaining to cost, yield and input combinations for the Wabash Basin were based heavily upon the Ohio River Basin study coefficients. The Ohio study coefficients were develOped by a retired Ohio State University agronomist under contract 46 to USDA.1 Except for reviews by SCS and ERS personnel, the Willard data was utilized without major revisions. The preharvesting costs included all charges for land prepara- tion, spraying, planting, cultivating, and other preharvest activities. This category included labor charges, depreciation of equipment, taxes, insurance, and repairs, service, fuel, and lubricants for equipment. The materials category covered such items as twine for baling, herbi- cide Spray for corn, and seed costs. The plant nutrients included fertilizer and lime applications sufficient to sustain soil produc- tivity at the level specified in the RLP model. The harvesting costs were computed on a per acre basis for field operations such as com- bining, mowing, picking, and chopping. Costs associated with owning and maintaining fixed investments were not included. The Specification of the RLP model required all cost estimates to be based on out-of-pocket costs incurred at the farm gate. An Op- portunity cost for the land input was not included in the calculations, since land is a residual claimant and its value is a function of the net value of its output. Transportation costs also were not included under the assumption that most of the output produced would be con- sumed on the farm or delivered to nearby handlers. Bulky, perishable commodities such as fruits and vegetables, which have high transporta- tion costs, are of very limited importance in the Wabash Basin, and were not included in the planning model. 1C. J. Willard [Technical Consultant, Ohio River Basin Survey], "Estimated Production of Crops in 1980 and 2010 in the Ohio River Basin," Columbus, Ohio, May 20, 1964, 15 p. (Unpublished paper.) 47 Two sets of cost of production crop budgets were made for those LRA/LCU'S which were identified as having a water problem. The water problems included inadequate drainage, flood hazard, and a drought hazard (potential for irrigation). The first set represented the costs associated with the LRA/LCU in its current state of develOpment. The second set represented the production costs that would be incurred after the water problem was eliminated. In the case of flood plain lands, the first set included average annual crop production costs incurred with the existing flood hazard; the second set represented average annual costs incurred under flood-free conditions. Projected Technology One of the principal features of the RLP analytical system is the explicit manner by which it evaluates future conditions. In addi- tion to the projected commodity requirements for 1980, 2000, and 2020 discussed above, the analytical system also requires specific estimates of yields that are anticipated in the projection years. Crop and pasture yields utilized in the Wabash Basin RLP planning model were projected on the basis Of historical yield trends and potential future yield increases based on findings of current agronomy research. Both crop and soils specialists were consulted in developing yield projections for the soil groups in the Wabash Basin. Average farm management capabilities and average weather conditions were assumed for the target years. The yield estimates represent increased levels of inputs over time, such as improved seed, insecticides, fertilizer, and improved timeliness of farm Operations. Specific crop yield changes were estimated for the various LRA/LCU soil groups in the 48 Basin. The general anticipated trends in crop yields, as derived from regression analyses of past trends and judgment of crop Specialists, are presented in Table 2. TABLE 2.--Projected Yield Increases Per Year, Wabash River Basin Per acre yield increases per year CrOp , 1964 to 1980 I 1980 to 2020 ----- Bushels----- -----Bushels----- Corn 3 1.00 .67 Soybeans : .33 .26 Wheat f .50 .33 Oats : .50 .33 Barley E .12 .12 Constraints Built into the Model The use of a linear programming model to determine land use allocation in a river basin for future time periods will provide a solution based on the comparative advantage of the various soil groups. Analysts conducting river basin studies have observed that certain economic and institutional rigidities Of the agricultural economy may prevent achievement of an economic optimum based wholly on comparative advantage in crop production of soil groups in different parts of the basin. Because of this, certain constraints were built into the RLP model. Upper limits are placed upon the rate at which shifts in river basin cropping patterns can occur. Another way to view these constraints is to recognize that a decision made during the current time period affects the Opportunities 49 and choices during subsequent time periods. In an effort to better correlate the RLP model's Optimal farmer behavior with historically- observed behavior, these limitations on changes in cropping patterns are imposed. This approach enables some specification on limits of changes in the acquisition and accumulation of resources. Six economic subareas were delineated in the Wabash River Survey to facilitate the analyses by the participating agencies. These subareas generally encompass major trade centers having similari- ties in industrial, manufacturing, and retail trade activities and also approximate the major hydrologic areas. These subareas were used in the RLP model. Constraints were imposed upon the model by speci- fying a minimum percentage of each geographical subarea's historic share of the crop output to be produced in that subarea. The use of these constraints assure that cropping patterns in a subarea will not consist of only one or two crops in which it has a high comparative advantage. The constraints also reflect the historical fact that farmers in the Wabash Basin generally have found it desirable to have a certain crop mix to enable maintenance of a balanced operation. The implication of employing the constraints is that crOps will likely con- tinue to be grown in areas where production has occurred historically. The constraints employed in the Wabash model specified that for those crops that are expected to decline or remain constant in total output, at least 50 percent of the historic output of each crop would continue to be produced in the respective subareas. These crops included oats, wheat, barley, and hay. For crops in a relative rising demand situation, the subarea output minimums for 1980 were 50 percent 50 of the 1959 base; for 2000--40 percent of the 1980 base; and for 2020--30 percent of the 2000 base. Crops in the rising demand cate- gory include: corn, soybeans, corn silage, and pasture. Limits on the extent of crOp pattern change were based on criteria established by NRED analysts using the Wabash Basin planning model. Restrictions on extent to which crop adjustment within each subarea could occur were arbitrarily set at levels indicated in Table 3 so that acreage of crops decreasing would only approach zero asymptotically by target year 2020. The limits were applied only at the subarea level. No constraints were placed on particular groups of soils, such as flood plain soils. TABLE 3.--Production Minimums for Selected Crops by Subarea, Wabash River Basin Production as percentage of base year Base 3 Target year : year Decreasing 3 Increasing cropsa ; crops ---Percent-—- ---Percent--- 1959 1980 50 50 1980 2000 50 40 2000 2020 50 3° aOats, wheat, barley, and hay. b . Soybeans, corn, corn S1lage, and pasture. Source: Wabash River Basin Type 11 Survey data. CHAPTER V MAJOR ISSUES AND PROBLEMS IN MODIFYING THE WABASH RLP MODEL FOR PROJECT EVALUATION Much of the information required by a RLP project evaluation model (RLP-PE) can be drawn directly from the RLP basin planning model (RLP-BF). However, since the project area will generally be smaller than the basin planning area, it is desirable that the RLP-PE model be based on more detailed data. There are three major problems or issues associated with adapting a RLP-BF model to a RLP-PE model that must be resolved prior to moving ahead with the empirical test. The first and most critical problem is concerned with the reliability of land resource data for the specific flood plain areas to be pro— tected. Several potential sources are available. A second problem is concerned with the derivation of accurate estimates of crop and pasture yields Of flood plain land under with and without flood pro- tection. The third problem is associated with a modification that is required in adapting a RLP-BF model to a project situation. This problem involves the development of procedures to reflect yields as- sociated with partial protection actually afforded by a project, in lieu of yields based on assumed 100 percent protection in the RLP-BF model. 51 52 Obtaining Land Resource Data for Ppoject Flood Plain Areas One of the inputs necessary in any flood control benefit esti- mation study is reliable productivity data regarding the flood plain acreage subject to inundation. Land resource data for many flood plain areas has not been systematically collected. If the productivity information is to be evaluated in a RLP model, it must be compatible with the data format of the model in order to be incorporated into the analysis. There were three relevant alternative data sources to consider in this study. First, the Conservation Needs Inventory data, which served as the source of land inputs for the Wabash River Basin Type II Survey. Second, the soil survey reports published by the Soil Con- servation Service, USDA, in cooperation with the Stateagricultural experiment stations. Third, land use data from Corps of Engineers pro- ject justification studies. These studies evaluated crop yields and land use for all project impact areas downstream from prOposed Corps flood control reservoirs. A discussion of the features and limita- tions of each of these sources is presented below. The Conservation Needs Inventory The initial national CNI land use and land capability clas- sification data were collected and evaluated between 1957 and 1961. Soils with a flood hazard were identified in this inventory. Two yield estimates were derived to reflect average annual flooding condi- tions and flood-free conditions. The CNI is not a complete inventory 53 of all lands but rather, a 2 percent random sample of quarter sections (160 acres) from each county in the United States. A 2 percent sampling rate was used in the CNI in order to pro~ vide an acceptable rate of statistical reliability for counties con- taining between 250,000 and 500,000 acres. Since the impact area for a given flood control project frequently will either traverse more than one county or be less than 250,000 acres, it was necessary to evaluate the extent to which CNI data could be used in determining the composition of flood plain lands. Statistical tests were con- ducted in an effort to determine the acceptability of CNI data for use in a flood control project evaluation study. An evaluation of the CNI'S reliability in estimating the true proportions of flood plain soil types was made by applying chi-square tests to the CNI data and soil survey information for ten Indiana counties for which modern soil survey reports were available. These soil survey reports were assumed to contain the true or population parameters of flood plain soil since the soil surveys consist of on- site inspection of the entire area. Flood plain information for the ten counties is summarized in Table 5. In order to test the effect Of increased size of the sampled area, flood plain lands from the ten counties were combined serially. Five different orderings of the counties were made--alphabetic, reverse alphabetic, size of flood plain--1arge acreage to small, size of flood plain-~small acreage to large, and by county chi—square agreement-~smallest to largest. In each ordering of flood plain lands were summed, starting with the first county. A record was kept of the 54 TABLE 4.--Chi-Square Test of Agreement Between Soil Survey and CNI for Selected Soil Types, Ten Indiana Countiesa C County County : Chi-Sq. ounty number 5011 survey : SS/CNI acreage : agreement Bartholomew 1 53,504 .049 Carroll 2 18,330 .001 Cass 3 14,326 .620 Fountain 4 18,232 .410 Gibson 5 99,200 .600 Knox 6 72,064 .200 Miami 7 19,968 .036 Owen 8 30,589 .001 Parke 9 31,398 .815 Tippecanoe 10 27,629 .407 aFor three flood plain soil types: well-drained, fair drainage, poorly drained. Source: (1) U.S. Department of Agriculture, Soil Conservation Service. (2) U.S. Department of Agriculture, Conservation Needs Inventory, State of Indiana, 1958. statistical agreement between the CNI and soil survey (88) as the counties were aggregated serially. Composite information from the five orderings of counties is presented in Table 5. The second column of the table, "Number of observations," indicates the number of times a particular class interval of flood plain acreage was found among the five different orderings. As indicated in Table 5, the CNI/SS probability agreement be- came progressively higher as the size of the CNI sampled area in- creased. The outcomes indicate that an acreage level of approximately 200,000 acres is generally necessary for an agreement to exceed .90. 55 TABLE 5.--Composite Test Of CNI Reliability for Estimating Flood Plain Acreage, Ten Indiana Counties Flood plain 2 Number of E Class égiigée E Rggfijsgf acreage i observat1ons : average I agreement I agreement --Thousands-- 3 -Acres- -Chi-square- -Chi-square- 14.0 - 25.0 2 16,328 .310 .001 - .620 25.1 - 50.0 3 36,369 .180 .001 - .407 50.1 - 75.0 6 62,449 .337 .001 - .790 75.0 - 100.0 4 93,365 .319 .050 - .600 100.1 - 150.0 4 116,360 .396 .100 - .840 150.1 - 200.0 4 176,960 .704 .308 - .927 200.1 - 250.0 5 220,947 .873 .550 - .990 250.1 - 300.0 7 282,881 .834 .690 - .990 300.1 - 350.0 6 326,605 .936 .880 - .990 350.1 - 375.0 4 358,687 .986 .980 - .990 Over 375.0 5 385,240 .990 all .990 However, in one ordering, the agreement at the 224,000 acre level was only .55. The findings indicate that in order to be on the safe side, the CNI may be used as the sole source of flood plain land resource data for projects having impact areas exceeding 200,000 acres. If this rule of thumb is followed, the CNI estimate will be equivalent with the population parameters approximately 80 percent of the time. Soil Survey Reports Modern soil survey reports would serve as an excellent source of information concerning the acreage of various soils found on the flood plain, and would also indicate current or potential use of flood plain soils. If soil survey reports exist, flood frequency lines from flood plain maps could be superimposed on the soil classification maps. 56 Unfortunately, however, modern soil survey reports do not exist for all counties of the United States nor those in the Wabash Basin. Twenty-two Indiana counties and seven Illinois counties in the Wabash Basin have flood plain acres affected by the six operational and two proposed flood control reservoirs being evaluated by the RLP planning model. Of these 29 counties, only 10 have soil survey reports that are adequate for flood control benefit evaluation purposes. Soil surveys for the remaining 19 counties are either based on obsolete classification schemes, have a scale too large for accurate acreage assessment, or have not yet been published. Since full coverage of modern soil survey reports for all 29 counties was not available, this source of flood plain soils informa- tion was rejected for use in the study. If such information had been available, however, it would have been used in place of estimates of flood plain soils as derived from the Conservation Needs Inventory. Project Justification Studies The Corps of Engineers undertakes a detailed study for each flood control reservoir. In the Wabash Basin, fifteen flood control reservoir justification studies had been completed by mid-1967.1 In the course of the Corps' study, a systematic strip sample is made of agricultural areas that are to be protected. This sample is designed to include from 15 to 25 percent of the total area in the protected 2 . . . . . stream reach. The sample 15 used to determlne crOp d1str1but1on and 1U.S. Army Engineer District, Corps of Engineers, "Wabash River Comprehensive Study," Interim Report No. 3, Louisville, Kentucky, March, 1967. 2lbid., p. C-67. 57 yields, as well as to gather non—crOp agricultural damage information. In each completed project justification study, data that were current at the time the study was conducted were used for crOp distribution, yield, and commodity price levels. The crop distribution data are used to derive a composite land use acre for estimating flood damage reductions. The simplifying assumption made by the Corps is that the crop distribution for a given stream reach will hold for each acre of land in the reach. Use of the Corps' land resource data, in effect, implies that there is one flood plain soil type for each stream reach, with an accompanying set of crop yields. By contrast, the Wabash River Basin Type II Survey identified 17 flood plain soil groups. The present average corn yields for these 17 soils range from 41 to 89 bushels per acre. If the Corps' land resource data are used, they will provide a single weighted average figure representing the contribution of all the soil types found in that reach. The task of converting the Corps' average reach figure into the LRA/LCU soil groups system used in the RLP is an impossible one. If the single Corps figure for each reach is introduced into the RLP analysis as an output, the model may Specify that whole reaches should be used entirely for a single crop. This is due to the fact that in the LP model, comparative advantage in crOp production is largely responsible for the Optimal crOpping pattern in the basin. Another major problem in using the Corps' data for this partic- ular study is that much of the information is out of date. The flood 58 damage surveys used in project justification studies for the six Op- erational Corps' reservoirs in the Wabash Basin were completed prior to 1956. As studies are completed and projects are authorized by Congress, no resurveys are attempted on stream reaches affected solely by authorized projects. Since the Corps' project justification land resource data are not delineated by LCU soil groups and are not consistently updated, they were rejected as a source of information about the productivity of flood plain land soils for this study and CNI data were used instead. Since the impact areas for the six Operational reservoirs and the Big Pine-Lafayette reservoir complex include 654,687 and 412,814 acres, respectively, statistical reliability problems are not considered to be an issue in the study because both figures well exceed the minimum 200,000 acres. Estimating Flood Plain Crpp and Pasture Yields In the Wabash RLP planning model, cost and yield coefficients for flood plain soil groups were estimated without regard to upstream or downstream location.1 If there is a significant difference between upstream and downstream flooding conditions, then the use of average upstream and downstream crOp yield data could cause a significant bias in the results of the investigation. There appear to be two types of potential differences between upstream and downstream flooding 1Upstream and downstream designations are institutional de— lineations which satisfy the agreement between the Corps of Engineers and Soil Conservation Service. Upstream drainage areas, those tribu- taries containing up to 250,000 acres, are of special concern to the Soil Conservation Service. 59 conditions that could contribute to differential yields. First, variation in natural flooding conditions due to flood frequency, seasonality, depth, and duration; and second, the variation in the degree of protection currently afforded. Comparison of Upstream and Downstream Flooding. An empirical test was undertaken in an effort to determine whether differences do in fact exist in the crop yields that upstream and downstream farmers obtain. Corps and SCS project justification reports were evaluated in the test. Two evaluations of the data were made. First, an evaluation was made of the differences in average annual crop damages for a representative sample of Corps projects and SCS upstream watershed projects. Second, the damage factors for corn and soybeans under upstream and downstream flooding conditions were evaluated. The first test by the author was made by comparing average annual crop damages on soils with similar productivities in the up- stream and downstream areas. These comparisons revealed that average annual crOp damages reported for downstream acres exceed upstream crop losses by only 4.3 percent (Table 6). Although the same set of com- modity prices was used in making the comparison, two other key vari- ables, land use and per acre crop yield, were not evaluated. Corps of Engineers project justification studies are conducted independently of SCS studies to determine yield and land use, but are coordinated through inter-agency reviews. Therefore, differences are likely to be unimportant. 60 TABLE 6.--Comparison Of Average Annual CrOp Losses Upstream vs. Down- stream Areas, Wabash River Basin “P2522“ E ““232” Number of reaches 8 12 Acreagea 42,437 634,910 Sampling rateb 3.7% 42.2% Average annual damage $13.12 $13.68 Range ($5.47 - $19.05) ($7.14-- $23.24) Price set usedc AN AN aFlood plain defined by acreage inundated by 50-year recurrence interval storm for upstream and 100-year storm for downstream areas, respectively. bBased on total upstream acreage of 1,142,800 and total down- stream acreage of 1,504,800 as reported in U.S. Army Corps of Engineers, Ohio River Basin Comprehensive Survey Appendix M, "Flood Control," U.S. Army Engineer Division, Ohio River, Cincinnati, Ohio, December, 1967. Table WA-l, pp. 11-147. CThe 1957 USDA adjusted normal price set (AN). Sources: (1) SCS Preliminary Watershed Investigation Reports. (2) U.S. Army Corps of Engineers, Louisville District project justification data. The second comparison revealed differences in damage factors for specific agricultural crOps grown on the flood plain. Data for the upstream flood plain soils as compiled by the SCS used in the Wabash River Basin Type 11 Survey were compared with data from the same set of Corps studies used in the first evaluation. The Corps stream reach studies used in this evaluation were the most recent available and were located in four Of the six economic subareas Of the Wabash Basin. This comparison indicated that there is virtually no 61 difference (less than 1 percent) between Corps and upstream data of the weighted average damage factor for corn (Table 7). In contrast, there was a significant difference in the damage factors for soybeans, the second most important crop grown on flood plain lands. The Corps damage factor on the sampled downstream reaches was larger by 6.8 per- centage points than the weighted average damage factor used in the up- stream areas. This implies that there is a significant difference between upstream and downstream losses from flooding for soybeans, assuming that the Corps sample used is representative of downstream conditions. TABLE 7.--Comparison of Flood Damage Factors by CrOp, Corps and Wabash Basin Planning Model Data, Wabash River Basin Wabash planning Category model Corps projects Acreage 357,200a 735,900b Average annual corn loss 15.1% 15.0% Corn yield (average) 84.0 bu. 92.0 bu. Average annual soybean loss 9.3% 16.1% Soybean yield (average) 31.6 bu. 34.6 bu. aIncludes all upstream flood plain acreage for economic subareas 1-4. bRepresents a sample of eighteen Corps of Engineers stream reaches drawn from economic subareas 1-4. Sources: (1) U.S. Army Corps of Engineers, Louisville District project justification data. (2) Wabash River Basin Planning Model data. 62 The second major difference between upstream and downstream flooding is the variation in the degree of existing protection. There are 145 named levees in the Wabash Basin; most Of the levees protect downstream flood plain lands. The Corps has supplemented small pri- vate levees with an extensive system of levees, particularly along the lower reaches of the Wabash River. In contrast, upstream flood plain lands are seldom leveed. This is probably due to the fact that the narrower upstream flood plains do not have a ratio of area protected to levee miles favorable enough to permit economic justification. The net effect of this difference in the extent of levees is that the flood control increment for leveed downstream flood plain lands will be small relative to unleveed flood plain lands, all other things being equal. This is particularly true for Corps constructed levees along the lower Wabash, such as the Lyford Levee. The Lyford Levee, when completed in 1943, was designed to pro- tect against all floods of lS-year or more frequent recurrence inter- val. With the addition of four Corps reservoirs upstream, this levee will now protect against all floods of 50-year or more frequent re— currence. As upstream protection is added to supplement Corps reser- voirs, it is likely the Lyford will never be topped. The Corps' project justification reports do not explicitly state the extent to which existing levees affect prOposed projects since this would require estimating crop damages that would occur under natural (unleveed) conditions. This required that the investi- gator recompute the average annual damages for the reaches affected by the project, and is an extensive undertaking. In order to do a precise 63 job, all the available Corps data are necessary, plus information pertaining to private levees in the area. Since this is a highly technical, time-consuming, and costly procedure, it was not done for the individual stream reaches in this study. Since the vast majority of these private levees were con- structed prior to the assembly of crop yield data, the effect of these levees are reflected in the projected yields under the "without" develOpment condition. Additional protection to be provided by the proposed reservoirs is reflected in the projected yields under the "with" protection condition, through the application of the damage reduction factors obtained from the Corps analysis of these stream reaches. This procedure results in consistency between the Corps evaluation and the RLP-PE analysis with respect to considering the effect of the private levees. Pasture damages were found to be insignificant in the Wabash River flood plain evaluated in this study. Specifically, there are only 7,250 acres of pastureland out of a total of 525,800 acres of Wabash River flood plain below the Big Pine reservoir. Since less than 2 percent of the flood plain is in pasture and its value is so low, pasture damages can effectively be ignored.1 In summary, the examination of crop yields and the crop damage factors employed for downstream soils did not reveal differences that were felt to invalidate the use of basin-wide cost and yield data, as 1Gross cash rent per acre of pasture for the State of Indiana ranged between $9 and $10 per acre in 1964-66. U.S., Department of Agriculture, Farm Real Estate Market Developments, CD-67 (August, 1965) and CD-7l (December, 1968). 64 compiled for the RLP planning model, in the project evaluation model. The large differences between upstream and downstream areas in the damage factor for soybeans may have some effect on the estimated efficiency benefit to flood control as derived from the project eval- uation model; however, it will not be a significant factor in testing the general application of the proposed model to flood contrOl benefit evaluation. Partial Flood Control Protection In the Wabash RLP model, the with flood protection condition was based on total or 100 percent protection against all crop flood losses. This specification was adopted in the RLP model because the objective of the Wabash River Basin Type II Survey was to measure the maximum potential societal gains from the water and related land re- source develOpment activities. In the application of the project evaluation RLP model, the purpose will be to evaluate only the societal gains from the level of additional flood protection provided by that project. The protection level downstream from a given flood control project will be less than 100 percent for at least some portion of the protected flood plain. This is true for several reasons. First, there is no way to specify with absolute certainty the maximum possible flood. The ultimate size limit of a flood for an area is approximated by the Corps' maximum probable flood used in the design of a spillway to insure the area's 6S safety.1 Without exception, such a flood is considerably in excess of the design flood--the one against which the given area is to be pro- tected,--which reflects a balance between maximum net benefits and engineering safety and integrity standards. Second, the level of flood protection afforded downstream is, to a large extent, related to un- controlled drainage area. Therefore, the level of protection decreases with distance downstream from the reservoir (Table 8). TABLE 8.--Indices of Flood Protection to CrOpland on the Wabash River Flood Plain Index Reach 3 Miles below 3 Big Pine site i Present i After conditions8 3 project W-6 ; 2.5 - 32.8 .397 .598 W-SA = 32.8 - 52.3 .427 .619 W-SB ; 52.3 - 75.5 .395 .562 w-4 = 75.5 - 166.1 .371 .553 w-3 ; 166.1 - 195.8 .401 .516 w-z = 195.8 - 250.3 .161 .269 w-1 ; 250.3 - 290.3 .127 .216 aProtection index is the reduction in average annual dollar crop damages over a no protection condition, expressed in decimals. Present condition (1969) includes Salamonie, Mississinewa, Huntington, Cagles Mills, Monroe, Mansfield reservoirs. bIncludes existing reservoirs plus authorized Big Pine and Lafayette reservoirs. Source: Louisville District Corps Data. 1U.S. Army, Corps of Engineers, "Survey Investigations and Reports," p. 49. The maximum probable flood is the largest flood for which there is any reasonable expectancy in this climatic era. Its recurrence interval is unspecified but most infrequent. 66 The Wabash River Basin Type 11 Survey data utilized in the RLP model expressed crop yields and associated costs for flood plain soil groups in terms of 100 percent flood protection. These flood-free yield and cost coefficients were computed for all flood plain soil groups identified in the Wabash Basin. In order to estimate the effect of a specific flood control reservoir which will not provide 100 percent protection throughout the affected flood plain, it was assumed that the effect could be expressed as some fraction of the yield increase associated with complete protection. The portion of the yield increase associated with providing additional flood protec- tion is designated as the flood control increment. This increment for a given soil group reflects the single basin-wide estimate for average annual flooding condition. This figure represents the judg- ment of State soil scientists as they evaluate the particular soil group in the context of the flooding conditions under which it is found. Project justification data for the downstream reaches being evaluated in the empirical investigation were Obtained from the Louisville District Corps Office. These data evaluated the average annual dollar crop damages, under both natural conditions and the flooding conditions anticipated following the installation of a flood control reservoir. The percentage reduction in average annual crop and pasture damages following the installation of the flood control reservoir was used as an estimate of the effect of the project in this empirical investigation. This impact was incorporated into the analysis by way 67 of increasing the yields and associated costs of the flood plain lands affected by the project. For example, if the Big Pine and Lafayette reservoirs were added to the existing levels of flood protection, average annual crop damages for Wabash River stream reach W-6 would be reduced by $92,679 (1960 values). Since the Corps estimated average annual crop losses for this reach to be $275,862 (1960 values) following the completion Of the upper three Wabash reservoirs (Salamonie, Mississinewa, Hunting- ton), this is a reduction in average annual crop losses of 33.6 percent. Each flood plain soil group in the reach has an associated flood con- trol yield increment which represents the difference between present flooding conditions and flood-free conditions. The simplifying assump- tion was made that all affected flood plain soil groups in the reach will respond to this reservoir's partial flood protection to the same extent. Operating under this assumption, the flood control increment for each soil group was then obtained as a product of the damage reduction factor and the flood control yield increment. The sum of the crop yield increase attributed to the particular reservoir and the present average soil group crOp yield gives an estimate Of the expected yields following the project. The use of Corps dollar damage data contains two key implica- tions which may or may not have an effect on the outcome of the empiri- cal investigation. First, the reach-wide dollar damage reduction factor (computed from Corps data) was used to adjust upward the flood-prone yields for corn and soybeans. The resultant yields for corn and soybeans reflect 68 the estimated effect of the additional flood protection provided by the set of Corps flood control project which was evaluated in the empirical investigation (see Chapter VII). Since a Single coefficient reflects the effect on both corn and soybean yields, this implies that the damage factors for corn and soybeans are identical. The Corps estimates of yield reductions (Table 7) appear to bear this out for the downstream flood plain of the Wabash Basin. The estimates derived from the Wabash planning models in Table 7, however, imply that soy- bean losses are understated by 6.8 percentage points. The second key implication of using the Corps reach-wide dollar damage index is that the relationship between corn and soybean valuation per acre in the base year (1960) will also apply to the target year 1980. Variation in crop valuation over time, due to relative changes in market price and per acre yield of these two crops, were examined by comparing the effect of both prOportional and nonproportional per acre valuation increases for corn and soybeans (Table 9). Three simplifying assumptions were made in this hypothetical example. First, the prices paid for NO. 3 corn and No. 2 soybeans at Chicago are the relevant market prices. Second, the statewide yield averages for Indiana are the relevant per acre output coefficients. Third, the average annual damage factors for the corn and soybean crOps are identical at the level of 15 percent per year. The data in Part B, Table 9, represent a 20 percent increase in both the 1960 corn and soybean values reported in Part A. The data in Part C represent the actual trends in soybean and corn value per 69 TABLE 9.--Variations in Crop Loss Estimates Due to Relative Change in Crop Valuation Hypothetical Reach Crop 3 Value per 3 Land 3 Damage 3 Contribution to acrea I use I factor : composite acre Corn Soybeans Composite acre Corn Soybeans Composite acre Corn Soybeans Composite acre 3 a.--Estimate for 1960 $74.80 .50 .15 $ 5.61 58.59 .50 .15 4.39 $66.70 1.00 .15 $10.00 : b.--Estimate for 1969, with 20 percent valuation increase but no change in relative crop valuation f $90.00 .50 .15 $ 6.75 § 70.80 __£31 11E. 5.31 : $80.40 1.00 .15 $12.06 E c.--Estimate for 1969, using current crOp , valuations 2 $95.45 .50 .15 $ 7.16 E 70.22 ._£g; gl§_ 5.27 i $82.84 1.00 .15 $12.43 aStatistical Reporting Service, USDA, Crop Production Annual Summaries for the years 1960-69,(Washington, D.C.: Government Pr1nting Office). 70 acre has increased 27.6 percent, whereas soybeans have increased only 19.9 percent. A comparison between Parts 8 and C indicate that the reachwide dollar damages would only be underestimated by 3.1 percent if constant valuation relationships are assumed between corn and soybeans. This result indicates that the projection of a constant valuation relationship between corn and soybeans does not cause a significant bias during the 8-year period 1960-68. Therefore, the projection of a constant valuation relationship for the period 1960-80 appears to be a reasonable one. In summary, the use of the percentage reduction in stream reach dollar crop damages as the coefficient to adjust present flood- prone yields upward appears statistically acceptable. The bias in- volved in assuming that the soybean and corn damage factors are equal and the bias associated with assuming a constant valuation relationship between these two crops do not appear to be significant in this appli- cation. CHAPTER VI MODIFICATION OF WABASH RLP MODEL Introduction Although the major focus of the empirical analysis in this study is on the Big Pine-Lafayette reservoirs, data for the entire set of eight reservoirs were analyzed. The six operational reservoirs were included for two reasons. First, the Wabash RLP planning model did not explicitly consider the total effect of this set of projects. Yield data used in the model were collected in 1963, which meant that the effect of the four largest and most recent structures was not evaluated at all. In addition, the effect of the Mansfield reservoir on flood plain yields was not likely incorporated because the reser- voir had been Operating only two years prior to collecting the yield data, and it was doubtful that its effect on yields was incorporated in the yield estimates. Second, the sequential analysis of the effect of the six operational reservoirs, followed by the two authorized reservoirs, provided a firm base for analyzing the separate or incre- mental effect of adding the Big Pine-Lafayette reservoirs to the flood control system. 1The six operational reservoirs include Cagles Mills, Mans- field, Monroe, Salamonie, Mississinewa, and Huntington. 71 72 Modifications to the RLP Planning Model The Wabash RLP-8P model was formulated to provide general basin-wide information on the economic need for and effects of river basin development projects and programs. Adapting the planning model to a project evaluation model requires some refinement in the formula- tion of the model. More detailed information about the anticipated effect of flood protection on specific acres to be affected by the prOposed project must be incorporated in the model. Four modifications in data format and Specification were made to convert from a planning model to an evaluation model: (1) identification of project-affected flood plain acres by soil groups, as used in the planning model; (2) estimation of yield effects on project-affected land resulting from the reduction in flood hazard; (3) revision of cost of production estimates for project-affected land; and (4) separation of flood con- trol and drainage increments on inadequately drained flood plain soils. Identification of Project-Affected Soil Groups As noted in earlier discussion, the flood plain lands evaluated in the Wabash RLP—BF model included all land capability units (LCU) subject to flood 1055, without regard to their geographical location within the basin. An estimate of the acreage of the LCU's affected by the two levels of flood protection was required as an input to the linear programming model. Guidance was Obtained from Indiana and Illinois State soil scientists with regard to those flood plain LCU'S likely to be found in the Corps project impact areas. The following 73 procedure was followed in order to obtain an estimate of the LCU composition of these downstream flood plain soil groups. 1. The Corps project flood plain cropland acreage was determined by county, based on Corps data. Total cropland acres on flood plains in upstream areas were determined for the counties having Corps flood plain land. Upstream and downstream cropland acreages were summed. The total Corps project impact area was subdivided into geo- graphical subareas, each having over 200,000 crOpland acres. The CNI totals of flood plain soils by LCU, by county, were determined for each Of the subareas. Within each subarea, the upstream acreage was subtracted from the CNI flood plain total. The LCU prOportions present in the remaining CNI acreage was then computed. These prOportions represented the LCU composi- tion for the Corps project impact lands in each subarea. The LCU proportions thus Obtained were multiplied by the Corps cropland acreage in each of the respective subareas. The LCU acreages thus obtained for project impact areas were introduced into the RLP-PE model as project flood plain land resource inputs with their associated cost and yield coefficients. 74 Yield Estimates for Project-Affected Flood Plain Lands In the Wabash River Basin Type 11 Survey, the flood protection alternative was approached under a slightly different set of assump- tions than desired in this investigation in three respects. First, the Type II Survey's "no develOpment" evaluation for 1980, 2000, and 2020 represent the outcome if present flood protection levels persist into these target years. The cost and yield coefficients for flood plain soils reflect the average of all flooding conditions experienced by each particular soil group in upstream areas. Second, the flood protection alternative in the Type II Survey was provided only on the basis of total protection against all cropping losses. Third, the cost and yield coefficients used in the Type 11 Survey for flood plain soils reflect optimal drainage conditions, as well as total flood protection, for these soil groups. Since 77 percent of all the downstream flood plain lands require additional drainage, the co- efficients for these soil groups will require modification to permit the economic impact attributed to flood control only to be evaluated in this study. The first problem noted above concerns the comparability of yields on upstream flood plain soils versus downstream flood plain soils under present flooding and flood-free conditions. In the RLP planning model the flood-control yield increment was defined as the expected yield difference between present flooding conditions and flood-free conditions. A comparison of downstream and upstream damage levels on flood plain lands, reported in Table 7, indicates that there is little difference in the flood control increment for corn (less 75 than 1 percent). This implies that the potential bias in using Type II upstream flood-control corn-yield increment data as an estimate of the corn yield increment associated with additional flood protection on downstream flood plain lands will be negligible. On the other hand, the comparison of Type II and Corps-sampled reaches indicates a difference of 6.8 percentage points in the flood control increment for soybean yields--9.3 percent and 16.1 percent, respectively, for the Type 11 survey and the Corps project study (see Table 7). A possible explanation for this discrepancy may relate to the Corps' use of the composite acre concept. The conposite acre con- cept implies that the reach-wide crop distribution holds for each acre of land in the reach regardless of location on the flood plain. Soy- beans are much more susceptible to flood damages than corn during the bulk of the growing season (Table 10). If flood damage calculations assume that soybeans are typically grown in high-risk zones, when in fact they are grown in low-risk zones, average annual damages will be overstated.1 After weighing the evidence regarding the flood-control yield increments for corn and soybeans, it was determined to use the esti- mates developed for the RLP-BF model without further adjustment, recog- nizing that the use of the upstream soybean flood-control increment could subject the findings to downward bias. This could bias both 1The assumption of homogeneity is modified in crop damage cal- culations to reflect distribution of crops by flood-hazard zones in normal Corps procedures. While soybean damages in Table 16 indicated considerably higher losses in July-September periods, it should be noted that flood probabilities are much lower for this period. River bottom farmers also shift almost entirely to soybeans after May-early June floods, thereby planting soybeans in apparently high-risk flood zones. 76 TABLE 10.-—Samp1e Unit Crop Damages as Percent of Total Crop Value Per Acre, Wabash River Flood Plain3 Time of flood : Soybeans : Corn --—-Percent---- ----Percent---- 1-15 June f 30.5 41.0 16-30 June 1 51.8 64.0 1-15 July f 66.1 38.5 16-31 July ; 67.3 26.5 1-15 August 3 67.3 18.0 16-31 August i 67.3 21.0 1-15 September f 66.1 8.2 16-30 September 2 58.2 5.9 1-15 October S 37.3 4.2 ° 6 2.8 16-31 October : l4. aCrOp damages were estimated on the basis of an inundation up to two feet and a duration of flooding up to forty-eight hours. Note: Maximum damage is total value of crop minus labor and expenses not expended at time of flood. Source: Wabash River Basin Comprehensive Study, Interim Report No. 3, Vol. III (Louisville, Ky.: U.S. Army Engineer District, Corps of Engineers, March, 1967), Table 53, p. c-70. estimates of the extent to which the project impact areas would grow soybeans, and of the production cost savings associated with the eight flood-control projects evaluated. The second major modification regarding yields was to select a methodology to adjust flood-prone yields upward to reflect the partial protection afforded by the six operational and two authorized Corps flood-control reservoirs (Table 11). The first set of projects, affording the additional flood protection designated as Flood 77 TABLE 11.--Crop Acreage and Flood Protection Index, by Selected River Reaches, Wabash River Basin : o I a Total : Crop Product1on 1ndex Stream 3 Reach , , . acreage . acreage . Flood Flood :Control 1 Control Wabash River : W-l 108,000 70,848 .1265 .2157 ; W-2 114,000 89,376 .1609 .2694 : W-3 99,700 79,062 .4007 .5164 ; W—4 154,500 117,111 .3705 .5528 = W-SA 19,700 17,099 .3953 .5619 ; W-SB 13,700 11,892 .4271 .6185 : W—6 16,200 14,110 .3972 .5997 ; W-7 14,600 13,316 .6185 .8035 : W-8 3,800 3,165 .6766 .8309 ; W-9 19,500 16,965 .8804 .8804 : W-lO 11,600 10,451 .9300 .9300 ; W-ll 3,600 3,243 .9500 .9500 Raccoon Creek ; RC-l 2,450 2,050 .6000 .6000 : RC-2 5,050 4,100 .9500 .9500 Lowel Eel : LE-l 45,000 43,335 .3500 .3500 White River - Main Stem : WE-l 24,200 17,593 .1193 .1193 East Fork : EFW-l 30,200 24,999 .2500 .2500 Whlte Rlver ; EFW—Z 9,200 7,500 .3500 .3500 Salt Creek : SC-l 7,350 5,500 .8000 .8000 Total Acreage = 823,350 654,687 aProtection index is the reduction in average annual dollar crop damages expressed in decimals (1960 prices). Flood Control 1 includes Salamonie, Mississinewa, Huntington, Cagles Mills, Monroe, and Mansfield reservoirs. Flood Control 2 includes Flood Control 1 plus Big Pine and Lafayette reservoirs. Source: Unpublished data, Corps of Engineers, Louisville District. 78 Control 1, includes all currently (1969) operational Corps reservoirs with which the empirical investigation is primarily concerned. The methodology selected to obtain an index of additional flood protection provided by these two sets of reservoirs was straight- forward. The percentage reduction in average annual dollar crOp damage attributed to each level of protection was derived from Louis- ville District Corps data (1960 values). For flood plain LCU'S not requiring additional drainage, the partial-protection crop yields for project impact areas were calculated in a two-step procedure. First, the flood-control yield increment for each impact area LCU was multi- plied by the protection index for the particular stream reach and set of reservoirs (Table 11). Second, this adjusted yield increment was added to the "no development" yield which reflects current flooding conditions. AS noted earlier, a third difficulty in using RLP planning model yield data resulted from the inclusion of both drainage and flood control yield increases in the flood control yield increment for imperfectly drained flood plain LCU's. In the planning model, the flood control development alternative on inadequately drained flood plain land was derived on the basis that drainage would accompany the higher level of flood protection. Thus, the costs and yield coeffi- cients were calculated to reflect a "flood control plus drainage" con- dition. This was done for the planning model because it is generally held by soil scientists that the full benefit of flood control on inadequately drained soils will only be realized if drainage is also provided. In the project evaluation model we are first interested in 79 determining the effect of flood control as the initial effect on agri- cultural production; and secondly, the effect on adding drainage as a develOpment alternative. This approach will allow an evaluation of the efficiency benefits of providing additional flood protection under the "flood control only" and "flood control plus drainage" development assumptions. The following procedure was used to estimate the yield incre- ment on inadequately drained soils resulting from the partial protec- tion provided by the flood control project, in lieu of seeking field estimates of the effect of flood protection considered by itself. This procedure provides crop yield estimates for the "flood control only" analysis. 1. Obtain the yield increment between the optimally drained flood-free yield condition and the yield associated with optimal drainage under existing flood—protection levels. 2. Determine the percentage yield increase due to flood control only, by dividing the flood-control increment by the optimally drained flood-prone yield. 3. Multiply the present ”flood—prone inadequately drained" yield by the percentage derived in step 2 to obtain the "flood con- trol only" yield increment. 4. Multiply the flood control yield increment from step 3 by the flood protection index for the particular level and stream reach (Table 11). 80 5. Add the adjusted yield increment to the "flood-prone inade- quately drained" yield to obtain the partial protection "flood control only” yield. Estimates of yields and costs for the analysis, based on the assumption that drainage and flood control must be a joint development to achieve optimum yield increases on flood plain soils, were derived by the steps outlined below. This procedure provides partial- protection crop-yield estimates under the condition of the joint development, "flood control plus drainage" resulting from the instal- lation of the two sets of flood control reservoirs.1 1. Obtain the yield increment between the optimally drained flood-free yield condition and the yield associated with optimal drainage under existing conditions. 2. Multiply the flood control increment derived in step 1 by the flood protection index for the particular level and stream reach (Table 11). 3. Add the adjusted yield increment derived in step 2 to the optimally drained yield to obtain the partial protection "flood control plus drainage" yield. Cost of Crongroduction Estimates for Prpject-Affected Flood Plain Lands In the RLP planning model, the cost coefficients were stated in terms of the per-acre on-farm production costs for growing a 1Analysis of the flood control benefit under the "flood control plus drainage" assumption is presented in the Appendix A. 81 particular crOp on the respective LRA/LCU soil groups. The additional per-acre costs associated with the provision of additional flood pro- tection includes only those out-of—pocket costs incurred by the average flood-plain farmer. Thus, the additional costs include the cost of the increased inputs (fertilizer, seed, lime, etc.) necessary to raise the additional crOp output, plus the additional harvesting costs associated with the flood control crop-yield increment. No allocation of flood control project costs was made in either the RLP-8P model or this study. Revised per-acre costs for project impact lands were obtained by multiplying the flood control yield increment by the marginal cost per unit of output for each crop. This additional cost was added to the per-acre base cost for the project-affected LCU's, in order to obtain the partial development cost intermediate between flood-prone and flood-free costs. Summary of Input Revisions The costs, yields, and acreage of flood plain LCU's affected by the two sets of flood control projects were adjusted to reflect partial protection and to separate flood control effects from drainage effects. These data were prepared in a format for computer analysis that would enable the partial-protection flood control alternatives to be evaluated as revisions to the 1980 Wabash RLP-BF model "no develOpment" solution. There were four sets of cost, yield, and associated land re- source coefficients representing two levels of flood control and two assumed levels of drainage. Flood Control 1 Flood Control 2 Flood Control 1 conditions. Flood Control 2 conditions. under the under the under the under the 82 "flood "flood "flood "flood control only" conditions. control only" conditions. control plus optimal drainage" control plus optimal drainage" CHAPTER VII EMPIRICAL INVESTIGATIONS Introduction In previous discussion, production efficiency gains were identified as one of the primary benefits from flood control invest- ment. The RLP-PE model provides a direct estimate of these gains by calculating the reduction in total cost of crop production as a result of the flood protection. Two computer runs--without flood-control project and with flood-control project-~are required for each target year. The difference in the total cost of production between the two runs provides a point estimate of the savings that accrue to society as a result of the flood control project in each target year. An estimate of the total efficiency gains that are expected to occur is derived by extrapolating the three target year efficiency gains over the life of the project to determine the annual flow of benefits, and discounting them to a present value. In this study we were primarily concerned with determining whether a river basin planning model could be modified to serve as a project evaluation model. Therefore, only one target year was selected to test the model conversion. 1980 was selected as the target year, and the "no development" run completed for the Wabash River Basin Type 11 Survey was used to 83 84 estimate total cost Of production without the flood control project. Since we had chosen to test our model conversion procedures on the authorized Big Pine and Lafayette reservoirs, two "development" runs were necessary. One "development" run was based on revised yield and cost coefficients for flood plain lands affected by the six Operational flood-control reservoirs in the Upper Wabash Basin-~this is referred to as Flood Control 1.1 The information on total costs of production and land use patterns resulting from Flood Control 1, provides a new base to represent the "without development" situation for the analysis of the effect of the Big Pine and Lafayette reservoirs. The second "development" run included yield and cost coefficients which reflect the additional flood protection provided by the Big Pine and Lafayette reservoirs-~referred to as Flood Control 2. In the analysis below, it was assumed that no additional on— fanm drainage costs would be required or installed by flood plain farmers in order to realize higher levels of output associated with the reduction in flood risk. Additional analysis was made, however, based on an alternative assumption that additional on-farm drainage would be necessary to realize the flood protection benefits. This assumption implies that inadequately drained flood plain lands will not reSpond to flood control alone and that the joint development of flood control plus drainage is necessary to realize the full potential of reducing the flood hazard. Analysis based on this assumption is presented in Appendix C. Cost and yield modifications which were associated with the project-protected lands were introduced as revisions to the input 85 matrix used in the 1980 Wabash RLP-BF model. In order to lower total basin on-farm costs, the unit costs of crops grown on the soil groups in the development alternatives must be lower than unit costs of soil groups utilized in the "no development" solution. Thus, soil groups having cost and yield coefficients reflecting partial flood protection which enter the 1980 land-use allocation will displace lands on which it is more costly to grow the same quantity of total basin output. Low and High Demand Analysis The initial Wabash Basin Type II river basin projections were made on the basis Of the regional allocation assigned by the joint Office of Business Economics-~Economic Research Service Committee (OBE-ERS). However, in mid—1969, after the initial projections were completed, the OBE-ERS unexpectedly revised the regional share allo- cated to the Wabash Basin. The extent to which the OBEcERS realloca- tion affected the land use, production cost and output projections for 1980 are indicated in Table 12. The effect of the reallocation on the other target years is indicated in the USDA final report: The projected agricultural economy of the Wabash River basin could meet the specified level of crop and livestock production for 1980, given the constraints imposed upon the model (in the form of costs, yields, acreages of various soils, etc.) without supplementing the current level of water and related land resource develOpment . . . the same condition held for the 2000 and 2020 projection years as well. However, the "new" OBE-ERS regional allocation of national demand requirements resulted in a substan- tial increase in the share of some agricultural commodities being allocated to the Wabash River basin. These increased demands were significantly larger for milk and eggs in the livestock sector and for corn silage, wheat and soybeans among the field crops. The increase in demand requirements was sufficiently large to require a minor amount of additional resource development (flood control and on-farm drainage). By 2020, a substantial amount of additional resource development would be needed to 86 TABLE 12.--Land Use, Production Cost, and Output Comparisons, Low vs. High Demand Levels, Wabash River Basin, 1980 Projecteda Total Total ' Crop production 2 Output acres . COSt Low Demand Level = 100 Feed Grain Corn 3 91.06 91.37 91.51 Oats : 101.96 101.67 101.01 Barley 3 109.71 103.86 100.00 Total : 91.80 92.33 91.73 Roughage Hay E 85.06 84.89 89.22 Silage : 262.68 267.77 257.34 Pasture E 99.90 101.43 100.00 Total : 105.09 130.43 114.18 Feed Use Totalb : 95.98 97.86 96.04 Wheat : 180.14 177.88 172.43 Soybeans : 124.89 124.75 124.03 Grand Total : 111.32 110.45 ------ a . . . . . . PrOJections w1th s1x present Corps reserv01rs operational. bFeed grain plus roughage. 87 obtain the level of output Specified by the "new" OBE-ERS demand requirements given to the land resource base available for agri- cultural production.1 In 1980, the impact on overall basin land use indicate a 11 percent increase in agricultural lands utilized, with total basin on-farm costs up 10.5 percent (Table 12). The total output of feed grains was reduced. Total roughage output increased with pasture being held constant, hay reduced, and silage output more than doubled. Wheat and soybeans were also increased. In order to evaluate the effect of the revised demands with Corps reservoirs in place, the Wabash RLP-PE model was rerun under Flood Control 1 and Flood Control 2 conditions. The results are re- ported in the sections which follow. Efficiency Gains The total on-farm variable costs which would be incurred in producing the estimated levels of agricultural commodities under alternative flood control conditions are indicated in Table 13. Under "low demand" conditions, the total on-farm costs for 1980 are reduced about $318,000 by the addition of the Flood Control 1 set of six reservoirs. Distribution of these efficiency gains across the 654,687 cropland acres indicates a gain in annual return of about $0.49 per acre. Efficiency gains resulting from the Big Pine-Lafayette project are associated with a flood plain impact area of 412,814 crop acres. 1Wabash River Basin Comprehensive Study, Vol. IX, Appendix H, Agriculture, June, 1971, p. 571 88 TABLE 13.--Effect of Flood Control Reservoirs on Agricultural Produc- tion Costs, Wabash River Basin, 1980 Projected Demand and Cost Conditions Status of flood E Total on-farm Q Incremental Inc::m:::al protection - costs - difference p c ' ‘ change Low Demand Conditions No development ' (no projects) 3 $471,084,363 ---- ---- Flood Control 13 ; 470,765,824 $318,539 .07 Flood Control 2b 470,547,350 218,474 .05 High Demand Conditions 3 Flood Control 18 § 519,953,210 ---- --—- Flood Control 2b 519,583,100 370,110 .07 aFlood Control 1 includes Salamonie, Mississinewa, Huntington, Cagles Mills, Monroe, and Mansfield reservoirs. bFlood Control 2 includes Flood Control 1 plus Big Pine and Lafayette reservoirs. 89 An additional basin-wide cost saving of $218,200 would be realized by the addition of Big Pine and Lafayette reservoirs in 1980 to the six reservoirs of Flood Control Set 1. These cost savings represent the net enhancement to the Big Pine-Lafayette flood plain lands after considering offsetting effects elsewhere in the Wabash Basin. The addition of the Big Pine-Lafayette project would result in the efficiency gains to the entire flood plain of about $0.53 per acre. If the efficiency gains are attributed only to the 400,631 flood plain acres enhanced by this project, the annual return would be about $0.55 per acre. The "high demand" analysis was made only to evaluate the effect of the addition of the Big Pine-Lafayette project to existing flood control levels. The cost savings from this project were $370,100 under high demand conditions compared with $218,600 under low demand conditions. Thus, the cost savings under high demand conditions would be about 69 percent more than would be realized under the low demand conditions. No additional flood plain land would be utilized on the project flood plain impact area (Table 16). Thus 409,946 of the pos- sible 412,814 acres of Wabash River flood plain would be utilized with or without the Big Pine—Lafayette project. The efficiency gains to the entire flood plain are $0.90 per acre. The efficiency gains attributable to the Big Pine-Lafayette reservoirs represents the net societal cost savings available to cover project cost for the agricultural sector in the Wabash Basin associated with investing in these two reservoirs. Total production costs of meeting the demand for commodities from the Wabash River Basin are 90 lowered, because output losses due to flooding are reduced substan- tially. This costs reduction is offset somewhat by slightly higher input costs, especially to be applied to flood-protected lands in order to sustain the increased yields over time. In addition to esti- mating efficiency gains accruing to a reservoir, the RLP-PE model can also provide estimates of probable land-use changes. A breakdown of on-farm costs by crOps are indicated in Tables 14 and 15. In both the low and high demand runs, reduction in basin- wide costs of producing soybeans are responsible for the bulk of the efficiency gains. 91 TABLE l4.--On-Farm Agricultural Production Costs, Low Demand Conditions, Wabash River Basin, 1980 Projected No : Cost : Six : Cost : Eight :reservoirs : change : reservoirs : change :reservoirs ($1,000) : (0 to 6) : ($1,000) : (6 to 8): ($1,000) ------------------------ Dol1ars----------------------- Feed grains Corn 242,735 - 13,191 242,722 11,132 242,733 Oats 23,949 14,091 23,963 543 23,964 Barley 653 - 1 653 0 653 Total 267,337 899 267,338 11,675 267,350 Roughage Hay 15,586 4,521 15,591 - 6,997 15,584 Silage 9,619 -151,361 9,467 - 51,866 9,416 Pasture 20,348 0 20,348 0 20,348 Total 45,553 -146,840 45,406 - 58,463 45,348 Wheat 31,659 - 89,889 31,570 378 31,570 Soybeans 126,535 - 82,709 126,452 -l7l,664 126,280 Total costs 471,084 —318,539 470,766 -218,474 470,548 92 TABLE 15.--On-Farm Agricultural Production Costs, High Demand Conditions, Wabash River Basin, 1980 Projected Six : Cost : Eight reservoirs : change : reservoirs ($1,000) : (6 to 8) : ($1,000) --------------------- Dollars--------------------- Feed Grain Corn 221,781 96 221,781 Oats 24,364 0 24,364 Barley 678 - 6,584 671 Total 246,823 - 6,488 246,816 Roughage Hay 13,235 0 13,235 Silage 25,351 -229,462 25,122 Pasture 20,639 6,964 20,632 Total 59,225 -236,426 58,989 Wheat 56,155 24,792 56,179 Soybeans 157,750 ~151,988 157,598 Total costs $19,953 ~370,110 519,582 93 Changes in Land Use The 1980 ”no development” solution obtained in the RLP-BP model indicates that there is excess capacity in the agricultural sector for 1980 under either low or high demand conditions. As indicated in Table 16, approximately 19 percent of the crOpland would be idle, as- suming no Corps reservoirs would be operational by 1980. The high demand increased both the cropland and pastureland use significantly. Although only 1.5 percent of pastureland would be idle in 1980 under high demand conditions, 5.5 percent or about 764,000 acres of cropland are projected to remain idle. The extent to which the two levels of flood control protection affect the Wabash Basin land use pattern for 1980 is indicated in Table 16. A comparison of the "no development" land use pattern and the pattern associated with the installation of the six reservoirs of Flood Control 1 indicate a 19,149 acre reduction in cropland required to meet anticipated 1980 Wabash River Basin agricultural demand condi- tions. On the project flood plain, however, land enhancement occurred through the conversion of 44,189 aCres of idle land to cropland because of the additional flood protection afforded by Flood Control 1. In addition, 490,909 acres of project area floodplain lands would be used more intensively. This represents approximately three-fourths of the 654,687 crOpland acres on the impact area flood plain. These lands, which produced corn and soybeans in the base situation, would continue to raise these crops, but with higher application of inputs (ferti- lizer, seed, and lime). While land enhancement occurred on flood plain land, a total of 63,285!up1and acres were idled because of the 94 TABLE l6.--Major Land Use, No Development Situation, Wabash River Basin, 1980 Projected Available : b : Share of Category landa : Land used : land ------- Acres------- --Percent-- Low Demand analysisd Cropland 13,911,727 11,316,200 81.3 Pasture 2,087,500 1,969,400 94.3 Total 15,999,227 13,285,600 83.0 . . d High Demand analysis Cropland 13,911,727 13,147,831 94.5 Pasture 2,087,500 2,055,771 98.5 Total 15,999,227 15,203,602 95.0 3Based on 1958 CNI data, less land withdrawn for urban expansion. bBased on the 1980 Wabash River Basin "no development" solutions. cOriginal demands. dDemands as modified by OBE-ERS in 1969. Sources: (1) Wabash River Basin Type II unpublished data for low demand analysis. (2) Wabash River Basin Comprehensive Study, Vol. IX, Appendix H, Agriculture, June, 1971, Table H-12, p. 64. 95 increased economic efficiengy associated with_producing on the flood protected lands below the six reservoirs. This equivalent to idling four rural townships which are comprised of about 70 percent cropland. The composition of lands idled is indicated in Table 19. In the low demand analysis, the addition of the Big Pine- Lafayette project resulted in a reduction of 9,186 acres in the basin- wide cropland requirement. On the flood plain below the reservoirs, an additional 8,293 crOpland acres would be converted from idle to pro- ductive use and 392,338 crOpland acres would be crOpped somewhat more intensively (Table 17). In the non-impact areas, 9,798 more acres of cropland would be idled. The land use changes under the high demand conditions were evaluated with respect to the difference between Flood Control 1 and Flood Control 2 only. The basin-wide cropland requirement was reduced by 7,710 acres (Table 17). The lands idled include about 27 percent roughages and 73 percent soybeans (Table 19). The distribution of crops were fairly stable within low and high demand categories as the Big Pine-Lafayette project was added (Table 18). In comparing the effect of demand levels, the high demand runs used more soybeans relative to corn (Table 20). 96 TABLE l7.-—Adjustments in Basin Land Use Due to Corps Reservoirs, Wabash River Basin, 1980 Projected Low Demand 3 High Demand Category Flood 3 Flood 3 Flood Control 13 1 Control 2b ; Control 2b ---------------- Acres--—------------- Basin-wide Reduction in cropland 19,149 9,186c 7,710 Non-impact area Reduction in cropland 63,285 9,798c 7,710 Flood_plain impact area Total acreage 823,350 823,350 823,350 Cropland available 654,687 654,687 654,687 Percent crop used 82.50 94.92 99.46 Land use changg_ l. Idle to cropland 44,189 8,293d 0 2. CrOpped more e intensively 490,909 392,338 409,946 aFlood Control 1 includes Salamonie, Mississinewa, Huntington, Cagles Mill, Monroe, and Mansfield reservoirs. bFlood Control 2 includes Flood Control 1 plus Big Pine and Lafayette reservoirs. CLands idled in addition to Flood Control 1. dIdle lands converted to cropland in addition to Flood Control 1 lands converted. eIncludes lands affected by Flood Control 2, not including lands converted per note (d) above. 96 TABLE l7.--Adjustments in Basin Land Use Due to Corps Reservoirs, Wabash River Basin, 1980 Projected Low Demand 3 High Demand Category Flood 3 Flood 3 Flood Control 13 : Control 2b 2 Control 2b ---------------- Acres------------—--- Basin-wide Reduction in cropland 19,149 9,186C 7,710 Non-impact area Reduction in cropland 63,285 9,798C 7,710 Flood plain impact area Total acreage 823,350 823,350 823,350 Cropland available 654,687 654,687 654,687 Percent crop used 82.50 94.92 99.46 Land use changg_ 1. Idle to cropland 44,189 8,293d 0 2. CrOpped more intensively 490,909 392,3388 409,946 8Flood Control 1 includes Salamonie, Mississinewa, Huntington, Cagles Mill, Monroe, and Mansfield reservoirs. bFlood Control 2 includes Flood Control 1 plus Big Pine and Lafayette reservoirs. cLands idled in addition to Flood Control 1. dIdle lands converted to cropland in addition to Flood Control 1 lands converted. eIncludes lands affected by Flood Control 2, not including lands converted per note (d) above. 97 .mouom voumo>nmmm O. maa.owm.ea mmo.eam.eH HRH.NNQ.NH emm.amm.NH oom.omm.NH on: eenfi Heooe mm.vm Ho.mm 5H.Hm om.Hm mm.Hm om: pcmfi acounom ooe.Hmo.m wom.mmo.m wem.emo.e ema.mmo.e nmm.eeo.e neeoomom 0H.NH mH.~H Hm.n Hm.n mm.n om: puma pcounom Hmw.mee.a mme.mea.a HHo.onm mmo.oem mmo.mam anon: wa.wa ma.mH wm.aH RN.mH om.aH on: eeeH scooped www.mao.m eow.wao.m moo.oa4.m ooH.Noe.N eme.eae.m Heooe noe.owm.a who.oma.a emo.~wa.H emo.mwa.H emo.mwa.H assumes aen.am~ mwo.mom Hme.eHH Hmm.mHH mem.eHH omeaam Hon.mmm Hoe.mmm eoe.eam Heo.eam Hae.emm an: owmnmsom mo.em mo.em eo.~e No.Ne ma.He on: papa ocooeoa wae.mma.e eee.awa.~| mno.mme.m ame.mme.m Nam.nme.m Heooe moo.om ooe.o~ ooo.wH ooo.mH ooo.mH Refinem Ree.aem Ree.fiem ooa.mmm eofi.mmm mHH.mmm mono wma.u~o.e Hmo.mmo.e mam.oeo.m Neo.meo.m oao.mwo.m eeou mcflmmwwoom .. ................................ mopo< ................................... whack/homo." mhwo>Hommh mhwotomOH mhwozommh mhwotomOH onwam H Xum H osmfim H Ram H oz H mHo>oH namEmv swam " mHo>oH vcmEou 30g " eoooohoea omaa .cfimmm uo>flm :mmpmz .mHo>og panama swam msmho> 304 .m:0wpfipcou Houucou woon o>wumchopfl< nova: mom: vam4-1.wH mameonou w m~o.emm.eH newosuonou o .H oeeEoo rm“: mmm.w~ eoaoa mfifida Hmuop HmuCOEOHUCH mw.Nm ma. mo.mH 00.4 owH.m oofioa apnea ARH.NNQ.NH omeao>uomoe m Rmm.ama.~a omeaosuomoa o .N Nm.om em.mH Re.NH eo.om meH.mH oofioa noses amm.Hmm.~H omewo>eomoe o oom.omm.mH mmufio>Homou mahou 02 .H ucmEop 304 .................... ucoohom---uu-u---u-u---uuuuu m m m a... m m mcwonxom H puma: H owmnmsom H poem . . woflufi mwcmfi eoooohoea omaa .camem eosum mo :ofiufimanou pom: mmuo< :mmnmz .mufio>homom mahou mo cofiuusnonucH xn voHpH cofiumsufim moses Hensoaooaom<--.ma mamcfi voxfiu v .oyom you moxmp Hmsccm mm mafia flwm @ oovmv ucmfi co amonoucfi om» co nommmo .mooouoemmooo ma-aam owns no power a .muoonoua N Honucou voofiuumlum ”muooMOHm H Honucou voonuauum mucoemofio>ou OCWQZm mm.~m Hm.nm om.oo : : : n.mm N-um : mv.om mm.om mm.oo : z : «.mm Huuu : oa.e~ om.mo oa.mo : : om.om o.mm oz meeooaom ofi.mH oo.HoH vw.Hw : : : N.om N-um : om.wH mo.mm mm.Hm : : : m.vm Huum : am.ma Rw.mm mo.om : : mo.me m.Hm oz enou ooeaeee aaeooa n~.om oo.moH mw.oo : : : n.Hv muum : mo.mm on.HoH No.00 : : : N.Hv H-0u : um.mm ow.mm mN.oo : : mm.Hm o.ov oz mcmonxom mo.m~ om.moH HH.mw : : : o.moH ~-uu : NH.VN mn.cofi no.mw : : : n.HoH Huuu : on.HNm -.m0aw ov.wa oo.on oo.mmm o~.ovm m.wm oz Chou vocfimpp HHoz MWMMM m mansuon m mumoo wwwmmw omoMHmnu W wmumoo m pm.snv wawmww W mono max» HwOm ensues H Heooe H Heooe nonoo papa H o beeee> H papa» oooaa H ahead eooaa ome .m monmnsm .mfimmm ouo< nod .CHmHm voon po>fim :mmnmz .mddOho Hwom awmfim voofim .mpcofiowmmoou cusuom use uncut-.vm mqmoo .oz H N-ua " H-6a " .>oo .oz nmmmm u done one show you ufimonm m whom you ufimonm nod m ogxp Hwom owmfi .m noeeosm .enefia oooaa eo>um apnoea .naaoeo Haom afiefia ooofia .nunafiee< apnoea flaw umoulu.mm mqmOU “.U.Q afioumchmNZV “cowwmoui Gnu. mo whomlmlm UwEOCoum .mhomw>fi< UHEOCOUM mo HMUCSOU "mohsom 0.H 0.H H.m 0.0 0.0 0.0 poo» eoa ucmouoa ommho>< mummm .lmmm .mqmmm n.00H mmmmm. 0.5HH 000a n.0HH 00H 0.0m” H.00H H.005 0.0HH 500a 0.00M 0HH 0.00H 0.00H 0.0HH 0.000 ooma 5.0HH HHH N.HNH 0.000 0.HHH 0.0HH 000H 0.00H 00 H.0HH m.NmH 0.000 N.NHH 000H m.NHH 00H 0.00H 0.00H 0.000 0.0HH moma 0.00H 00H n.00H n.0HH H.000 0.00H ~00H 0.000 00 0.000 5.00a 0.00a 0.50H 0000 0.500 00H H.000 5.00H 5.00H 0.00H 000a 0.00s 00H 0.000 0.00H 0.00H 0.00H 000H m.NOH 00H 0.00 5.00 4.00 5.00H 000a 0.00s 00 0.00H 5.000 0.00a 0.00a e000 - . H nooauemaaV H H . . mMWmMmRWMMW H A00Hu0m-emmfiv H esooea H n00Hn00-5m0H0 H h00Huemaa0 H fi00Hu500H0 H ..o: own H :ofiuospoum H Hmcomhom H usapso H ucoeonan H cod 0 ago H umo> o 0 . chasm 0660 . oaoemoonao H HensonoeeH H ceaaasau H .o H 0 H sham mmouo " u «venue you wounmmH .m.:numhoumowvcH owEocoom nouooflom How moofivcH nuonUIu.mN mqmo~ u:mEov zo~ nu“: mcofiufiucou :cho flouucou voofimz have: n Houucou poomm u N-ua .moUMAQ vouwfiasuoc Haucaou cousomom 6600: u z .owma on xfiaazccn ucouuoa n ommouucw o» vouuonoua mnaowx coma mahou u a .omaa oom~ .mofivaum :Owuaunfiuman uuonoua muoocwmem mo mauou u 0 ”60oz mv~.onm.nam oucouewwfio .u .mozfim> N-Ud ou o>flua~ou mos~m> usauzo maaou nmm nnm.vfi ~-ud a u m 66:00; aaoeeoauaeman 6666206u me~.0em.eaw 06 6606660000 000.046.500 z a 0 H 0mm.moo.m m oocokomwwo .n a 4: . .mvflowx mono mahou any on o>wum~ou moo oer “M 7 a u H mvfiofix N-Ua pozmwc muoofiwop 0mm.moo.wn mo oocouommfio «mo.HHm.mmm z n-ua u 0 “00.n0 0 6606u60050 .6 mm .meho vosfim> am“: we :0wuuonoaa gunman xdugmflfim mmo.wwn.fim z a u H 1L 6 60066006 0-06 6606 666665005 ~00.e00 66 6606660050 000.0H0.Hmm z a n-0a m mousuoooaa ceazuom moocouomuwo mo weaneduauumm .N mam.om~.oflm ougoaowwwo .6 .xuozoemuw Hesuaoucou ecu .o:~e> unnuao .ufiodx nouu . . - 1- - .mm: one” ucwpaamou mousuoooun can one coozuon moocouou mofi Hem mH N on e on N on m -wfiv mo uoomme we: on» muuofimoa mom.om~.o~u mo ouceuomwwo mmo.mvn.~mu z a u H oucoaommfio H~6u6>o .H unease m on a m o=~m> m m 04 H 603 :0 H o>a mayo . U H > ufiauso m: .> v a” .u Had .600d 000 36260 H6:0 06606: .0005 660 600600 0560a pooaa 0o 03mm.» HGHOH muwo>uomom ouuoxmqu mm-a4¢ ecu nahou coezuom moucouomwwo mo newum>wuuo--.om mammh 139 TABLE 37.~-Summary of Differences Between Corps and RLP-PE Total Value of Flood Plain Output for 1980, Wabash River Below Big Pine-Lafayette Reservoirs Difference : Description $17,370,145 Reflects higher Corps output values - 67,301 Shows FC-2 land use more intensive than Corps - 8,063,556 Indicates FC-2 crop yields higher than Corps $ 9,239,288 Explained difference — 16,186,995 Extent to which FC-Z less than Corps (Row 3, Table 35) -$ 6,947,707 Unexplained difference attributed to conceptual variations between the two procedures Source: Table 36. 140 .60600606 06660066 mmumqm any .606006060 0606660006600 6660606 0606009660: 606600w0m 06 66060 0N0 .0N 60060 000 0666006m .om60-mom0 0600 066% 060 6066066 m 06 66006060 00 6660 nuzoum 06 066600 .owm01m0m0 5600 0660 066 0066060 we. mo 60006060 00 6660 nusoam 06 0666m6 6006M 606600 06060m60 00 opo.pwmm 6606060000 60060 0606060066 06060662 00 000.060 06.600 6600606 606600 66600 n- 0000.0m00 60060 6060600066 006060 nwmm 600mm 656600 06060mom m0 pom.00mm 6606060000 60060 0606060666 06000662 mm 000.600 06.600 m6m0606 606600 mmohu u: £000.0m00 66060 6060600066 006060 360 6006w 606600 06060w6m mm pom.000m 6606060000 60060 6606060666 06000662 00 000.060 06-600 m6m0606 656600 66600 t. 600m.00mm 66060 6060000066 006060 0wmm 6006M 606600 060o0m6m 0m oo0.mmmm 6606060000 006060 606660600 006660066 0600: 6006m 0606060006 06060062 m0 oom.w0m mmumqm 006060 6066606 000660066 0600: m6w0606 606600 66600 .. 6oom.00m0 mayou 6060000066 006060 360 n606606av 6506> 606nm . 0006060 060um00666o 0 600066606 6660600 660606060 -6006 m0m .omm0 pom m6ums0umm 000606m 66060 006 mmnmqm coozuem 66606060009 mo KAumaesmuudm m0m .vaum o>0mcocomeou c000m Ho>0m £00903 "mounom 168 0.000 0.000 0.000 000.00 000.00 000.00 00000 0.0 0.0 0.0 00 00 00 00000 0.0 0.0 0.0 00 00 00 00000000 00000 0.0 0.0 0.0 0 0 0 00000000 0.0 0.0 0.0 00 00 00 000000000> 0.0 0.00 0.00 000 000.0 000.0 000 0.0 0.0 0.0 00 00 000 000 0.0 0.0 0.0 00 00 00 000000 0.0 0.0 0.0 000 000 000.0 0000 0.00 0.00 0.00 000.0 000.0 000.0 00002 0.00 0.00 0.00 000.0 000.0 000.0 00000000 0.00 0.00 0.00 000.0 000.0 000.0 0000 -------uuuuucoouom .......... . ---- ....... 00000 000.0 ........... 0000 m 0000 m 0000 0000 m 0000 m 0000 m . . . . 0000 0:009000 c0009 00000 mo ucoohom Quechua :Hmmm czoho macho cflmmm Hmmfiocfihm we ommouo<--.0-< mam00 000000005 :000000M00030 0000000 000mmo mapou 0000000o 0000>0wsog "oopsom .00c05000m .0oucoz .0000M0cmz .0000: 000mmu .coumcflucsm .030:00000002 00000m :0 0000>00000 09000 00000000090 x00 £0020 00.0 0 000.000.00 00000 --- --- 0.00 0.0 0.00 00.0 000.00 0000 00-2 0.00 --- 0.0 0.00 0.00 00.0 000.00 0000 00-2 0.00 --- 0.0 0.00 0.00 00.0 000.00 0000 0 -2 0.0 0.0 0.0 0.00 0.00 00.0 000.00 0000 0 -2 0.0 --- 0.0 0.00 0.00 00.00 000.000 0000 0 -2 ---- 0.0 0. 0.00 0.00 00.00 000.000 0000 0 -2 0. 0.0 0.0 0.00 0.00 00.00 000.000 0000 00 -2 0.0 0.0 0.0 0.00 0.00 00.0 000.000 0000 00 -2 0.0 0.0 0.0 0.00 0.00 00.0 000.000.0 0000 0 -2 ---- 0.0 0.00 0.00 0.00 00.0 000.000 0000 0 -2 0. 0.0 0.0 0.00 0.00 00.0 000.000 0000 0 -2 0. --- ---- 0.00 0.00 00.0 0 000.000 0 0000 0 -2 c090: 00>00 M cofiumpuoamcmnh M mono-:02 mono 0000 000 W 00000 W 0:00> M 0 000» 0 :000m fiucoou0mv 000000 mo :00ufi0anou 00000> 0000 000 0000 .00>00 000002 .000000 00000 000000 0000000 003::0 0w0~0>< 00000><-.0-< 00000 173 According to Boxley1 Kates, in his study of seasonality of flooding on the Ohio River Basin,2 fOund that flooding in the Wabash and White Watersheds was less seasonal (less concentrated) than in the other major watersheds of the Ohio Basin. The cumulative concentration of flood events by months did not exceed 70 percent until the month of May, and the mode of occurrence was March-April. Thus, flooding in the basin encroaches on the crop-planting and early growing season to a much greater extent than in other watersheds of the Ohio Basin. Flood Control in the Wabash Basin Private levees have been used for many years as a means of providing some means of flood protection. In the March-April 1913 flood, "every existing levee in the Wabash River basin was breached."3 Private levees continue to be used to afford protection to areas lacking publicly constructed levees.4 The earliest comprehensive report of flooding problems and other water resource problems was the survey completed in 1932.5 1Boxley, Institute for Water Resources, Wabash River Basin, IWR Report 69-4, p. 62. 2Robert W. Kates, "Seasonality," Papers on Flood Problems, ed. by Gilbert F. White, Research Paper No. 70 (Chicago, I11.: University of Chicago, Department of Geography, 1961). 3Wabash River Coordinating Committee, Wabash River Basin, Appendix-E, p. 35. 4There are 145 named levees in the Wabash Basin, with the majority constructed by private interests. In addition, there are other small unenumerated levees. Source: Existing Levees, Map #1, Wabash River Basin Emergency Flood Control Activities, January, 1966. Obtained from Louisville District, U.S. Army Corps of Engineers. SU.S., Congress, House, Wabash River and Tributaries, Indiana and Illinois, H. Doc. 197, 80th Cong., lst sess., 1932. 174 This survey according to Boxley "found that improvements of the wabash River by the Federal Government were not advisable at that time."1 The 1944 survey report recommended flood control by levees as the most feasible method.2 Since the 1944 report, which led to the construction of several levee projects, emphasis switched to multi-purpose reservoirs as the primary flood device. Since single-purpose flood control reservoirs were fbund to be marginally above unity in benefit-cost analysis, project authorization was more likely with multiple uses. The Corps reservoirs authorized in the 1950's and 1960's were generally multiple purpose reservoirs capable of providing flood control plus low-flow augmentation and water-based recreational activities. At the present time, six Federally-financed Corps reservoirs are operational in the Wabash Basin: Cagles Mill, Mansfield, Monroe, Mississinewa, Salamonie, and Huntington. The Cagles Mill was the first one having been completed in 1953, the Huntington is the most recent with its 1969 completion date. Construction of five additional Corps reservoirs was authorized 3 in 1965, but to date the Patoka is the only one under construction. 1Boxley, Institute for Water Resources, Wabash River Basin, IWR Report 69-4, p. 62. 2U.S., Congress, House, Wabash River and Tributaries, Indiana and Illinois, H. Doc. 197, 80th Cong., lst sess., 1944. 3U.S., Congress, Senate, Lafayette and Big Pine Reservoirs, Wabash River Basin, Indiana, S. Doc. 29, 89th Cong., lst sess., 1965. U.S., Congress, House, Lincoln, Clifty Creek, and Patoka Reservoirs, Wabash River Basin, Indiana and Illinois, H. Doc. 202, 89th Cong., lst sess., 1965. 175 In addition, a Corps report published in 1967 found favorable consid- eration for five more reservoirs.1 The last in the series of Corps reports dealing with a com- prehensive survey of water resources in the Wabash Basin is the final report.2 The findings of the Corps were that: One hundred and eighty seven potential major multipurpose reser- voir sites were screened in the Wabash Basin as a part of the general appraisal and preliminary screening studies. This re- sulted in fifty potential reservoir sites being selected for Phase I study as potential alternatives fbr meeting defined water resource needs. Sixty possible levee units were con- sidered in the general appraisal and preliminary screening studies. Engineering and economic analysis of the levee projects not dis- placed by recommended reservoir projects indicated that only ten levee units were economically justified in second position to the recommended system of major reservoirs and watershed projects.3 The June 1971 report was also significant in that consideration of non-structural alternatives, in particular flood proofing, were seriously evaluated fer the first time. Detailed analysis of the feasibility of flood proofing two communities was undertaken. In the Villa Grove, Illinois study the Corps evaluated the costs of raising houses to the 30-year flood level, the maximum of record at Villa Grove. At Anderson, Indiana, flood proofing to the 100-year level was analyzed.4 1U.S. Army Engineer District, Interim Report No. 3, Vol. III. 2Wabash River Coordinating Committee, Wabash River Basin Comprehensive Study, Vol. XIII, Appendix L, Project Engineering Studies (June, 1971), pp. 1-43. 3Ibid., p. 4. 4Ibid., p. 13. See wabash River Basin Comprehensive Study, Appendix D, June, 1971, for additional details on the flood proofing evaluation. U. S. DEPARTMENT OF AGRICULTURE ~ VIEIIIITY IMP E SOIL CONSERVATION SERVICE --——— (3'; 0 In“ if HIEHIEAN U" 1"" I __—I_— | . . q, ‘ INDIANA J" IJNID Tr,’ ILLINDIS L I “Tn FORT :1 N LEGEND ..ssm- Z IABASH STATE BOUNDARY .1 LIVINGSIDN J COUNTI BOUNDARI __ - BASIN BOUNDARY _/ \/ IROQUOIS anNS AND CITIES I % 0 254; STREAMS fl: .20 TIPTnN LAIIEs [JR RESERVOIRS m “HEN “CLINTON _""I NAL 'I an R ...—l—‘L— NATII] FDRES UNDA I “0|an INTERSTAIE NIGIIIIAII . , u. s. NIGNVIAI -__.._____ FAVETTE O IUULTRIE SHELBY nEcATun\ I SULHIVAN CRAWFORD l SAXNTE DAVIES; IASNINGTON EDIARus JASPER DuaaIs . mason Figure A—l WABASH RIVER BASIN ILLINOIS, INDIANA AND OHIO SCALE I/I,7oo,ooo SHAWNEE u. 5. DEPARTMENT or AGRICULTURE Saurcr NOTE NAT‘ONAL SCALE I0 o In 20 30 no MILES ECON‘jrgfiggiflggcl:ERV'CE . “SW-""9“- 5v""-°“' INTERSTATE NluIIIIAIs As sNDwII IIII THIS MAP HIE FOREST son. CONSERVATION SERVICE "I; -‘5-“ 33:52:?.mz;.::?::"-§:X?§-23‘I:- COMPLETED. uNuEII EUIISTRUCIIUII an PIIUPUSEII 176 5,: 25'8“” Field Yuhnichnl. USDAvSCSvLINCDLN. NEBR. I565 SOIL CONSERVATION SERVICE REV. 9'1-70 5,8-26, I25 Std-IKE: scs Dunn-g Numb" 5,&15,671 ma inlormuuon 'urrilhnd by FIeId Tnchnieinns. LEGEND STATE BOUNDARY __.___ COUNTY BOUNDARY ——-—-— BASIN BOUNDARY _./"‘\./ TnIII-Is AND CITIES O & STREAMS f6: LAKES CIR REsERvuIRs LAND REsouRcE REGION BOUNDARY N IIAJnR LANu RESOURCE AREA BOUNDARY /'\/'\ KEY TO LAND RESOURCE REGIONS MAJOR LAND RESOURCE AREAS LAKE STATES FRUIT. DAIRY REGION TRUCK. AND snIITHERN MICHIGAN DRIFT PLAIN CENTRAL FEED GRAINS AND LIVESTOCK REGION IIEI IE ILLINOIS AND IONA DEEP LDESS AND DRIFT ES NORTHERN ILLINOIS AND INDIANA HEAVY TILL PLAIN INDIANA AND OHIO TILL PLAIN II3 cENTRAL CLAYPAN AREAS SOUTHERN ILLINOIS AND INDIANA THIN LOESS AND TILL PLAIN CENTRAL MISSISSIPPI VALLEY HOODED SLOPES EAST AND CENTRAL GENERAL FARMING AND FOREST REGION .20 KENTUCKY AND INDIANA SANDSTONE AND SHALE HILLS AND VALLEYS HIGHLAND RIM AND PENNYROYAL LmneII EcnlumaI Guru: PluIEDIIuI- YIEINITY IIAP t g ILlINDIS L 3. IISSWII ea"oo' ‘ PAXTUN CHAMP-A I UN UREA-VA N0”OO' RequRIE I 33°DO' _‘_7 WOOL-DIS 19% r" ‘0 o 0 0 U 8. DEPARTMENT OF AGRICULTURE \ ILLINOIS \ INDIANA ES Figure A—Z _- ' ‘ WABASH RIVER BASIN H ILLINOIS, INDIANA AND OHIO COMPREHENSIVE BASIN STUDY lAND RESOURCE REGIONS AND MAJOR lAND RESOURCE AREAS US. DEPARTMENT OF AGRICULTURE JULY 1970 Economic Research Service. Forest Service and Soil Conservation Service ..ERR‘JRGR EVANSVILLE SCALE I/I ,700,000 40 MILES 177 \0 USDA-SCS‘LINCOLI. NE". 1970 PLATE H-4 U. S, DEPARTMENT OF AGRICULTURE VILINIIY m SOIL CONSERVATION SERVICE KANKAKEE OUTWASH woo-i 9’ 4 0 Z Z 4 j 0 DESCRIPTION or THE PHYSIOGRAPHIC AREAS 3 3 random: / TIPTON TILL PLAIN - Nearly flat to gently rolling gla- MUSCATATUCK REGIONAL SLOPE — Broad and flat to gently cial plain traversed by several low terminal rolling. dissected upland on glacial drift over moraines. Mainly ground-moraine deposits with ’ddle Paleozoic limestones and dolomites “mum TIPTON some end-moraine, valley-train and outwash—plain SCOTTSBURG LOWLAND - Nearly level to gently rolling depoi . topography on glacial drift over late Devonian and BLOOMINGTON RIDGED PLAIN - Low, broad, morainic ridges with intervening wide stretches of Iy Missrssippian nonresistant shales. NORMAN UPLAND - Dissected plateau of strong relief characterized by flat-topped narrow divides relatively flat LIINION tly rolling ground- moraine. SPRINGFIELD PLAIN - Flat to gently rolling upland char- pAPRE RANDOLPH IIIINTEITIIERI fIIUHTAIN steep H " slopes, and V-shaped valleys developed on rela— acterized by relatively well-developed shallow tIvely resistant siltstones and Interhedded softer x O entrenchment of drainage Into Illinoian drift over— shales of the Borden Group of early to middle Mis- E - lying Pennsylvanian sandstones. shales, coal, and sissippian a e. g I limestones. MITCHELL PLAIN — Essentially a rolling plain developed = rIrrrrE O MOUNT VERNON HILL COUNTRY - Rolling topography on thin on middle Mississippian limestones. Sinkholes, _ lllInoian drift over Pennsylvanian sandstones, and disappearing streams are comm mm”, /' shales, coal. and limestones. Valleys extensively - CRAWFORD IJPLAND - Rugged angular topography 0of strong ATUCK aggraded with Illinoian till glacio-lacustrine, relief developed on interlayered sandstones, and glacIo-fluvial deposits. shales, and limestones of the Chester Series of REGIONAL KANKAKEE OUTWASH PLAIN - A nearly flat area of ground- late MIssissippian age and sandstones and shales SLOPE moraine, outwash-plain. glacio-lacustrine, glacio- of early Pennsylvanian age. Areas of karst fea- fluvial and sandy eoIIan deposits. es including caves are common. STEUBEN MDRAINAL LAKE AREA — Uneven, morainic topog- WABASH LOVILAND — Rounded land forms of comparatively raphy with numerous lakes of kettle origin within - lIttle relief developed on thin glacial drift over ”W moraines or outwash-plains. Pennsylvanian shales and coals. Valleys are exten- sively aggraded by deposition of illinoian till, glacio-lacustrine and glacio-fluvial deposits LOWLAND MAIILTON IARRICK \0- MAP MODIFIED FROM: L ton, M. M. Ekhla E . and Horbe i94 Physiographic divisions of Illinois: Iour. Geology V. 56, No ne, V1.1, 1956, Thickness of drift and gbedrock physiography oi Indiana north of the Wisconsin gia acial boundary. SECTION tMODIFIED FRO I I . I, p. I6-33. . Rept. ol Progress No. 7, Indiana Geological Survey. . B. aznd others, 1956, Geologic Map uI a, Indiana Geological Survey. WeIler. J. M. and others 1945, Geologic Map of Illinois, Illinois Geological Survey. "d~ WISCONSIN GLACIAL BOUNDARY \ .00....0'0.o ILLINOIAN GLACIAL BOUNDARY 0 CRAWFORD MITCHELL NORMAN SCOTTSBURG MUSCATATUCK MOUNT VERNON HILL COUNTRY WABASH LOWLAND UPLAND PLAIN UPLAND LOWLAND REGIONAL SLOPE d SCALE 25 o 25 so MILES 2 5 2 °_ <1 onx "5‘5” '75:; :32; EAST roux Int/r5 fill/ER El“ ”NIT! RIVER I Ion/£1: A OUATERNARY TILL. CLAY.SAND AND snszL—N i I SCALE ”2.500.000 _ ..-0’ I000 '- —— , Figure A—S WABASH RIVER BASIN . ' " PHYSIOGRAPHIC MAP ILLINOIS, INDIANA AND OHIO 2000 est“; 7 . - - ‘Mi SISs’IPPIAN L‘“ __ .07 3000 U. 5. DEPARTMENT OF AGRICULTURE ECONOMIC RESEARCH SERVICE FRO EST SER soon SCHEMATIC SECTION SHOWING RELATIONSHIPS OF BEDROCK TO PHYSIOGRAPHY SCSE Drawing Number 5. S— 25, 671. and information fu rn nished by Field Technicians. VIC SOIL CONSERVATION SERVICE 178 m. M... 5.5-NJ“ USDA-SCS-LINCOLN. NEBR, I968 APPENDIX B SELECTION OF STUDY AREA APPENDIX B SELECTION OF STUDY AREA Four primary criteria were considered in selecting the down- stream flood area to be evaluated in the empirical investigation. First, the agricultural flood plain lands in the downstream project benefited flood plain must have been evaluated recently with respect to both hydrological characteristics and damage estimates due to flooding. Second, the acreage of the impact area must be sufficiently large so that statistical difficulties associated with using Con- servation Needs Inventory (CNI) land use data will not bias the out- come. Third, the flood control project must cause a significant re- duction (5% or more) in average annual crop losses for all areas to be included in the evaluation. Fourth, the reduction in flood fre- quency due to the flood control project will likely cause enhancement benefits through the conversion of idle land to productive use and through the more intensive use of existing cropland. Louisville District Corps office hydrologic and economic data were collected and reviewed for a number of Corps flood control pro- jects which were either completed or in advanced stages of planning. The projects considered in Illinois included the authorized Lincoln reservoir and the proposed Louisville and Helm reservoirs. Indiana Corps projects considered included twelve authorized or proposed 179 180 reservoir projects located on major tributaries of the Wabash River. These included the Big Pine, Lafayette, Patoka, Downeyville, Big Blue, Big Walnut, Annapolis, Eel River, Danville, Tippecanoe, Clifty Creek, and Shouls reservoirs. As single reservoirs were considered, it became readily ap- parent that none of the single reservoir Corps projects was able to exert a statistically significant effect over a sufficiently large impact area to meet both the CNI minimum acre requirement of 200,000 acres and the significant flood loss reduction requirement (see cri- teria two in this section). The authorized Big Pine and Lafayette reservoirs, located on major tributaries of the Wabash River, were considered jointly in Corps hydrologic and economic computations and were selected for this study. This set of reservoirs was found to satisfy all four selection criteria. First, the project impact area has been evaluated since 1960.1 Second, the downstream flood plain contains over 400,000 cropland areas (affects all reaches below W-6). Third, the hydrologic effect of the combined Big Pine-Lafayette project is sufficient to reduce damages in Wabash River reach W-l, the most distant impact area evaluated, by over 9 percent (1960 values-~see Table B-l). Fourth, there is a reasonable expectation that enhancement benefits will be realized if the Big Pine—Lafayette project is 1U.S. Army Engineer District, "Review of Wabash River Basin Covering Reservoir Sites on Wildcat, Big Pine and Sugar Creeks, Indiana for Flood Control and Allied Purposes" [Survey], Interim Report No. 1 (Louisville, Ky., March, 1963). U.S., Congress, Senate, Lafayette and Big Pine Reservoirs, Wabash River Basin, Indiana, 5. Doc. 29, 89th Cong., lst sess., 1965. 181 TABLE B-l.--Crop Acreage and Flood Protection Index, by Selected River Reaches, Wabash River Basin Production indexa Stream : Reach : ac::::: : acgzzge _ Flood . Flood ‘ ' 2 Control 1 2 Control Wabash River W— 1 108,000 70,848 .1265 .2157 W- 2 114,000 89,376 .1609 .2694 W- 3 99,700 79,062 .4007 .5164 W- 4 154,500 117,111 .3705 .5528 W- SA 19,700 17,099 .3953 .5619 W- SB 13,700 11,892 .4271 .6185 W- 6 16,200 14,110 .3972 .5997 W- 7 14,600 13,316 .6185 .8035 W- 8 3,800 3,165 .6766 .8309 W- 9 19,500 16,965 .8804 .8804 W-lO 11,600 10,451 .9300 .9300 W-ll 3,600 3,243 .9500 .9500 aProtection index is the reduction in average annual dollar crop damages expressed in decimals (1960 prices). Flood Control 1 includes Salamonie, Mississinewa, Huntington, Cagles Mills, Monroe, and Mansfield reservoirs. Flood Control 2 includes Flood Control 1 plus Big Pine and Lafayette reservoirs. Source: Unpublished data, Corps of Engineers, Louisville District. 182 installed. This is particularly anticipated in the lower reach of the Wabash River, as indicated in the Congressional document contain- ing the justification for the Big Pine-Lafayette project: Below the White River, the flood plain contains many large tracts . . . which are uncultivated or cultivated only inter- mittently because of frequent flooding or prolonged in- accessibility. Cultivated lands in flood-free years yield high crop returns; however . . . some of these lands are only cultivated one or two times every five years.1 p. 1U.S., Congress, Senate, Lafayette and Big Pine Reservoirs, 30. APPENDIX C FLOOD CONTROL PLUS OPTIMAL DRAINAGE ALTERNATIVE APPENDIX C FLOOD CONTROL PLUS OPTIMAL DRAINAGE ALTERNATIVE In a previous section the effects of the "flood control only" alternative were evaluated. A second alternative was evaluated, based on the assumption that drainage would be required on inadequately drained flood plain lands to realize increased crop yields and land conversion that is anticipated as a result of flood protection. 0n- farm drainage costs were included to represent the additional costs that would be borne by farmers if they are to realize higher yields on the protected land in the flood plain. Efficiency Gains Treating flood control and drainage as a joint development on project affected lands resulted in an efficiency gain in 1980 of $201,200 from Flood Control 1 (Table C—l). An additional efficiency gain of $109,300 would be realized with the increased flood protection afforded by the Big Pine and Lafayette reservoirs. The efficiency gain under the flood control plus drainage assumption was only half as large as was estimated under the flood control only assumption. This was due to the fact that additional on-farm drainage costs are 183 184 TABLE C-l.--Effect of Flood Control plus Optimal Drainage Alternatives on Agricultural Production Costs, Wabash River Basin, 1980 Status of E Total 3 Incremental E Incremental flood protection : on-farm costs : difference : change ---------- Dollars-----—---- --Percent-- No development $471,136,200 ---- ---- Flood Control la 470,935,000 $201,200 .043 Flood Control 2b 470,825,700 109,300 .023 aFlood control plus optimal drainage of project area flood plain lands with six reservoirs--Salamonie, Mississinewa, Huntington, Cagles Mills, Monroe and Mansfield. bFlood control plus optimal drainage of project area flood plain lands with eight reservoirs--Flood Control 1 plus Big Pine and Lafayette. incurred which tends to increase unit costs. In some cases, unit costs fer some "flood control plus drainage" soil groups were actually higher than unit costs for the "no development" alternative on the same soil groups. This implies that the provision of flood protection to some inadequately drained land does not automatically result in increased crop yields and/or land conversion. Including the cost of draining the land reduces the relative comparative ad- vantage of this land. Other soil groups in the Basin can be used to meet the projected output level at less cost than if drainage costs are incurred in order to realize the higher potential crop yields from flood protection. 185 Changes in Land Use The "flood control plus drainage" alternative would reduce the amount of cropland required to meet the anticipated 1980 agri- cultural demands (Table C-2). The addition of the Big Pine-Lafayette project to the Flood Control 1 reservoirs under "flood control plus drainage" conditions would reduce cropland requirements by 2,572 acres. No additional lands would be converted to higher value uses, and only 4,872 acres would be cropped more intensively. In addition, 2,575 cropland acres would be idled throughout the Basin. These changes in land use were less than the changes under the "flood con- trol only" assumption, because of the higher unit costs on flood plain lands under the joint develOpment assumption. The correSpond- ing land use changes under the "flood control only" assumption were as fellows: (1) Reduction in basin-wide crOpland--l,505 acres; (2) idle flood plain to cropland--8,293 acres; (3) flood plain cropped more intensively--60,201 acres; and (4) basin-wide cropland idled-~9,798 acres. 186 TABLE C-2.--Adjustments in Land Use due to Big Pine-Lafayette Reservoirs under Flood Control plus Drainage Assumption, Wabash River Basin, 1980 Category ; Flood Control la ; Flood Control 2b -------------- Acres—------------- Flood plain Total acreage 823,350 540,400 Cropland 654,687 412,814 Land use adjustments Flood plain: Idle to cropland 11,978 0 Cropped more intensively 180,339 4,872 Basin-wide Cropland reduction 7,929 2,572 Cropland to idle 9,469 2,575 aFlood Control 1 includes Salamonie, Mississinewa, Huntington, Cagles Mills, Monroe, and Mansfield reservoirs. bFlood Control 2 includes Flood Control 1 plus Big Pine and Lafayette reservoirs. APPENDIX D WABASH BASIN LINEAR PROGRAMMING MODEL APPENDIX D WABASH BASIN LINEAR PROGRAMMING MODEL The basic analytical tool is a regional cost minimization LP model. It was utilized to determine total on-farm costs of pro- duction and associated land uses in 1980, as required to meet speci- fied Wabash River Basin demands for the major agricultural commodi- ties. These demands are expressed in terms of bushels for wheat and soybeans and feed units for the other major field crops. "Low demand" analysis levels are shown in the equations below. (1) Minimize Z = C1 X1 + C2 X2 + ... Cn Xn where Z = total on-farm production cost excluding any pay- ments to land and management. Subject to: X X ..., X l’ 2’ n C , C , , C = costs of production per acre for various 1 2 m . potential X1, X2, .., X land uses. n X1, X2, , Xn = acres of various land uses: by crops, land capability unit groups (LCU's), land resource areas (LRA's), economic subareas, and water development activities (level of various activities). 187 188 The commodity demands for each of the nine specified commodity groups was specified in the following form: (2a) Feed grains (corn, oats, and barley):1 . 2 a11 X1 + a12 X2 + ... + a1n Xn :_d1 - 290,912,416 feed units (2b) Barley: a X + a X + ... + a Xn 3_d2 = 460,080 feed units3 2n (2c) Wheat: a X + a32 X2 + ... + a3n Xn :.d3 = 37,986,000 bushels (2d) Soybeans: a X + a X + ... a n X > d 150,764,000 bushels (2e) Silage: a X + a X + ... + a X 000 feed 51 I IV a. n 7,772 units‘I (2f) Alfalfa hay: a61 X1 + a X + ... + a X |v d6 = 22,417,274 feed units5 . 6 (2g) Oats. a71 X1 + a72 X2 + ... a7n Xn Z_d7 - 6,336,000 feed units 1Corn, oats, and barley were permitted to compete for meet— ing the total feed grain demand. 2One bushel of corn provides .56 feed unit. 3One bushel of barley provides .43 feed unit. 4One ton of silage provides 4.0 feed units. SOne ton of alfalfa hay provides 11.0 feed units. 6One bushel of oats provides .29 feed unit. 189 (2h) Other hay: a81 X1 + a82 X2 + ... + a8n Xn Z_d8 = 760,090 feed units = 39,000,000 feed (2i) Pasture: a X + a X + ... + a X > d n - . 8 units 91 l 92 2 9n 9 Where: 9) 00 :1 ll amount of product (feed units or bushels) supplied from a unit of activity (har- vested acre). 311’ a21, ..., d1, d2, ..., d = commodity demand for each of the nine specified commodity groups. Land availability restraint: (3) b11 X1 + b12 X2 + ... + b1n Xn :_r1, 21 1 22 2 "' 2n xn - 2 sl 1 $2 2 "' sn xn - 5 Where: b b ..., b = acreage of land required to supply one harvested acre for the activity. r = amount of land available for each set of activities utilizing the same land. Combinations of: (1) no additional water and related land development and (2) flood protection were permitted to com- pete for specified land availability in the solution. 7One ton of other hay provides 8.0 feed units. 8One animal unit day provides .15 feed unit. 190 Subarea minimum restraint: (4) e X + 3 X + ... + e X 3_w 11 l 12 2 1n n l + + .., > e61 X1 e62 X2 + e6n Xn — "v Where: e11, ... evn = yield of activity from one acre toward minimum production requirement for each economic sub- area for crops. wl, wz, ... wv = minimum production requirements for each subarea. Computer Analysis The "no development" problem was comprised of a matrix con- taining 1,069 rows and 4,893 columns. The LP problem was run using the IBM System 360, Model 65/75 computer, and other facilities at McDonnell Automation Company in St. Louis, Missouri. The least cost solution for the basic or "no develOpment" solution was obtained in approximately 40 minutes. The revised land and resource coefficients for the four flood control alternatives were entered as additional rows and columns to the existing matrix. This included 43 rows and 645 columns as well as 43 right-hand—side revisions to incorporate flood plain lands affected by the eight reservoirs. Using the IBM revise procedure, the four solutions were obtained in a total of 10 minutes of computer time or an average of 2.5 minutes per revision. 4520 7IIIIIIIIIIIIII II 3 1293 03 III "II All IIIIIII'