LIBRARY l ', N. ._ 3 1293 10242 5315 cant-may This is to certify that the thesis entitled AN ECONOMIC AND FINANCIAL PROJECTIONS MODEL OF THE U.S. FARMING SECTOR presented by Timothy Guy Baker has been accepted towards fulfillment of the requirements for _P_lL_D_.__ degree in figm— Major professor ii‘A/TAL Il/j [31:54}, / Date—MIML 0-7 639 ‘ - Q ~-- h” -i- -..~ .‘0 *- ’--*.1 +‘ "Y‘ k4 "'1 ufé‘j III III III.IIIIF'NIII*. 'Wpfll’lflndln' AN ECONOMIC AND FINANCIAL PROJECTIONS MODEL OF THE U.S. FARMING SECTOR , By S _ ' ' arch-‘4: -. . A Timothy Guy Baker ’ KIWI: ‘H I VT: ‘ ‘ 315d 1." ""“' ‘ , A r J; ( loud : .‘u’ Lfir _, _h . Iii»...- io'r: umu' " ' A DISSERTATION far 5m min: 1..< [1"1‘ (meek! I I , Submitted to acted hy‘ "1' l Michigan State University . in partial fulfillment of the requirements center for the degree of DOCTOR OF PHILOSOPHY enactment of Agricultural Beauties fi-J ; vf ‘ " butcher-Ls .y‘ - 5‘1938 .l «e Stqtmm:., mn’; W» I'MIttMs-mms -e.w=‘.t.:~w W ‘é‘flrtal ga [as an“ -'-§}:1li11' zuw‘u‘i ’34‘ 2W3? ‘ ' ‘4'“: =13" - Ated from the” tteemjial ”in fat pcafid‘emmu. r, - ' ' 1--- . f. .90 .c5e3eka.3 J! n . . . s.- :‘| ' “7 ‘ ‘ _ O .l 3‘» L\. . ‘- 7 V! ‘ A... ‘kfl'r‘w ' 5‘2”“ V... . ““5: “Witt. :0 fawn n-‘A-s sectcr. * 4 In Lh‘s SouC: “"v “('ELJR' ' P MC {or use flat u~ . Etfits Cf ."" .- k, r r. Afit ABSTRACT AN ECONOMIC AND FINANCIAL PROJECTIONS MODEL OF THE U.S. FARMING SECTOR By TimOthy Guy Baker Decisions affecting the financial structure of the U. S. farming 13;. ' . Ere continually being made in the public and private sectors. "Whiledge of the relationship among system variables, both controllable s '1 O'ncOntrollable, is necessary for informed decision making. Addi- ,3; anally, prOjections of system performance under alternative policies £5 Of capital’fgains and capital formation. “Numerous ”analytical A: :“3 w .r.e:.‘e:re:1;3. :.'. —‘4-O.- . u ._.I L...vh hL'uoa Agdk r-‘*~-_—~, r- ~ y: . .. {.1‘5 \ =.. . .-1_..O";‘_ _'.3 J u“ sealib \‘AJ 5- 0'. p 9" - u... 1...: “Ct OAL‘B U: 'n. ‘ a e... n . r7 . J but lA-rdL5‘ w - r p». : .' on HVULA .r.\.. v l‘n - ' . s-L'Hruant E}; J! n». ‘..| C lfltllde the r ‘6‘ a :J.‘ ‘HA "‘dk‘uno PIDJECtiQCS ar 5‘1“ . Bug—“M g Nu) IL. _ at 1- . ‘44.,A us of dry- 1.“: ’Le , ‘nl "‘~ 0°59 Sir“, \uljr y. .LQ‘HI fiuhi ts fmT _ s: JCk ',_1nor modifications. Equations were estimated using econometric » "u: II‘EndO‘genous variables in the model include the supply and utiliza- ::ig§5§?duantities and prices of aggregate crops and livestock. These The model includes equations for projecting intersector flows :uSR-fiseociated with entering and exiting proprietors. Consumption of 'us-‘ 4! §hmndurable goods and services by farm operator families and other .V‘ NgggéLOf funds are also endogenous variables. These components are on very weak data. fidfibtcdnc exogenous variables in addition to those mentioned kciuée‘ the rate oflinflation, gross ““10““ Pmd‘mt’ and "'5' iOEOEECtions are.made under several scenarios. A scenario with “none variables at levels considered likely to occur is the base Alternative scenarios are generated by changing one exogenous "ble at a time. The alternative projections give indications of ‘. In...h ‘ o ':h.u A‘l :,_' B“’As " u h». . wing—[3r EV. ' ‘3%5 av M Dgfgx .3 ~9 ‘ .t‘ L. n ' W :Y‘. L ier Iiiic o. .4 \ Isl C”n‘:“ ”“ttu year. In comparison to the base scenario, substantial reductions .' - ‘CL ; -*T. Net capital formation for the sector is projected to be small. . u: ‘ ari, net capital formation for continuing farm proprietors is very This is a result of continued purchase of assets by continuing ‘ ietors from proprietors exiting the sector. The resulting large "dd . .. w. ”3.3:, 5“ “ ‘59uuu~t I s. 't'“b‘r -: - At~wr‘s L“ ‘ ‘ ~"_ ;' 198 £17.41]. D 01A\.'. . Lin-321:5, v. I , .“vo.-’.u, ' .--.s. “‘3‘.‘ ‘ . . ‘ , v5.14“. or : ":- "KAN.” ACKNOWLEDGEMENTS The first work of thanks goes to Professor John R. Brake, my major 'fesaot, for his encouragement, advice, and support during all phases Imy graduate program. Appreciation is also expressed to the other ‘~ .v- p \ I. k . ’4 A A ~-- Jus:i“‘2: XE:E.:’.4“_:£: Criteria :1 Praise-3U“. CCUCIulifig 53-2112 III ‘ L; Intrcdgcti. Agétééate F Flog C. E17?- 51 Lixjfix IV ‘ Sa's: DéfinitiOn O. Aar: CPfi NOHLPGY; Farming TABLE OF CONTENTS '.('.;J"I:13TOFTABLES IIDOIIII_OOII oooooooooooooooooo ...-..........u..-.s..Viii ‘1_ LISI‘OFFIGURES IOOOOCOIIIOJOIIOIOIIIO .......... no.0... nnnnnnnnnn Oxiii 1bf.1h§APTEB I - Introduction ‘ W Justification .......... ......... ... ....................... ... ose .............. .......... ‘.... ...... . ............... .... Research Objectives ...... .... ..................... ... ...... Dissertation Organization ....... .............. ‘ ........ ....... “bu-3H Methodological Approach ............................ .. ...... .. 7 Criteria for Objectivity ........ ........................... .. 9 Procedures ................. ...... . ........................... 10 Concluding Remarks .. ..... ...................... ..... .. ....... 12 ';..-.; " “ III - Literature Review Introduction ................................................. 13 ,1 Aggregate Financial Analysis .......... ..... .... ......... ..... l3 . Flow of Funds .......... ..... ........................... 13 Emphasis of This Research ............................... 20 1 Economic Models of the Aggregate Farming Sector .............. 21 concluding Remarks .............................. ..... . ...... . 22 'ny'fv - System Definition and Theoretical Model 1 Farm Operator Households ................................ 25 Noncpérator Landlords ................................... 26 Farming Businesses ...................................... 26 Sector Interface (Intersector Flows) .................... 26 11A_ Systems View of the Farming Sector ......................... 27 ;:¥E1rl Level Structure ......................................... 32 " industry Structural Equations ................................ .35 t: IIOOOOOICIOIIOOOIOOIICII.IOIOII.IOOOI'IIIOOOIOOI'Q ‘6 ffinmn .....IIOOICOOOIOOII00-01.......QIQOCDOCIII,I‘IO so ‘111 p ‘ Q,‘ a? .__ Ut.11 ‘E 3": .I I '- - - ' nitrfficil‘ 1‘ \l.-- ,V'" uw-t-MA ' ov "‘I".""h v 1v- -.r ~I‘ - n- u... ....\ L..t r-J,.~>l lehte’uu\5‘ '9 "1 - I|.'tst~L—‘" h“n - It J5“ :13 EC!" iv Feed Supply ......... .................................... Seed Supply ........ ................ ..... ........ ...... Demand for Hired Labor ....... ... .......... .... ......... . Fertilizer Demand ..... ................. . ........... ... Demand for Other Nondurable Inputs ... ...... . ............ Demand for Durables ..... . ........ . .......... . ....... .. Deflation of Prices and Money Flows .. ................. .. Alternative Approaches . ............. . ................ .. ...... Summary ............ ................................. . ........ CHAPTER V - The Empirical Model Introduction .................... ......... . ......... ... ...... Time Index ...... ..... ......... .............. ..... ...... Data Period ................ .................... .. ...... Statistical Considerations ......... ......... ............... . Crop Supply ....... ........... ...... ......... . ............ ... Demand for Crops ............ ........ .. ..................... . Food-Industrial Crop Demand .......... ........... ....... Feed SUpply ........ .................... ....... ......... Solved Feed Demand ...... ..... ........... ....... ........ Seed Demand .... ............ . .......... ...,....... ..... . Seed Supply .......... .......... . .............. . ....... . Solved Seed Demand ... ........ .......................... Livestock Demand ....... ..... .. ..... ............... ....... ... Livestock Supply ......... ....... ............ ..... ... ..... ... Demand for Other Non-Durable Inputs ...... ....... ............ Demand for Hired Farm Labor ....... . ............ .. ....... .... Demand for Fertilizer ............................... ...... .. Petroleum, Fuel, and Oil Expense ....... ................... Investment Demand for Machinery and Motor Vehicles .......... Demand for Machinery and Motor Vehicle Maintenance and Repair .................................... Investment Demand for and Repairs of Farm Buildings ................................................. Investment Demand for Service Buildings, Other Structures, and Land Improvements .................... Repairs and Maintenance of Service Buildings, Other Structures, and Land Improvements .............. Expenditures for Farm Operator Dwellings ............... Repairs and Maintenance of Farm Operator Dwellings .................... ..... ................... Real Estate Price and Transfers ............................. Real Estate Price ...................................... Real Estate Transfers ............. ........ ............. Value of Real Estate Transferred ............................ The System of Equations ..................................... Reduced Form Equations ................................,..... Other Equations and Relationships ...........s............... "Near" Identity Equations .............................. ' Y (c. p r 7‘ ..ol ‘. ‘ 0"Jfi‘t ,. .. d‘ .'81 ,I‘ . O m -\’\ \ .‘~ u ‘0‘ D.‘ ‘ - Q ”,- ‘lo'_‘ 1 _vv, 4 Cu. "‘ Q sutj‘r "- T L") a:' _ ukrk‘Jk: 1)- ~41. ..g‘ 1‘. Jub‘zto‘. y. ... l ‘. \L' l 4- ‘“~CI Dev. V‘ L-._‘ said“ Page Household Purchases of Nondurable Goods . and Services ..... ........................ .... ....... .. 112 Personal Tax and Non—Tax Payments .......§ ...... ......... 113 Financial Assets ....... ............... ..... ............ 114 Number of Family Workers .............................. .. 116 Household Expenditures for Automobiles and Trucks ....... 118 Depreciation of Household Durables ................. ..... 118 Stocks of Household Durables ......................... ... 119 Non-Money Income Equations .............................. 119 Other Farm Income . .......... . .............. . ......... ... 121 Off-Farm Income . ........... . ................. . .......... 121 Demand for Loan Funds ..... . ................ . ......... ... 122 Handling of Intersector Flows ..................... 123 Calculating Stocks of Durable Assets and Depreciation ... 125 Investment Credit .............. . ..................... 129 Price and Quantity of Total Output ....... . ............ .. 131 Property Taxes ............ .... . ........................ 132 Disposable Income ......................... . ......... .... 132 Real Estate Value ....................................... 132 Farm Operator Dwellings ...... ................ . ..... ... 133 Service Buildings and Other Structures ........... ....... 134' Value of Land and Improvements .......................... 135 Accidental Damage to Farm Buildings ........ ...; ......... 136 ' Other Uses of Funds ........................ ............. 137 Summary ... .................... ...........- ................... 141 CHAPTER VI - Model Evaluation Introduction .......... ..L ...... ...1....I.. ..... . ......... ... 143 Adjustments to the Model ............ ......... ........... ..... .145 Endogenizing Crop Inventories .......... ................. 145 Revised Feed Demand Equation . ............ { .......... .... 147 Mbdel Performance ....... .............. ..... ;............... 149 ' Evaluative Data ............... ................... ....;.. 150 Crop Quantities .........-........ ...... . ..... ........... 152 Real Price of Crops ........... ............... ........... 163 Livestock Quantities .................................... 166 Livestock Price .................................. ...... . 166 Real Price of Total Output ...................,.,........ 166 Demand for Variable Inputs .............................. 173 '- Further Model Evaluation .................. ..... .............. 180 l ‘ Initial Ex Ante Simulation .. ....... . ...... .............{ 187 Crops Component ......................................... 187 Livestock Component ..................................... 189 Other Components .......................... ..... ......... 189 - ".»: awry GOOD-IOUOIIOOODIIIoopcancel-000.0000...Cocoon-IIODIIOO 190 ~.,... ‘.' I '71? 4111 ‘ .3,” L _ .IC :‘uausd 17..“ a L!- Y ’7" , Kl- hm \n .\.v ttf‘ Int vi p5211321333; VII - Financial Accounts and Other Performance Variables Financial Accounts ..............u.. ...... .................... The Balance Sheet ........... ..... ......... ......r............ .v,- The Income Statement .. ........................... ...........¢ 4 " Nominal Capital Gains ... ........ ..... ............ . ...... ..... ‘ " ‘Land and Improvements ..... ........ ....... .,.,.......,,.. Non-Land Durable Assets .................. ......t........ Crop and Livestock ........................ . ...... ....... Sources and Uses of Funds ............................ . ....... Capital Formation in the Farming Sector ......... ............. Other Data and Analytical Ratios .. ............ . ..... .... ..... Savings .. ..... . ...................................... .... Values of Exogenous and Endogenous Variables ... ...... ........ | Summary .. ......................... . .................... ...... -CHAPTER VIII - Projections under Alternative Scenarios Introduction ....... ........................ .. ......... , ..... Projection of Exogenous Farm Input Prices ..... ..... .......... Price of Motor Supplies ................... . ..... ........ Price of Buildings .................... ................ Price of Machinery and Motor Vehicles ......... .......... 'Price of Dwelling ..... ....................... ....... a Implicit Price Index for Household Equipment '3 and Furnishings ..... ........ .. ..... ................ V. Price of Fertilizer ..................................... 1. Price of Supplies .................. ..... ................ “ Scenario Development ........................ ,,.,,.,_., ..... .. Base Scenario for Exogenous Variables ..... ...... . ..... .. Alternative Scenarios for Exogenous Variables .... ..... .. ".Tu. Summary of Scenarios ........ ..... ................ ....... ' Results from Projecting Alternative Scenarios ...... .......... . Time Paths of Exogenous Variables .......... .. ........ ... - Commodity Supply, Utilization, and Prices .. ...... ....... Projections for Indicators of Farmer Welfare ....... ..... (Farm Input Projections .................... .............. Comparisons of Results from Alternative Scenarios ....... Summary ....... ............ . .......... . ............. ..... ..... tflfllPTER Ix — Summary and Conc1u31ons Introduction ....... ..... ......... ............ ................ The System to be Modeled ............. ....... .. .......... ..... The Simulation Model .................. ...... .........;....... " Empirical Results ............,.. ...... ....................... EValuatiOn of the Model ......................;............... ‘.Projections ........ ...... .................................... Pbeus of Research .......................................L.... ‘ ‘1tiona1ResearCh IOIOOOIUIIIOIIOIIOIOUIIIICCOOCIOIOI".OOOIO Page 192 194 197 200 201 202 202 202. 205— 208 208 213 213 214 215 215 215 216 216 217 2,17 218 218 218. 221 225 226 227 229 239 247 262 271 273 274 275 276 278 280 282 282 283 I . s o . .I . a o . . ... [DA ‘V. s s. .4 4 7L \ t a a u .. 4 r; u . a ..u v a a L In v 7‘ a M a... a I ”J x '8 Us «l. rt ox t . y. .1 . u L "1 0L 9‘ . r. o c . 7.. . f .1 o s .\q 1'. a, . .. .. ... ....u n.. 7k v _ u . Van Hm Ix , Ls pk -U vi .. r. r. r. w . 1 . I» '1. r1 . . V 1 .d .4 4 rs ~ Pu my. .14 r .tv .5. r: 3. y. . . 2 r1 YL 5L . u _ . .1 TA 3. ..-n T1 v. 71 L .-H Co v\ yr .rL V... VIA ad ad n\u ... {L a... t 3 .J r: .J .m C Y2 .1. vs. a: a.” 0.. s. a... am HI... flu WK“ 1. v . v.. flL .d .d P.. .va PH .1 .C r. .. . .3 PL a}. a . «A n: ~\. 1L .-L \\!. ‘v- fl .YV» ‘J . ‘lhcq ~‘j‘Jr-db- V! -...l o . . - Model Improvements ............................................ Livestock Capital Stock .........................-......... Crop Price ............................................... Supply-Utilization Aggregates ............................ Productive Durables Prices of Farm Inputs .................................... Purther Model Evaluation ....................... ...... ......... z 'APPEHDIX A -'Construction of Aggregates Commodity ' i ~Supply Utilization Data C Construction of Commodity Aggregates .......................... Livestock and Livestock Products ......................... Crops .................................................... Data ..........J............................................... Data IIO’CIDIOIOOUOIIIIIIIIIOIOOICOIOIIOIOOOIIOCDDDIOOOIIIIOIOO f“‘ .APPENDIX B — Statistical Consideration in Model Development ‘Simultaneous Nature of the Model .............................. ‘ ‘Identification ........................................... "P operties of ZSLS Estimates ............................. R in ZSLS ............................................... . Instrumental Variables ................................... Summary Statistics ............................................ Linearity ..................................................... Lag Structures ................................................ Auroregressive Nature of the Model Page 283 284 284 284 284 285 285 286 287 288 289 295 298 298 300 300 300 301 301 302 302 kl! . Le) rm, ' o . “‘521rlxn‘ ‘§ .w - ...... LIST OF TABLES Tob1e ~ - Page . 4.1 Symbols Used in Figure 4. 2 ....... . .............. . ....... ..... 29 , C 5.1 Empirical Results for Equation (5.1), Crap Supply, CRPROD .. ................ . ......... ..._ ....... . 64 1'1 ~ 5927 Empirical Results for Equation (5.2), Per Capita 9' Food-Industrial Demand for Crops, CROPS(7)/POP .....,...; ..................................... 66 5:3 Empirical Results for Equation (5.3), Feed ' Demand for Crops, CROPS(6) .. ............................... 68 . .5.4 Empirical Results for Equation (5.4), Feed Supply, ' RPFEED ..... ........ ....... . .......... . ................ .. 69 5.5 Empirical Results for Equation (5.5), Solved Feed ' Demand for Crops, CROPS(6) . ........ .. .......... . ........... 70 I Empirical Results for Equation (5.6), Seed Demand for Crops, CROPS(S) .................. ................... ... 72 $15,?) Empirical Results for Equation (5J7), Seed Supply,‘ 7‘" ”SEEDV‘IOOIIIOOOICO0.0.60.0.OI. oooooo e ....... cesauecou-eoc-l 73 5 Enpitical Results for Equation (5.8), Solved Seed ‘ Demand for Crops, CROPS(S) .... ........ . ........ ...... .... 74 EEnpirical Results for Equation (5.9), Demand for Livestock, LIV(5)/POP ...................................... 75 "A;0 Empirical Results for Equation (5.10), Livestock .1... supply, LIV(2)0 senescence-onus eeeee ounce-coleueeoooco-oo-oo 76 I; Empirical Results for Equation- (5.11), Derived Demand pvfor Other Nondurable Inputs, OTHER .................;....... 78 Empirical Results for Equation (5.12), Demand for - :Iirwd .‘Fam_.ub°r’ moRICIIIODOII......CCOCO-.....‘ICU‘CI'O 79 flgpdrical Results for Equation (5.13), Demand for , 1 v Bernlizer, PERT ......IOICU...’..I.'CI‘..‘.C.....‘C....'I.. 81 \l‘ U‘ LI. .1i '1 y '; '1 L1..u,t. r“ ‘fs ~- ....I .1 1L.: V, r. v ..u L', :Wr. '-.. u~rlrlta. ,' v‘ ..r r~f Cx‘fi fl\ 52mm. :5116 Empirical Results for Equation (5.16), Machinery and Motor Vehicle Maintenance and Repair , 5.17 Empirical Results for Equation (5.17), Investment 'r.,¥n 1 Demand for Service Buildings, Other " " Structures, and Improvements to w. 5.18 Empirical Results for Equation (5.18), Repairs and Maintenance of Service Buildings, Other Structures, 5.19 Empirical Results for Equation (5.19), Expenditures 5.20 Empirical Results for Equation (5.20), Repairs and 5. 22 Empirical Results for Equation (5.22), Acres of Real Estate Sold, Voluntary and Estate '5.23 matrix Showing Placement of Endogenous Coefficiegts for Structural J . . ~3¥ :f‘5.24 Reduced form Matrix Representation of the ‘ “ Simultaneous Equations . ....... . ............ ................ E6 Empirical Results for Equation (5.26), Cash “ . Receipts from Crops Near Identity, CSHCRO + CHCRV ... ...... . ............ ~ ......... .............. .i‘ t .- :3! 3127 Per Farm Stocks of Financial Assets .. ...... .................. crops, and Rental Value of Farm Operator Dwellings ......OOIIIIIOIICOOOIOII IIIIII ‘OIOIIIIIIIIIOODIIOII ;Imumenccra1tnata...lIUOIIIOIIOIIIIIIIOOOIIOICIIIUICIOCI 1'Eistorical Data for Other Uses of FUnds and a Al‘tauat1VEInC0memeasures solo-soocone-aooooo-ao-ooooooooo DeriVed Demand, RREPM ..... ... ........... . ........ .... ..... . Land,EXPBLD .................... .... ..... ...... ......... .... and Land Imrpovements, REPBLD ..... ............... . ...... ... for Farm Operator Dwellings, CE(3) ................. ........ Maintenance of Farm Operator Dwellings, REPDWL .... ........ . .5.21 Real Estate Price Equationsa ............... ' ..... . ....... ...... Sales of 10 Acres or More, ASLD ........ .......... .......... Equations ................................................. 90 92 93 94 97 99 102 104 110 115 120 130 138 141 ..‘V -‘n‘. .J;.€ ‘ "o-V‘V“I‘1 " ' g... 2.; _._:... '- 7 PI :.t , ..-; : r.2 Eur.“ 7' - 7" L\ -"-“ !'~.- '9 ..vdl‘u-l. fl 6 Y ..- , - . 4 . —\u.¢u‘.. 3'9 r311 aii‘. 6-1U5‘-‘al;u:,. 6.11 Evian-tum 6'12 Enlvfii‘. 6'13 “d um; 6‘11. “31411.1(. 61‘ 21.311431. 6'1EEV31133: . r . i 1 f E E L Table 1 Page 6.1 Empirical Results for Equation (6.1), Demand for Crop Inventory, CROP(8) . ...... . ............... . ........ 146 6.2 Empirical Results for Equation (6. 2), Feed Demand, CROPS(6) ........ . ..... . .................. . ..... 149 6.3 Evaluative Data for Crop Production, CROPS(Z) ................ 153 6.4 Evaluative Data for Quantity of Feed Use, CROPS(6) ....... .... 155 6.5 Evaluative Data for Quantity of Seed Use, CROPS(S) ........... 157 6.6 Evaluative Data for Quantity of Food—Industrial Use of Crops, CROPS(7) ..................................... ....... 159 6.7 Evaluative Data for Ending Inventories of Crops, CROPS(8) .... 161 6.8 Evaluative Data for Real Price of Crops, RPCROP ............ .. 164 6.9 Evaluative Data for Livestock Production, LIV(2) ...... . ...... 167 6.10 Evaluative Data for Livestock Consumption, LIV(5) ....... ..... 169 6.11 EvahatuveData for Real Price of Livestock, RPLIV .... ........ 171 6.12 Evaluative Data for Real Price of Seed, RPSEED ... ............ 174 6.13 Evaluative Data for Real Price of Feed, RPFEED ... ....... ..... 176 6.14 Evaluative Data for Real Price of Total Output, RPTOUT ....... 178 6.15 Evaluative Data for Quantity of Hired Farm Labor, LABOR ...... 181 6.16 Evaluative Data for Quantity of Fertilizer, FERT ... ........ .. 183 6.17 Evaluative Data for Quantity of Other Non-Durable Inputs, OTHER . ...... .......... ....... . ...... . ........... ... 185 6.18 Forecasted and Actual Results from Ex Ante Simulation with No Adjustments ........................................ 191 6.19 Forecasted and Actual Results from Ex Ante Sumulation with Some Adjustment Made to the Model ............. ....... . 191 7.1 Balance Sheet of the 0.3 Farm Production Sector, January 1 .......................... ....... ................. 196 7.2 Income Statement for the U.S. Farm Production Sector, Year Ending December 31, 1974 .............................. 199 ( 1:) . H v,‘3'l tn.» :4 ..,' ' ?T(4t_|_ '1 S... 019C115, SCICLLL‘: “Jr; 1".“ 41‘. HISEOrlkC‘ (a. , xi Page Implicit Nominal Capital Gains on Physical Assets for the U. S. Farming Sector, Year Ending J December 31 ........... . ...... . . .............. . . ..... 203 . '714 .Cash Sources and Uses of Funds for the U.S. Farm 9 Production Sector, Year Ending December 31 ............ ..... 204~ "0.5 ~Gross and Net Capital Formation in the U.S. Farming ‘ ' Sector for the Year Ending December 31 .................... . 206 ‘ 71'7°5 'Other Data and Analytical Ratios, U.S. Farming Sector, Year Ending December 31 .................................... 209 1 18.1 Values of Selected Exogenous Variables, Selected , Years, Scenario One ............ . ................ . .......... 228 M '8.2' Projections of Selected Farm Prices, Scenario One ....... ..... 232 A - 8.3 Projections of Selected Farm Prices, Scenario Three ..... ..... 233 ' 8.4 Projections of Supply-Utilization of Livestock, ‘ ‘ . Scenario One . .................... . ........................ . 235 .' 8.5 Projections of Supply-Utilization of Crops, ' ' Scenario One .... ................. . ..................... . .. 237 240 ,Eistorical Data for Capital Gains and Net Farm Income ........ ‘86 7 '93 3.7 Projections of New Farm Income, Alternative Scenarios ........ 241 :18.8 Projections of Capital Gains, Alternative Scenarios .......... 243 i539 Historical Data and Projections of Farm Operator Household Expenditures and Other Uses of Funds, Scenario One ..... .............. ......... ....... ... . L 246 00...... 251 ...... I. ...-IIIOIOOOOO.I 254 vroj_ectison of Selected Balance Sheet Data, Scenario One ..... 256' Projections of Capital Formation, Real Estate Transfer, 1 258 mi Savings,- Scenario One .................................. 108 dunno-coo....-oncocoonnoun-cano-oooI-OI-o-uooos'dse 259 ,'.‘_1 " PIN-ta: ‘- 1 :~ 5.1! Sttflori '11 CCT'fofih .‘ K DO. I.) YFV" O l' 0‘: ‘ {DE A o v .‘- L \.-K . a ‘1‘. Cu 1 ‘LLA .‘ . ~¢ Identi xii 'rPage V3¥3LPr§jections of RPCROP and RPLIV, Alternative Scenarios ........ 266 *Pficjections of Selected Variables, Scenario Six ............... 26? Idf9reomparison of Net Farm Income Projections under Alternative .}y¢ . Scenarios ................................................... 268 I Q0 Comparison of Nominal Capital Gains Projections under Alternative Scenarios ................................. 269 (‘I -2 Camparison of Net Flow of Loan Funds under Alternative _ 3 Scenarios ................................................... 270 .,, l Commodities. Price Weights, and Data Sources Used in @#Q _ Construction of Livestock Aggregates ........................ 290 \ :EZJ commodities, Price Weights, and Data Sources Used in 3.5' :Construction of Crop Aggregates ............................. 292 3, ,1 Aggregate Crop Supply-Utilization Data ........................ 296 ‘1 mgaregate Livestock Supply-Utilization Data ................... 297 :UIdenmifiability Teét Statistics .......;...............;....... 299 »LIST OF FIGURES . Page 5.1 The Decision Process . .' ....................................... 8 I 4.1 The Farming Sector ........ . . . . . . . ................. . . . . ....... 24 . r I 'r'General Block Diagram of the Farming Sector ........ . . . ....... 28 LIV Crop Production .......u... ....... ...... ......... . ..... ....... 154 rv6.2li‘eed Use of Crops . ..... .. ...... ...... . ......... . 156 fh'lfifi‘g'Seed'Use........... ........... .. .................... ..... .158 16.4 Food-Industrial Use of Crops .............................. . . . 160 ‘ ’2 Ending‘lnventory of Crops .... . .. ......................... ..r 162 . ifReal'Price of Crops ...... 165 _LivestockPr-oduction‘............... ......... ......... 168 @1’.:Pr1cve°f Feed .....I'IDGQCDIIIIIIII.III~ IIIIIII .....IIOOII 177 . CHAPTER I Introduction .Justification ‘Decisions affected by the financial structure of the U.S. farming sector are continually being made. Implicit in the if; :decision-making process are expectations of-the future financial . . _strficture of the farming sector. Informed policy decisions require knowledge of relationships within the farming sector and an under— $1: standing of the effects of variables exogenous to the sector. The impetus for this study is provided by a perceived usefulness :in public and private decisions of economic projections for the U.S. 44dfarming.sector. Specifically, it is believed that a computerized ; simulation model that emphasizes the financial structure of the Jilarming sector, as represented by a set of sector financial state- ‘13.}? ,1. ‘ 7 acute, and that is capable of projecting complete financial state- 333? of t13n5hi9s bi is “559533r§ controllaElé her of pu51e long‘run 0'6! DeciSic Credit‘orier HElfare of I pglicies); r durable imp; ttOt with AM...“ that a model could address are: the rate of growth in gross national product, population growth, technological change, prices of farm inputs, governmental farm price and income policies, export demand for farm products, and grain reserve policies. These factors are manifested through behavioral and economic relationships in the form of cash and noncash flows which, over time, affect the assets and liabilities of the farming sector. The relationships of interest can be specified largely through the use of economic theory and can be empirically estimated using econo- metric techniques. Many of the factors affecting financial flows in the farming sector are determined, in part, by public policy. Knowledge of the rela- tionships between system variables, both controllable and uncontrollable, is necessary for informed public and private decision making. In addi— tion, projections of system performance under alternative policies (for controllable exogenous variables) and under alternative scenarios (for uncontrollable exogenous variables) would provide input into a large num— ber of public and private decisions dependent upon or affected by the long—run outlook for the financial structure of the farming sector. Decision makers who would be potential users of information proveded by the model are: the Farm Credit Adminstration and other credit—oriented clientele: public decision makers concerned with the welfare of farm proprietors (i.e., formulation of farm price and income policies); suppliers of farm inputs, including both durable and non- durable inputs; and those analysts wishing to compare the farming sector with nonfarm sectors. Specific questions that an aggregate economic projections model could address can be posed in terms of the kinds of scenarios the model is designed to handle. Scenarios in this context are alternative pro- jections of variable exogenous to the model. The usefulness of a model comes from its ability to predict the effects of alternative scenarios on endogenous variable. Endogenous variable of interest to policy . makers are often called performance variables. In the model proposed here, these include: net farm income, the level of farm production, the price of farm products, capital gains, equity accumulation, debt flows, capital formation, savings, leverage, intersector flows, and consumption levels in the farming sector. Exogenous variables are , often divided into the categories of variables controllable and un- controllable by policy makers. Thus, information concerning effects i of alternative policies (levels of controllable variables) on perform- } ance variablesis particularly useful to public decision makers. In addition, levels of performance variables under projections based upon a "most likely" scenario have implications for such private decision makers as farmers, agricultural lenders, and input supply firms. Furthermore, there is a continuing need to formulate, conceptualize, } and study the structural interrelationships of the farming sector in order to better understand what is happening and why. Purpose With justification provided by the above factors, the purpose of this research is to design an aggregate U.S. farming sector economic projections model. The model will, in general, be of an aggregative l E ' nature, but will contain sufficient detail to handle a broad set of A policy questions related to financing the aggregate farming sector. The focus of the model will be on making long—run projections. Research Objectives The proposed research is oriented toward policy decisions and disciplinary knowledge. With respect to policy, the intent is to con- tribute a portion of the information necessary for a class of policy decisions. Specifically, the class of policy decisions affecting or affected by the financial structure of the aggregate U.S. farming sector is of interest. The research is not problem solving per se in that all of the information necessary to make policy decisions will not be provided. That is, policy prescriptions cannot be reached without information in addition to that provided by this research. The specific objectives of this project are as follows: 1) To develop a theoretical model of the aggregate U.S. farming sector in order to provide a conceptual framework for the estimation of an empirical model 2) To identify structural relationships among variables within the U.S. farming sector and the effects of variables exog- enous to the sector through empirical investigation 3) To construct an operational aggregate economic projections model of the U.S. farming sector capable of making long- run projections of financial variables under alternative futures in order to provide input into public and private decision making l....‘.- >_—-.'V"‘nr~w~ ~‘. '4‘— .. we. qv—v r...-.. i K Dissertation Organization Chapter II of this dissertation gives a short description of the methodological and philosophical approach of the research. Chapter III is a broad review of previous studies relating to the research reported here. This includes economic modeling efforts at the farming sector level which are not necessarily focused on finance. Chapter IV identifies the system to be modeled and develops ’the theoretical basis for equations to be estimated for the model. Chapter V presents the empirical results from estimation of equations. This includes ancillary equations required for a complete model as well as structural equations. Statistical results for each individual equation are presented. Economic properties (i.e., price and income elasticities) of the individual equations are shown and discussed. Elasticities of the set of simultaneous equations are -pursued further in Baker (1978). Modifications to structural equations are discussed in Chapter VI. In addition, the model's ability to track the historical data is evaluated for a set of endogenous variables. The results of simulating over the historical period are presented numerically and graphically. Chapter VII provides background for and develops the financial accounts and other financial data projected by the simulation model. In Chapter VIII, results from the simulation model are presented. Additionally, the methods and assumptions used to project exogenous variables are explained.‘ ‘Chapter IX presents a summary and conclusions. The appendices include both explanations of the construction of A glossary of variables used in equations, financial statements, and elsewhere in the simulation model is published in Baker (1978). '“ The glossary is arranged in alphabetical order by variable name (the symbols used in equations and/or computer variable names). The 3lossary should be used by the reader to obtain additional infor- 'fiafi 3 intion on the following: alternative variable names, descriptions of ~£T7Txthe‘variables, units of measurement, variable type (i. e., endogenous ‘br exogenOus), historical data source, and sources of the variables .-1§ the simulation model. l __-_--. CHAPTER II Methodology Methodological Approach The methodological approach to this study follows a modification of the system's problem-solving methodology outlined by Manetsch and Park (1974, Chapter II). The methodology described by Manetsch and Park can be viewed as a decision—making process encompassing the steps from problem recognition through implementation and system operation with resulting feedback. The process is conceptually similar to the decision—making process discussed in farm management. Figure 2.1 illustrates the process as typically presented by farm management researchers (Bradford and Johnson, 1953; Hopkin, Baker, and Barry, 1973; and Johnson, 1954 and 1961). The approach here is not identical to the problem-solving pro— cess for it is only a subset of the process. In addition, the re— search reported here is intended to be applicable to a large number of problems pertainintho the topic researched. However, the purpose of emphasizing, here at the outset, the "problem-solving process" or "the decision process" is to place the research explicitly within a Palicy framework—-even though theinformationassembled in this 7 PROBLEH DEFINITION ' I V [7 OBSERVATION p] 4k V NORMATIVE I ANALYSIS l POSITIVE om BANK ‘_—""" *—* DATA BANK A r \ ‘I" / DECISION J I \f r EXECUTION I NESPONSIIIILIIY BEARING Figure 2.1 The Decision Process .rwb-‘=. -- u—vv w, m," research may fall short of that required for policy prescription. Criteria for Objectivity The author's philosophical view of the criteria for objectivity has important implications for the development and testing of a model such as that constructed in this research project. The four criteria for truthl/ of a concept are: internal con- sistency (coherence), external consistency (correspondence), worka— bility, and clarity (inter—personal transmissability). Internal consistency requires an analytical system that is coher- ant and conforms to logic. Economic theory provides the basis for the ,system of logic applied to the model developed in this research. External consistancy requires correspondence of the relation- ships developed to those of the real world. Tests of correspondence include statistical tests and estimates based on empirical data. However, the author does not feel confined to statistical tests. Through experience, one develops concepts of what exists. These concepts are useful to provide informal tests when there are no data to Support formal testing. Workability requires that the concepts being developed be use- ful in the problem—solving context. This provides a portion of the impetus for putting the equations developed into a model to use in policy analysis. l/ “ One may substitute "objectivity" or "validity" for the word "truth" in this sentence. 10 Iszocedures I The research procedures for this project have been divided into 'three parts: premodeling analysis, system modeling, and postmodeling 'ianalysis. The following is an outline of the research process relevant to this study: A. Premodeling analysis 1. Needs analysis. Identify and examine the consistency of the -needs of public and private decision makers and other system participants. 2. System identification. In a general way, identify and define the system, including classification of variables. a) System inputs (1) Exogenous [(2) Endogenous (a) Controllable by policy makers (b) Not controllable by policy makers b) System Outputs (l) Desired ‘w(l) undesired 11' inorderto satisfy the determined needs b) Develop performance criteria and determine performance variables 4. Generation of system alternatives -a) Alternative management strategies or policies b) Alternative models that might be constructed to address ~ ,:I-,- . - policy questions ~ 5. Select a subset of feasible system alternatives 3. System modeling L 1. Select a final subset of system alternatives to model 2.’ Develop a conceptual model in the form of equations and/or explicit block diagrams, ‘ a) Determine the hypothesized relationships between the Ovariables . b) Check for internal consistency in the model formulation c) Check for external consistency where possible . 3. Computer implementation ' . a)‘ Estimate parameters I 5) Develop a computer model using analytical and numerical .H techniques . e) Test the viability of concepts used in developing the I..3mode1 12 sensitivity tests a) .Carry out "sensitivity tests” on the model coefficients for which accurate estimates are not available b) Establish priorities for further information-gathering and model refinements ‘5‘ 0‘ Stability analysis .‘ ’- I a) Identify the stability boundaries of the model .3} 1, ,. e i. 3. ,x -b) Test for stability based on stability theory, use of 93K .J repeated simulation runs, or both “is” “I .-'.-4- 3'5 "s 6‘ ‘ I - *irs $‘.” ‘ 'Tostmodeling analysis .V, ‘5'"? I .0 ” '1. 1. Use the model to describe past system behavior .. J'L‘ \ I in). > ') I r ‘qur-5. ‘0‘ o .1. Hn~ 2. Examine the effects on system variables of alternative poli- - cies through simulation runs V . 3. Project future system performance under alternative scenarios We >1 thThe procedures of this research have been affected by the philo- ~ . lcéhical orientation of the author--hence the need for this chapter. CHAPTER III Literature Review :Introduction The body of literature relevant to this study is extensive ‘ “‘lf’ . and diverse. This chapter will review previous research in areas re- “ lated to this study on a general basis. In later sections, more :specific references will be made to previous studies as they pertain ~ to the particular topic being considered. ‘ Aggregate Financial Analysis . Financial analysis of the aggregate farming sector differs from Hi_ghat one might refer to more generally as economic analysis in that awn" xm‘ ......- F' ,“fivm‘l 'w‘ i. l4 (Tostlebe, Tables 35 and 36, pp. 136—139). The approach was to estimate the uses of capital (land and buildings, machinery and motor Avehicles, inventories of crops and livestock, and cash working bal— ances) and estimate the external sources of capital (loans, credit, and financial reserves) in order to derive the implied internal sources. Also constructed were two sets of ratios of the following variables: Savings to income, savings to capital formation, and capital formation to income. One set of ratios is on a net basis, the other on a gross basis (Tostlebe, Table 38, p. 146). The formulation of the account by Tostlebe is consistent with the type of account one might initially deem appropriate as a sector disaggregation of a national account. In a closed national economy, the financing task can be reduced to diverting just enough funds from current income to equal total gross capital expenditures on new tangible assets. Tostlebe's disaggregation includes the addition of financial assets and liabilities to the account. However, for a particular economic unit, the task is larger than simply including financial assets and liabilities, since other intersector flows must be financed. The flows omitted by Tostlebe and others doing the early flow of funds research were those related to proprietors entering and leaving the sector. D. Gale Johnson (1963) updated Tostlebe's accounts through 1958, based on data published annually in the Farm Income Situation and The Balance Sheet of the Farming Sector. The research was similar to Tost- lebe's in that it was basically an examination of historical trends. “-..—.... - ..- 0.. 15 Brake (1966) used a flow of funds model as the basis for pro- jecting increases in farm debt. Brake's approach combined current flows (operating income and expenses, consumption, taxes, etc.) into a savings variable and included a flow for real estate. In other research, Brake (1970) conducted a more extensive exam-- ination of fund flows in the Canadian agricultural sector. Projections of capital and credit needs to 1980 were made. The method used was similar to Brake's previous work with the addition of a detailed break- down of capital and current flows. Melichar (1973) estimated equations to project capital flows for real estate transfers, machinery and motor vehicles, buildings and land improvements, livestock and crop inventories, and financial assets. Internal financing was determined by projecting net cash flow from operating, then multiplying by a savings rate. Credit needs were thendeterminedresidually. Melichar's flow of funds account extended Tostlebe’s and modified the Brake account by including an important intersector flow, purchases of real estate from discontinuing pro- prietors. The Brake and Melichar models project credit needs, but do not provide a great deal in the way of structural parameters. Melichar pointed to the need for building structural models lIIWthh independent financial variables are simultaneously determined and to the need for examination of the factors determining internal financing. While the accounts used by Tostlebe and Melichar focused entire— ly on capital formation and the scurces of its financing, the later Brake account moved in the direction of including all cash flows. 16 During this period, an effort was underway in the USDA to construct a flow of funds social account for the farming sector to serve as a basis for examining financial aspects of policy questions (Penson, Lins, and Irwin, 1971). This account was designed to link Balance Sheets. The Penson, Lins, and Irwin account may have grown out of the use of the Sources and Uses of Funds (SAUF) statement at the firm level. Van Horne (1971) describes sources and uses of funds for a cash basis statement as follows: Sources: 1. A net decrease in any asset other than cash 2. A net increase in any liability 3. Proceeds from the sale of preferred or common stock 4. Funds provided by operationl/ Uses: 1. A net increase in any asset other than cash or fixed assets 2. A gross increase in fixed assets 3. A net decrease in any liability 4. A retirement or purchase of stock U! . Cash dividends The_majority of firm level accounting is done on a historical cost basis. In a cost accounting system, a cash basis SAUF statement will account for all changes on the Balance Sheet between periods. 1] Funds provided by operation are defined by Van Horne as net income after taxes plus noncash expenses (e.g., depreciation). 17 Much of the social accounting for the farming sector values assets at current price levels (e.g., values of real estate and inventories of crops and livestock published in the Balance Sheet of the Farming Sector). Thus, if one requires that a SAUF statement account for all changes on the Balance Sheet, it will include a mixure of cash and noncash flows. This view has been taken by researchers dealing with the nonfarm as well as farm sectors. Hendershott has viewed the SAUF statement as the link between two Balance Sheets when doing financial modeling for nonfarm sectors. He hypothesized the following financial statements: Hypothetical Balance Sheet Hypothetical SAUF Statement FA1 FL1 AFA1 AFL1 FA2 FL2 AFA2 AFL2 FA FL AFA -CG AFL r n r RA NW INV1 - SAV RA2 INV2 RA INV m m FAi = the ith financial asset. ‘ FL1 = the ith financial liability. RAi = the ith real assets. NW = net worth. A =9 change CGn 8 capital gain on the nth financial asset. SAV = savings. INV1=net investment in the ith real asset. I y 18 The sectors which Hendershott consideres endogenous are households, nonfinancial businesses, state and local governments, commercial banks, other savings institutions, sponsored federal agencies, and other finance. Three exogenous sectors are the federal government, the monetary authority, and the rest of the world. Hendershott has concentrated research efforts on financial as- sets and liabilities. In general, he treats investment in real assets and savings exogenously. His approach is "general equilibrium" in nature (with respect to financial markets) in that all interest rates are determined by the direct interaction of demand and supply in the financial markets.l/ , The modeling by Hendershott differs substantially from the re- search proposed and reviewed here due to its multi-sector general equilibrium approach for financial markets. The approach normally taken for farming sector models is that the farming sector does not have a great influence on the money markets and that the key endogenous variables should be investement in real assets and internal financing. The Penson, Lins, and Irwin account for the farming sector was criticized by Brake and Barry (1971). Brake and Barry objected to the conception of the SAUF statement as completely bridging Balance Sheets between two periods. The SAUF statement resulting from this view included a mixture of cash and noncash flows which Brake and l/ - The majority of present day financial models explain a short-term rate of interest by analyzing a bank reserve market and then determining long-term rates through a term structure relationship. .v-‘xwtr-"" .. . < “‘1'“ w—p— _ l.... A —-r"" 1'“ I f, . I 19 Barry felt was conceptually incorrect. They, in turn, proposed a SAUF statement onia cash basis that included gross flows where possible. The cash basis account is appealing because it is limited to and includes all items that require financing and the sources of their financing. Extensive modeling of fund flows in the farming sector has been completed by Lins (1972 and 1973) and Penson (1973). Lins developed a model concentrating on sources of external funds, including exten— sive disaggregation by lender groups. Penson viewed the flow of funds in the farming sector as result- ing from adjustments in the portfolio of farmers from actual to desired asset levels. The theory of portfolio balance suggests that the desired balance between physical and financial assets in the port- folio depends upon relative pecuniary and nonpecuniary services and, .hence, utility provided by the asset. This reasoning led Penson to a set of structural equations, simultaneouslydeterminingthe year- end stocks of physical and financial assets. Coefficients of stock variables indicate complementarity (positive sign) or substitutability (negative sign) in determination of year-end stocks. The empirical model of Penson is the basis of an aggregative.income and wealth (AIW) simulation model (Penson, 1973). Penson used a modification of the SAUF statement for the farming .sector proposed by Penson, Lins, and Irwin. The proposed SAUF state- ment was designed to be consistent with USDA income and balance sheet series, thus making it useful in terms of facilitating financial 20 analysis of the farming sector. Cash and noncash flows are included in the simluaiton model developed here. However, only cash flows are inculeded in the SAUF statement. The AIW simulation model in— cludes gross flows where data are available. Gross income and gross cash expenses are included rather than net income, capital con— sumption allowances, and net change in inventories of crops and livestock. Data needs and limitations are pointed out by Penson, Lins, and Irwin and are major factors in the development of their SAUF statement. Emphasis of This Research The research reported here differs from earlier financial studies in one broad area. The economic relationships underlying the financial accounts are modeled to the extent possible, based on stan- dard theoretical relationships suggested by static economic theory. The model differs from the AIW simulator in its theoretical basis and in the variables endogenous to the model. The model here includes as endogenous variables the supply and demand for aggregate farm out— put, including price determination. In addition, the demand for durable and nondurable farm inputs is included in the model. The approach was to provide a complete set of equations in sufficient detail to project an Income Statement, Balance Sheet, and cash basis Sources and Uses of Funds Statement for the sector. The information required for these accounts is sufficient to construct any of the ac- counts discussed earlier, as well as the Capital Finance and Capital Flowsaccountsproposed by Simunek (1976). 21 Economic Models of the Aggregate Farming Sector The emphasis of this study on supply and demand for farm output and on demand for farm inputs (the factors relating to internal financing) leads to examination of complete models and component studies of previous researchers. The concept of aggregate supply is especially important for this study. Early developments of the concept of aggretate supply in the agricultural sector include articles by D. Gale Johnson (1956), T. W. Shultz (1956), and W. W. Cochrane (1955). Among the more complete studies of the farming sector is that of Heady and Tweeten (1963). Their study included extensive estimation vWCC'“ ‘wé of structural equations for durable and nondurable inputs and aggre— _ v gate suply. Earlier studies of factor markets include the demand for farm machinery by Griliches (1959, 1960, 1962) and Cromarty (1959), an analysis of the farm labor market by Schuh (1962) and Tyrchniewicz and Schuh (1969), and studies of farmland prices by Herdt and Cochrane (1966) and Tweeten and Martin (1966). Policy-oriented aggretate models have been developed by Tyner and Tweeten (1968), Ray and Heady (1972), and Nelson (1975). These . _‘,-._.._.-—1..—"~vo.—' -..“. / models have not focused on the financial implications of economic ac- tivities and policies. Long-run projection models have been compiled into a national interregional projections (NIRAP) system by the Economic Projections and Analytical Systems program area in the National Economic Analysis W.— W‘v—~—~ v--- -... _ ..l , ”‘Y fifi—Vw 22 Division of Economic Research Service, USDA. The capacity of the system includes the ability to make projections of aggregate output, prices, and net farm income. However, a complete set of financial flows is not provided. Concluding Remarks It is hoped that the reader perceives the gap in previous research at which this project is aimed. It is intended that this research will differ from previous financial modeling in terms of theoretical basis, structure modeled, and variables that are endogenous. The inclusion of variables such as farm prices and quantities of output as endogenous' variables is intended to make the model useful for longer-run projections. The research differs from other farming sector studies which have endogenized input and output prices and quantities in that it goes one step further to trace the impacts of these economic activities on seetor financial statements. I." \CWM . 'w ervm—nJ- . ... CHAPTER IV System Definition and Theoretical Model Definition of the Farming Sector Defining the farming sector is not an easy task. Any workable definition must be conditioned on the availability of data and the use or purpose and must fit into one's conceptual framework. The definition used here is best described by Figure 4.1. The individuals included are farm operator families and nonoperator landlords. The decision-making units included are farming businesses, farm operator households, and the farm investment portion of nonoperator landlord households. TThe definition is based in part upon data. The individuals in- cluded are those whose assets appear in the Balance Sheet of the Farm— ing Sector and whose income is reported in the farm income accounts . of the USDA (see Farm Income Statistics). Conceptual difficulty occurs to the extent that the three de— 'cision-making units are not distinct. This is troublesome with respect to the interrelated nature of farm operator business-household decisions. It is felt that decisions of nonoperator landlords are made in a somewhat different manner than decisions made by operators. There- fore, the sector could not be considered farm operator households 23 A 24 ' Farm I Operator Households ' In I m S. :3 .p ’l" l a c a) CI. l x Lu 2 ,_ .8 .3 E "‘ g; 5? ° 2 5 a u- Q g * Farming + ' E BusInesses .c + 1 U) 06 .‘ U r. , ID 2 Cash Capital Expenditures Nonoperator Landlords Figure 4.1 rrThe_Farming Sector Nonfarm Income Nonfarm Investment 25 ,:flé§b9nt Subtracting out nonoperator landlord activities. With the :s:.*jf1fisredsing size of the firm, level of mechanization, and income, the l :lfiaecision-making units of farm operator households and farm businesses sf. I'san'be conceptually separated, even though decisions may be interde- & -' i pendent. Thus, the definition of the farming sector was based largely ,f . on the ability to identify decision-making units and the activities of ».o'.‘>-. Eacho‘ kFarm Operator Households ~.§}1§" :‘ . Activities and decisions of farm operator households will include 'the following: 1) Consumption of nondurable goods and services requiring cash transactions {IIII'-‘,2),Consumpt10n of services from durables (no cash transaction) :3)rConsumption of nondurable goods and services that are noncash in nature 4)'Expenditures' for consumer durables 9) Making net cash investment in the farming business I 26 Nonoperator Landlords Nonoperator landlords are not treated as extensively as opera— tors. Nonoperator landlords are included in the farming sector only to the extent of their farm business activities. Income of nonoper— ator landlords other than from farming is not considered to be income of the farming sector. A portion of the net cash flow from farming goes to and a por— tion of net cash capital expenditures comes from nonoperator land— lords. The difference, net cash investment, is determined by the investment alternatives available and by consumption patterns. Farming Businesses Decisions made by the farm business decision—making unit are based upon costs and returns in farming. Produciton and distribution theory will be used extensively in deriving relationships. Decisions will include those related to current flows, such as crop and livestock sales, government payments, and operating expenSes. Capital flows for purchases of real estate, machinery, motor vehicles, new buildings, and improvements will also be included. Debt acquired and retired will be considered activities of farm business. Sector Interface (Intersector Flows) Over time, individuals move into and out of the farm sector as (defined herein. This entry and exit is partially voluntary as in 3 instances of labor migration off farms, and partially involuntary as it: the case of death. There are fund flows associated with entry and extit. These include the following cash flows: 27 1) Cash inheritance and gifts (in and out) 2) Equity introduced by new proprietors 3) Equity removed by discontinuing proprietors It is reasonably clear that the net flow of these items is a deficit for thefarmingsector. That is, more cash leaves the sector via the above flows than enters. Derivation of equations explaining the flows for inheritances, gifts, and equity accompanying entry and exit is conceptually diffi- cult. Some portion of the reduction of human resources in agriculture can be explained via relative returns to labor. It is not clear to the author how one theoretically explains the equity captial flows associated with the human resource flows. Data series for these variables are in general nonexistent. It was mentioned earlier that the anticipated net flow is out of the sec- tor. Thus, ignoring flows associated with sector interface will bias a residual factor substantially. The concept of the farming sector described here is a combina- tion of establishment and product concepts. A Systems View of the Farming Sector Figure 4.2 is a general block diagram for the farming sector. The purpose of the diagram is to give an overview of the farm sector as a system. Major physical and behavioral processes are indicated by blocks. Inputs (stimuli) are indicated by arrows into blocks, and outputs (responses) are indicated by arrows out of blocks. Much of the detail has been excluded, but major linkages have been retained. 25 v corpora; “cmrumm>ch scumnwnccox . \ .I m l H.-._. . . a _ _ i-.....----I---.%o . l _ >. M umxpmx “Dunno — ...—”-..“..- ———. _..--_ .... ._—_. L. Sthtor P D —~> .--»a—qfl corpucsu . cowwurnmcou - _ cofiuocam If) 3 O J- — ) ’ ).). . _ _ J rrcrw : r COwuunmoue _ m / repatmnc . . l . z . _ l /, IIIIIIII -IIIIIII IIIIII- “ .II N! \l l ,- .- ./\ l _ m. f, . IIIII " lII. - ( - .— ‘..__...__ ...—.--- _.-_._.-_-—..— .... .. ...... / (Jouu:rul BIIHtk llizurrunI of thte PaIInInI umxnmx unenH Li I.) I I. (k; I I. (J l‘ L) I) r O .2 4“ - /‘~ .------ Figure 4.2. 29 Table 4.1. Symbols Used in Figure 4.2 Symbol Definition/Explanation BS Balance Sheet for the sector C Vector of consumption expenditures, current and capital CN Credit needs CP Vector of expenditures for productive captial items D Debt flows Bi In general, a vector of exogenous factors E1 Includes weather, technology, and a subset of govern- ment programs E2 Includes government programs, population, inflation rate, GNP E3 Includes prices of captial goods, r, current input prices E4 Includes prices of consumer goods and services, r, nonfarm income E5 Includes rates of return on alternative investments IS Income Statement for the sector NOI Nonoperator landlord net investment Px Vector of current input prices Py Vector of output prices R Cash receipts from farm marketings r Interest rate X Vector of current input levels Y Vector of output levels 30 In terms of the procedures outlined earlier, the block diagram is one of the outputs of premodeling analysis. There is an indicae tion of the breakdown of variables into endogenous and exogenous categories. Specification in greater detail will be given later in this section. The choice of processes indicated in Figure 4.2 was based upon economic theory and considerations discussed in earlier sections. Each block will yield a set of one or more equations relating inputs and outputs. Sequencing of blocks does not necessarily imply recur- siveness. Simultaneity is broken by delay blocks. In the situations used here, delays may be interpreted as distributed lags. A block diagram places emphasis on structure. This causes some problems here because the structure for an individual firm differs from the structure in a market. An example would be the shift from a fixed output price for a firm to a demand function for output when firm cost curves are aggregated to specify industry price and output relationShips. The actual equations used to describe the. system and to be empirically estimated will focus on structure where possible. Reduced form equations are often estimated for forecasting models but will not be estimated directly for this model because of the emphasis on long-run projections. Structural equation estimation will involvesimultaneous equation estimators. The farm production process is described in traditional economic terms of input level determination, production, and price determina- tion. As we will see later, this is more complicated than expressed 31 in the block diagram because of the use of crop output in the produc- tion of livestock. The inclusion of farm operator households brings about consid- erations of consumption activities. These include expenditures for household capital items as well as currently consumed goods and ser— vices. The investment block determines cash flows for purchases of pro- ductive capital items. The level of the capital stock feeds back into the production block. Nonoperator landlords are treated as making investment into the farming sector based upon returns in farming (information from the Income Statement) and a set of exogenous factors. The lack of and poor quality of data on the current flows to and from nonopera- tor landlords will prevent estimation of this component. The author is not sure of the direction of bias that this may create. The financial market is treated as a residual supplier of funds based upon credit needs. The rate of interest is taken as exogenous to the sector. The block entitled "accounting subsystem" is included to avoid numerous and cumbersome summations. It is a set of identities pro- ducing the Income Statement, Balance Sheet, Sources and Uses of Funds Statement, and miscellaneous summaries for the sector. The following paragraphs will derive explicit equations to de- scribe the system. The derivation will first be the general theory without specific names given to the variables. Later, specific vari- ables will be identified as the theory is applied to the farm sector. _____——_.. _ .-....ld -_ _ 32 Firm Level Structure For simplicity, we will consider a single enterprise firm pro- ducing output Q with variable nondurable inputs X1 and X2 and fixed level of service inputs V from a durable good according to the fol- lowing production function: (4.1) n = PQQ - le x1 - sz x2 P P = output price, price of X1, and price of X 1’ 2 2’ respectively. All prices are exogenous to the firm. That is, firms are per- fect competitors in both input and output markets. Equation (4.2) is maximized by setting the partial derivatives with respect to the variables X1 and X2 equal to zero. (4.3) an -—— = P MPP - P = 0 s> VMP = P 3X1 Q x1 x1 x1 x1 (4'4) .gg = PQMPP - P = 0 s> VMP = P 2 x2 x2 x2 X2 Second order conditions require that the principal minors of the relevant Hessian determinant alternate in sign: 2 2 (4.5) -§—1 = P 3 f("1”‘2) 3x2 - Q 2 < O 1 3x1 2 2 and -341 ==P a fIxI’xz) < 0 3X2 Q 2 2 3x2 33 (4.6) fl 3% 3x: 3x13x2 f11 f12 = P3 > 0 2 2 8.n 3 n £21 £22 3x 3x 3x2 2 l 2 Conditions expressed in.04.5)imply that marginal products of both in- puts are decreasing. They also imply that profit decreases with addi— tional units of X1 or X2. Conditions (4.5) and (4.6) require the production function to be strictly concave in the neighborhood of the point at which first-order conditions are satisfied. The total cost equation is total cost in terms of input prices, input quantities, and fixed costs (FC). (4.7) C = Px X1 I P x + FC 1 x2 2 The total cost function expresses cost as an explicit function of the level of output plus fixed costs. (4.8) C = C(Q) + PC The cost function is obtained by solving the cost equation, production function, and expansion path simultaneously to reduce the system of equations to a single equation in the form of ( 4.8). The cost func- tion gives the minimum cost of producing each level of output. It has input prices as fixed parameters. Marginal cost (MC) is the cost of producing an additional unit of output. Thus, marginal cost is the derivative of the cost function. 34 d[C(Q) + FC] = dQ Q (4.9) MC = Looking at the profit maximization problem from the revenue and output viewpoint, we can consider the determination of the optimal output level. The profit equation is (4.10) W = PQQ - C(Q) — FC. Setting the derivative of profit with respect to output level (Q) equal to zero will give the profit maximizing output level. dn _ “'11) do Q Q (4.12) P = c = MC The second order conditions for a maximum require a negative second derivative. 2 _ 2 (4.13) 9—3- = 9—9- < 0 or cm2 W 2 (4.14) i—C» 0 do2 Equation (4.14) says that marginal cost must be increasing at the pro- fit—maximizing output level. The supply function for an individual firm is given by the pro- fit-maximizing rule (4.12) with output (Q) solved for in terms of output price. Equation (4.15) gives the supply function for the jth firm. The fixed arguments, input prices, are also denoted. (4.15) Qj = sj(PQ.Px1.Px2) Inpy SOl‘v’lng tr levels as gives the of firm 5 SUPPlF in At A dEEan. ..d Glantity r. LUEGS (P E The Ilrm lnp; 35 Input demand functions for an individual firm are obtained by solving the firm's first order conditions (4.3) and (4.4) for input levels as functions of product and input prices. Equation (4.16) gives the demand function for the kth input by the jth firm. (4.16) xjk = Djk(Px1’ PQ), i = 1, ..., k Industry Structural Equations The supply function for an industry is determined by summation of firm supply functions over all firms in the industry. Thus, the supply function for an industry with m firms is given by equation (4.17). I‘ll m _ _ _ (4.17) Q = E Q. = Z Sj(P P , Px ) = S(P Px , P ) At the industry level, output price (P ) is no longer exogenous. Q A demand function for output demand is shown as equation (4.18). Quantity demanded is a function of own price (PQ), prices of substi- tutes (PS), income (Y), and population (POP). (4.18) Q = D(Pq, PS, Y, POP) The input demand for an industry is obtained via summation of firm input demand functions. Thus, using (4.16), the industry demand for input X is given by (4.19). l m x, = z D (P , P , P ) = D (P , P , P ) 31 jl x1 x2 Q l x1 x2 Q m (4.19) X = X :1 j=1 l . J Assuming that the industry is a perfect competitor with other industries for inputs, the input prices would be exogenous (horizon- tal supply curves). Under this assumption, the structural equations 36 for an industry composed of firms producing one output with two in- puts are given by the following system of equations: (4.20) Output Supply: Q = S(P , P’ , P ) Q x x l 2 Output Demand: Q = l)(PQ, P , Y, 1’61’) Input Demand: X1 = Dl (PQ, Px , Px ) l 2 Input Demand: X = D (P , E- , E- ) Exogenous variables are denoted by a "bar." There are four equa- tions in four unknowns. These equations can be solved to get reduced- form equations. Reduced-form equations express endogenous variables only in terms of exogenous variables. ExtensiOns of the theoretical model presented above will be made in the following section as the theory is applied to the farm pro- duction sector. The modification will deal with demand for durables. In the empirical applicaiton, the view of decision making will be that levels of nondurable inputs are determined assuming fixed levels of services from durables. Thus, the level of durable services is a fixed argument of demand for inputs and supply of output equations. Then, when deriving the demand for durables, it will be assumed that the decision is being made for future periods when the durable and nondurable inputs may vary. Demand for Durable Assets The purpose of this section is to formulate a theoretical basiS‘ for specification of investment demand equations. This goal will be a quide and will lead to assumptions that may be less desirable if 37 the goal were to examine the economics per se of investing in and using productive durable assets. Investment, as used here, refers to the quantity of a durable asset purchased in a period (sometimes referred to as gross invest— ment). It will be assumed that it is possible to measure the quanti- ties of durable assets and that each unit of a category of durable productive assets is similar in the respect that it has identical length of life when new, produces identical services, and requires simi- lar inputs to produce services. The investment demand function can be viewed as consisting of two components: adjustment of the stock to a desired level, often referred to as net investment, and replacement investment. Many of the empricial studies of investment have concentrated on estimating net investment (gross investment less replacement investment) or an optimal stock, assuming that replacement investment is a con- stant portion of the durable stock. These include studies by Hall 4 and Jorgenson~(l967), Griliches (1960), and Melichar (1973). Investment demand is surely a derived demand, for there is no economic reason for a firm to desire a stream of investment per 58- Investment is desired because it is economic to acquire the produC’ tive inputs of durable assets. This intuitive reasoning leads one to a formulation of the problem as one of acquiring optinufil stocks of durable assets. Combined with Jorgenson's (1969) justifixnition of replacement investment as a constant proportion of capital stock, one is led to investigating net investment. 38 Feldstein and Rothschild argue convincingly that a technological- ly determined constant rate of replacement isincorrect,even as an asymptotic limit. Some of the terminology defined.in the following is derived from Feldstein and Rothschild. Deterioration is the increase in real resource cost per unit of service output as a durable ages. It is composed of input and output decay. Output decay is the decline in the level of service output as a machine becomes older. Input decay is the increasing requirement of inputs used by a durable (maintenance and repairs) to produce services as it becomes older. Scrapping is the withdrawal of a durable from the capital stock. Depreciation is the fall in the value of a durable as it ages. Depreciation reflects deterioration, obsolescence, and riskiness of older durables. Replacement investment is the actual purchase of durables to maintain the service capacity lost through deterioration (input and output decay) and scrappage. Replacement investment is not identical to deterioration, depreciation, or scrapping. On an intuitive level, one can see that both scrappage and de— termination of the level of service output from a durable are eco— nomic decisions. The level of service output depends upon the levels of inputs used in the production of durable services; this would de- pend upon prices of inputs and outputs as well as technological fac- tors. Scrappage of a durable would occur when the present value of 39 future quasi-rent is equal to the present salvage value of the machine and thus depends upon prices of inputs and outputs. With the consideration that scrappage is an economic decision, that the level of service output is an economic decision, and that replacement investment cannot be distinguished from net investment, the following will attempt to derive gross investment and maintenance (inputs into production of durable services) demand functions. First, an attempt will be made to define symbols and make assum— tions explicit. In a manner similar to the earlier model, f (4.21) is a function - relating input levels to the level of output. f(X V) (4.21) Q 1.‘X2, level of output. Q X1= a nondurable input. X2= a nondurable input. V = level of services from a type of durable. The level of durable services, V, is given by the function (4.22). Variables to the right of the vertical bar are fixed. (4.22) v-= V(X3, 1 1 St, Sc) X 3 I a nondurable input. gross investment in the current year. St = stock of the durable at the beginning of the year. Sc = scrappage of the durable in the current year. The variable cost function, here referring to variable nondura- ble inputs X1 and X2, is given by equation (4.23). It gives the least 40 cost: combination of nondurable inputs for any level of Q and V. It is suolved from the expansion path, cost equation, and production func- tion. (4.23) C = C(Q, V) C = total cost of variable nondurable used directly in production of Q. A concept similar to that for the traditional cost function can be af>p>lied to find a cost function giving the least cost combination of nondurable inputs to produce durable services (4.24). It is a func- tion <3f the level of services and level of investment. (4.24) M = M(V, I) M = cost of maintenance and repairs or the total cost of variable nondurable inputs used in the production of durable services. The objective function (4.25) defines the net present value of the firm, assuming a zero terminal value. It gives the present value 0f quasi-rent over the life of the firm. T (4.25) NPV = f[P Q - C(Q, V) - M(V, 1)] e‘rt dt - P I 0 Q d -l —rT = r <1—e )[PQQ - C(Q. V) - M invest in new durables. It says to equate the change in quasi- rennt (as more of the durables are purchased) to the amortized price of tflle durable. The continuous amortization factor is r(l-e-rT)—1. (Vlternatively, one could leave the discounting factor on the other side of? the equation and have the rule: equate the present value of the (finange in quasi-rent to the price of the durable. If one were to View the firm as making these decisions simulan- eOusly, then the firm derived demand for input equations would be a rEduced form of the four first order conditions. The quantities 44 demanded of X1, X2, X3, and I would be functions of the fixed vari- ables, output price, price of all the inputs (including Pd), the interest rate (r), and the beginning stock of durables. Traditionally, it is not assumed that the quantities of durables purchased can be put into service instantaneously. Thus, Pd and r would not enter the demand for nondurable inputs equations, although the beginning stock of the durable would. If one assumes a two-step decision process where the nondurable input levels are determined, assuming the durable input level fixed and then durable purchases (investment) determined assuming the levels of nondurables fixed, then the prices of the nondurable inputs do not enter the investment equation. This leads to the acceleration principle where investment demand depends, among other things, upon the rate 0f change of output (Eisner, 1969). A somewhat modified view will be used in the empirical portion Of this research. It will view the decision of nondurable input level determination to be made assuming the stocks of durables fixed. The level of durables (investment) will be determined as a plan for the future when nondurables will be variable; thus the prices of nondurables (as opposed to changes in outputs) will be included in investment demand equations. -m‘ .... 45 Further comment at this point is required relative to the investment equation being specified in terms of gross or net investment. The earlier portion of this section argues for scrappage Of durables to be an economic decision. However, the theory is pre— sented assuming scrappage predetermined because of the absence of good data on scrappage. (The purpose of the theory presented here is not to advance theory, but to derive a basis for the equations required in the model.) It is planned to estimate gross investment equations with estimated scrappage as an independent variable. There are several reasons for this. First, the measure of scrappage that will be used is not based on empirical data (Baker, 1978) and thus may not be an unbiased estimate of actual scrappage (i.e., expected Sc = population mean). If the measure is biased, including the scrappage variable as an independent variable would allow the regression coeffi- cient to correct for some of the bias. Second, assuming a reasonable estimate of scrappage, inclusion of scrappage as an independent variable would allow implied replacement investment as an economic decision. The coefficient of scrappage could be interpreted as the quantity of scrappage automatically replaced, with the balance of replacement depending on the levels of other variables (e.g., output prices, price of the durable, and the prices of variable inputs). Theory Applied to the Farming Sector In this section, the structural equations to be estimated for the model will be specified in a general manner. The final selection of 46 variables in each equation, along with the empirically estimated coef— ficients, will be presented in Chapter V. Output Output of the farm production sector will be-aggregated to the aggregate crop and livestock levels for this study. .This breakdown is based upon several considerations, including the following: (1) the demand for crops and livestock is distinctly different--livestock demand is largely derived directly from U.S. consumer demand for live- stock and livestock products, while crop demand is derived from live- stock (feed) demand, export demand, stock demand, and direct consumer demand for crop products; (2) input categories can be divided into inputs for livestock only, inputs for crops only, and inputs used by livestock and crops; and (3) linkages between exogenous and endogenous variables are more direct and have greater meaning when output is divided into more than one componenet. A more detailed breakdown of output could probably be justified based upon the above considerations. Such a breakdown might concentrate on disaggregation of crop output into the categories of feed grains, food grains, fiber, and other. Such a breakdown would add more detail than is necessary for this study. Aggregate data used in this study are presented in Appendix A. For purposes of the analysis of this research, variables of interest are cashremeipts,change in inventory value of livestock and livestock products, and crops. This requires knowledge of production, consumption, and price level. In the following paragraphs, equations for [E Baker units for 5; Lives: tuncti r171 47 for these variables will be specified. The glossary published in Baker (1978) should be referred to for variable name, definitions, units of measurement, data source, and source in the simulation model for specific variables used in the model. Livestock Supply A livestock supply equation expresses quantity supplied as a function of own and input prices and quantities of fixed inputs. The general specification of the supply equation for the model is expressed in equation (4.34). (4.34) Qt = f(Pt-i’ Rt-i’ K) Qt = quantity supplied in year t. P O t—i = price of livestock in year t—i. —t-i = vector of current and lagged input prices, K = fixed factors. It is anticipated that livestock supplied would be responsive to the current year's price, thus causing Simultaneity with demand for livestock. In addition, it is hypothesized that there are effects of past prices on current output. All own price coefficients would be expected to have positive signs. The major input into livestock production is probably feed. The Current year's price of feed, as well as past prices of feed, will be considered in the estimation of the equation. Other inputs that may be important are labor and miscellaneous supplies. Coefficients of in- put prices would be expected to be of negative sign. 48 The lagged effects are expected, in part because of the physical limitations involved in expanding livestock production. Crop Supply When considering the crop supply equation, one must consider the empirical data to be used in estimating the equation. The data series for crops is constructed on a crop year basis that begins with harvest for major crops (see Appendix A). Thus, by definition (based on the physical or technical factors), the supply variable cannot be influenced greatly by prices in the crop year. A supply equation gives quantity supplied as a function of own price, prices of variable inputs, and quantities of fixed inputs. The general form of the supply of crops equation is given by equation (4.35). (4.35) Qt = f1ied, the price of crops as an input in seed production, and the ;>rices of other inputs variable in seed production. These might.in- c=lude labor and supplies. \ 1H/ , “ The terminology "dependent variable" does not really apply here as the system of equations is simultaneous. Thus, as a portion of the identifying restrictions, one normalizes one variable in each equation (gives it coefficient one). This is essentially choos- ing the "dependent" variable for the individual equation, although _all of the endogenous variables are jointly dependent. 54 Demand for Hired Labor In general, the demand for a variable input is a derived demand. It is derived from the profit-maximizing behavior of firms. The quantity of an input is a function of its own price, the price of output(s), the prices of other inputs, and the quantities of fixed inputs. In the case of hired farm labor, it is anticipated that the (quantity demanded of hired farm labor will be a function of its own ptfiice, the prices of crops and livestock, and the quantities of fixed inputs--operator labor and machine services. It would be atlticipated that the fixed inputs are substitutes for hired labor. I’rfiices of other variable inputs such as fertilizer, supplies, feed, (Jr' seed could be included in some specifications of the equation but are unlikely to be important as substitutes or complements. Fe r tilizer Demand It is anticipated that the quantity of fertilizer demanded will 1>€3 £1 function of current fertilizer price, current and past crop prices, C=ITC>p11and acres, and possibly the prices of other variable inputs. It is not expected that the price of hired labor or quantity of oper- E‘t1C>lr labor would be significant complements or substitutes for fertili- Zer - The demand for fertilizer in the current calendar year is anti- CZLIDEitied to be a function of the price of crops in the current crop y‘aéilf (applying fertilizer for the next harvest), as well as a function of QTop price lagged one year. 55 Demand for Other Nondurable Inputs This residual category of nondurable inputs is anticipated to be a function of its own price, the prices of livestock and crops, the prices of inputs such as fertilizer and labor, and the quantities of machinery and cropland. Demand for Durables The model will include structural equations for investment de— xnand for two categories of productive durable assets. These are: 1;) machinery and motor vehicles and 2) service buildings, other struc- ttires, and improvements to land. In addition, a structural equation fc)r the demand for machinery repairs and maintenance is included. £3tzructural equations for investment or maintenance of other categories c>ff durables are not estimated, primarily because of the lack of or low (ltlality of data. Chapter V gives further explanation. Investment Demand for Machinery and Motor Vehicles. Following 111162 derived demand for the gross investment equation derived earlier, the demand for gross investment (quantity) is'a function of the price (>1? t:he durable, interest rate (either as a separate variable or combined W1 th the price of durable via amortization of the price), prices of (117(3133 and livestock, prices of nondurable variable inputs, and quanti- ties of fixed factors. It is anticipated that the price of machinery and motor vehicles “7j*]-1- be adjusted for investment credit. The prices of livestock and (:17(31353 (output prices) may also need to be weighted and combined into one 56 price. An additional alternative for output prices would be to use net cash flow as a variable to represent output prices and the avail- ability or cost of internal funds. The major variable input price anticipated to be included in the equation is the price of hired labor. Alternatively, the quantity of operator labor could be used in the equation. A substitute rela- tionship is expected with either labor variable. Fixed inputs, snach as acres harvested, might be important factors. An indicator of the level of technology may be required, as it seems reasonable that the demand for machinery would expand as new technology is availabe. It is pOssible that this fac- tc>r' is taken into account in the measurement of machinery prices. Estimates of the "quantity" measure stock of machinery and scrap- page have been made and are explained in Baker (1978). Machinery and Motor Vehicle Maintenance and Repair-Derived Demand T11£e view of demand for maintenance and repair of machinery and motor vehicles is that it is a demand derived from the use of nondurable it‘l’llIZS in the production of durable services. ’As such, the quantity demanded is a function of own price, the price of the durable input, tfléi :Llnterest rate (or amortized durable price), the price of output, pr ices of other'inputs, and fixed factors. Investment Demand for Service Buildings, Other Structures and Gross investment for this category of durable .EEII£1__ngprovements. assets is specified in a manner similar to that of machinery and motor vel"HI-Cles. The quantity demanded is a function of own price, the 57 interest rate (or amortized price), the price of output, prices of other inputs, and fixed factors. Deflation of Prices and Money Flows The first order conditions for a firm's profit maximization (or for a consumer-maximizing utility) are homogenous to degree zero in prices.l/ That is, a proportional change in all prices has no effect on the solution to the equations (quantities supplied, demanded, etc.). TThe'existence of money illusion would be contrary to the homogenous In this study, nearly all prices and money tc> degree zero condition. The hypothesis flxyws included in structural equations are deflated. of money illusion is, in general, not tested. One should distinguish between the type of deflating discussed here and the deflation of current dollar flows to get quantity flows. 111ee two concepts are completely unrelated. The choice of a deflator for the general level of prices is the cratiesumer price index for all items. One could construct arguments CC) Lise the CPI to deflate consumer-oriented prices and the implicit This is not done GIVE) (leflator to deflate producer-oriented prices. for the model, since to do so would involve adding another exogenous VEIITjLzalale. It was felt that one price level indicator would suffice. él;E;§33§11ative Approaches The amount of "structure" to include is always somewhat arbitrary, f0)? <>r1e could specify a nearly infinite amount of detail. The approach ‘_.______~§“_________ 1/ SEQ any standard economic theory text. outline L; arm :..1 I lut.UCt .. llfcct Ma .: :abl . U. ETC ,1 . «CIT, 58 outlined in this chapter has presented the traditional approach of market supply and demand functions. Some might argue for modeling of more structure. This might include estimation of production functions to get the direct relationships between inputs and outputs. In addition, one could directly derive input demand equations from the production function, assuming profit maximization. The production function approach has not been attempted for this model for numerous reasons, including the following.. Estimation of production functions for crops and livestock would require a break— down of data on inputs into those inputs used in livestock and those used in crops. Aggregate data in this form are, in general, not avail- able. In addition, it seems to the author, highly unlikely that input demand functions derived from an aggregate production function and first order conditions would fit the historical data series. A modification of the approach used here would be to incorpor- ate risk into the model. While this is an appealing idea, the author has not attempted it, based on the following factors. First, the theoretical basis for aggregate risk models is not well developed. . Static economic theoretical frameworks, such as those alluded to for "portfolio balance theory," basically do not exist. At a minimum, an aggregate risk model would involve variance and covariance estimates for the basic structural variables. Placing these demands on the meager data available_seems outlandish. __' EC. 59 Summary The first portion of this chapter outlined the system to be modeled. The major performance variables of interest were deter- mined to be those appearing on financial statements for the sector. These variables are the results of economic activities of the sec- tor. Economic activities can be expressed mathematically in the form of a set of equations. The solution of this set of equations gives values for the economic variables of interest. Economic theory relevant to derivation of these equations was presented. Application of the theory to the farming sector was made to derive general expressions for the equations to be empirically estimated. r“ ‘ bur". 320 .§ ‘1‘ 5.! I. .Q .1 i n i. {-5 )N CHAPTER V The Empirical Model Introduction The chapters to this point have identified the system to be Incnieled and have developed a theoretical framework. The performance vari— éitxles of interest were determined to be those appearing on financial statements for the farming sector. This chapter presents equations estimated for the model. Where the data and theoretical framework vacelre deemed adequate, structural equations have been estimated. This cic>e=s not, however, provide all of the variables required to prepare a. (romplete and consistent set of financial statements. A significant number of equations have been estimated that are not intended to reeprresent system structure. These equations are based upon many factors, but reflect only the author's judgment that they are the best eqlléltions considering the constraints. I Additional relationships are specified in the model based upon assumptions or single observations of relationships. These rela- tio“Ships are largely in the intersector transfers component, where hot}: (data and theory are absent. Again it should be indicated that the "guiding light" for which variables are necessary is the set of financial statements for the 60 61 sector. The financial statements are presented in Chapter VII. The research process began with a set of financial statements and desired economic relationships to model. These were modified as data were discovered to be adequate or inadequate and as the author gained insight into accounting at the sector level. Additional considerations in specifying the model were related to its use in policy analysis. Specification of the structural de- tail to model and variables in equations was influenced by the need for explicit introduction of policy—related variables whenever possi- ble. In the process of empirical estimation,some ad hoc procedures ‘vere employed in selecting variables for the equations. The total ‘number of variables theoretically appropriate for each equation in- ‘volved a greater number than expected to be included in the final :form selected. Signs unambiguously specifiedby theory were con- sidered aminimum requirement for all coefficients. Some equations :anlude time trends, productivity indexes, or other noneconomic vari- zibles which were required to get the proper economic relationships. Time Index One of the major sources of confusion when trying to Ilrrterpret the equations estimated for the model is the use of the tilnne label. The confusion may arise because the crop and Calendar years differ. “The crOp year variables were constructed fITDm data that began with harvest of the crops. 'In general, this is t(Dward the end of the calendar year. Thus, the crop year is labeled 62 with the second calandar year of the two it overlaps. For example, 4 the 1974—75 crOp year is labeled 1975. Data Period The data period used to estimate most of the equations is 1951 to 1974. There are exceptions which will be noted when necessary. If there is no reference to a data period, it may be assumed that the 24 observations from 1951 through 1974 were used. Statistical Considerations Appendix B gives a description of statistical considerations rele- vant to this model. However, the following items will be useful when interpreting statistical results: 1) simultaneous equations are estimated via two-stage least squares (ZSLS); 2) all nonsimultaneous equations are estimated via ordinary least squares (OLS); 3) all critical values are for one-tailed tests; 4) all equations are linear in the variables unless otherwise stated; and 5) unless other stated, t.statistics, not coefficient standard errors, are given. (Irop Supply Equation (5.1) shows the supply equation for aggregate crOp out— IDLIt (CRPROD). Appendix A gives a description of crop aggregation. frlieatime index for this variable is the calendar year in which har- Vest occurs. CRPROD is CROPS(Z) lagged forward one period. Thus CIKIPROD is crop production in calendar year t. It enters the crop supply 63 utilization identity for crop year t+l and is recursive to a set of simultaneous equations determining crop price and utilization.l/ The supply of crops equation includes as independent variables (1) output price, the real crop price for the year leading to har- vest (RPCROPX and the previous year's real crop price (RPCRMl); (2) the price of an important input, the real price of fertilizer (RPFERT); (3) quantities of fixed inputs, harvested acres (HARACR), and beginning-of—year stock of machinery and motor vehicles per acre (SM4/A); and (4) the productivity index (XPROD) to "capture" changes in technology and weather variations. Expected signs are shown under the variable names in equation (5.1). The empirical results from estimating equation (5.1) via OLS .are shown below in Table 5.1. The short-and long-run price elasti- cities of supply are less than unity (see Table 5.1) . This holds when 'the price and quantity used are mean values over the entire histori- <:al series or for the three most recent years of the data set. At tihe average price and quantity for recent years, the long-and short— Irun elasticities are less inelastic than for the mean price and quan— tzity over the entire data period. Addingzuiadditional lagged cr0p Price resulted in a statistically insignificant coefficient of negative EBifgn. Thus, the lag structure is cut off after using two prices. (S. 1) CRPROD = f(RPCROP, RPCRMl, RPFERT, HARACR, SM4/A, XPROD) + + - + + + 1/ ’— :[t can also be viewed as CROPS(Z), predetermined for the following Year's set of equations. 64 crop production in calendar year t, for use in the following crop year (t+l) supply-utilization identity CRPROD RPCROP = real crop price for crop year t RPCRMl = real crop price for crop year t—l RPFERT = real fertilizer price for calendar year t beginning of calendar year stock of machinery and motor ve- hicles per harvested acre SM4/A XPROD = index of farm productivity, United States Table 5-1. Empirical Results for Equation (5.1), Crop Supply, CRPROD 'Explanatory' Regression Selected Elasticities _Variable Coefficient t Statistic 1951-74 1972-74 Constant -5507.498 --- --— --- RPCROP 36.57623 4.1 .2058 .285a RPCRMl _ . 19.34327 1.4 .314 .435 RPFERT -65.93648 —5.4 --— --— HARACR 14.97038 1.7 .23 --- SM4/A 1.4963 .1 .013 —-- XPROD 232.00885 11.7 -—- -—- 2 _ = .. = = R -' 0993, Dow. 1.06, t.01 .‘ 20898, t.05 20110, t.10 1.740 a; Long run elasticity An "elasticity" calculated with respect to harvested acres is. shown in Table 5.1. This elasticity gives an indication of the re- Srnansiveness of the sector to changes in acreage. As one might ex- ptec:t, a 1 percent change in harvested acres will result in a less than 1 IDGErcent change in crop output. This result has implications for the effectiveness of acreage control policies in controlling crop output. 65 The beginning—of-year stock of machinery and motor vehicles is included in the equation, in spite of its statistically insignificant coefficent, in ordertx>maintain a linkage between the decisions of farmers to purchase machinery and output capacity. This linkage is weak, as indicated by the very low elasticity. One reason suggested for the low statistical significance and elasticity is that the existence of excess machinery capacity has had little direct effect on output resulting from changes in the machinery stock. Alternative specifications of the model included deletion of the productivity index and inclusion ofthe real wage rate and real price of supplies. The exclusion of the productivity index, while not having a great effect on the R2, caused problems with significance levels and signs of coefficients for other variables. Inclusion of additional input prices did not improve the equation. I)emand for Crops The demand for crOps is the sum of several components. These :itmflude feed demand, seed demand, export demand, food-industrial de- nmand, and inventory demand. In the model here, inventories and exports are exogenous. Thus, crop demand estimation consisted of estimating seeed, feed, and food-industrial demand functions. F0 od—Indus trial Crop Demand Equation (5.2) shows domestic per capita food and industrial deuuaIId for crops to be a function of the real price of crops (RPCROP), rea;1 gross national product per capita (RCMP/POP), and the logarithm 0f tiijne. 66 (5.2) CROPS(7)/POP = f(RPCROP, RGNP/POP, LoglOT) . + CROPS(7) = food and industrial crop usage in crop year t. POP = U.S. population. RPCROP = real price of crops in crop year t. RGNP = real gross national product for calendar year t. T = time (l950=l). This demand equation proved difficult to specify empirically in a satisfactory manner. The final equation, estimated via ZSLS shown in Table (5.2), includes a time variable. In addition, the coef- ficient of price is not significant at the normally acceptable level. From a theoretical viewpoint, the demand equation lacks a "price of :substitute" variable. At the level of aggregation for this model, the only price that seemed appropriate, the real price of livestock, teas not statistically significant. Less desirable results were ob— . 1:ained when the equation was estimated on a "total" rather than per (:apita basis with population as an independent variable. frhble 5.2. Empirical Results for Equation (5.2), Per Capita Food-Industrial Demand for Crops, CROPS(7)/POP Selected Elasticities IZXplanatory' Regression ._;Yariable Coefficient t Statistic 1951-74 1972-74 Constant 46.43573 —-- -—- ~-,-- ‘ RPCROP -.00976 .98 -.026 . -.0453 RGNP/POP .001949 4.75 .16 .205 I-aoglO(T) -9.02276 —8.85 --- --- ‘ 2 . R- .91, D.W. = 1.96, t.01 = 2.845, t.05 - 2.086, t.10 - 1.725 67 The variable logloT, the base ten logarithm of time, represents a set of factors causing a reduction in per capita consumption of ' crops over time not explained by price, income, or population. The price elasticity of demand for crops for food and industrial consumption is very low, —.026 using mean values for the 1951 to 1974 period and -.O453 for the recent 1972 to 1974 period. The in- come elasticities are .16 and .205 with variables at their 1951 to 1974 and 1972 to 1974 mean values, respectively. The low elasticities seem reasonable in light of thefacttflmnzthe equation is a "farm level" demand. That is, it is the response in terms of "raw" commodity demanded to changes in farm price. One would hypo- thesize that demand is much less price and income inelastic for services that may be added to the commodity. Feed Demand Equation (5.3) shows feed demand for crops to be a function of tihe real price of feed (RPFEED), current and lagged real livestock [>Iices,and the base 10 logarithm of time. (5.3) CROPS(6) = f(RPFEED, RPLIV, RPLVMl, RPLVMZ, LoglOT) - + + + CROPS(6) = feed usage of crops in crop year t. FU?FEED 8 real price of feed for calendar year t. RIWLIV = real price of livestock for calendar year t. RPLVMl = real price of livestock lagged one year. RPLVMZ - real price of livestock lagged two years- T = time (l950=l). 68 The empirical results from estimating this equation via ZSLS are shown in Table 5.3. Earlier specifications of the equation in- cluded additional input prices. In addition, attempts were made to estimate the equation using the real price of crops rather than the real price 0f feed, thus eliminating the need for a feed supply equation, These alternatives were considered less desirable than the specification (5.3). Table 5.3. Empirical Results for Equation (5.3), Feed Demand for Crops , CROPS(6) Explanatory Regression Selected Elasticities Variable Coefficient t Statistic 1951-74 1972-74 Cons tant —3628 . 84548 --- --— --- RPFEED -l8.07679 -2.28 7 -.22 -.19 RPLIV 42.5208 5.95 --- --- RPLVNl 20.74522 2.03 --- --- RPLV'MZ 11.50612 1.22 --- --- Loglo'r 6079.80507 13.04 --- --- Rzr 97 Dw-164 =2878t =2101t =1734 - ° ’ ° ' " ° ’ t.01 ° ’ .05 ' ’ .10 ° The price elasticity of demand for feed-11 with the variables at their 1951-74 and 1972-74 means is -.22 and -.l9, respectively. The base ten logarithm of time has a positive coefficient and might be interpreted as representing the longer term trend toward greater feed usage not induced by real livestock or feed prices. 33 This elasticity is with respect to feed price. Later, the supply function for feed will be substituted into the demand for feed equation to derive the elasticity with respect to crop price. This is useful when looking at feed demand as a component of aggregate crop demand . 69 Feed Supply Equation (5.4) shows the supply of feed equation (price depen- dent) as a function of the quantity of feed, the price of crops, and the base ten logarithm of time. The price of feed was choSen as the normalized variable in the equation as a result of viewing the system as having feed quantity somewhat more predetermined than feed price on the supply side. That is, on the demand side, feed price allocates the quantity of feed, while on the supply side, the price of crops and quantity of feed used determine the "supply price" of feed. '(5.4) RPFEED = f(CROPS(6), RPCROP, LoglOT) + + RPFEED = real price of feed in calendar year t, CROPS(6) = feed use of crops in crop year t. T = time, 1950 = 1, etc. Table 5.4 gives the ZSLS results from estimating equation (5.4). Table 5.4. Empirical Results for Equation (5.4), Feed Supply, RPFEED M ¥ Explanatory Regression Selected Elasticities _ Variable Coefficient t Statistic 1951-74 1972-74 Cbnstant 4.92227 --- —--a . ---a CROPS(6) .003283 2.47 .27 .32 ItPCROP .86112 11.76 --- --- lgogloT -20.74726 -2.68 —-- —-- Fiz = 96 D W = 2 15 t = 2 845 t = 2 086 t = l 725 ' ’ ° ° ° ’ .01 ' ’ .05 ' ’ .10 ' In this case, these are supply flexibilities. 7O Solved Feed Demand The feed supply and demand functions can be solved to eliminate feed price. This provides an alternative form of the structural . equation for crop demand which allows direct examination of the elas- ticity of feed demand with respect to crop price. 'Table 5.5 shows the solved demand for feed equation. This equa- tion may be viewed as a partial reduced form in that an endogenous variable (RPFEED) has been eliminated from the system of equations but other_endogenous variables (RPCROP, PRLIV) still remain. Table 5.5. Empirical Results for Equation (5. 5), Solved Feed Demand for Crops, CROPS(6) Selected Elasticities Variable Coefficient 1951—74 1972-74 Constant -3509.5467 --- ——- RPCROP —l4.6942 -.186 -.26 PRLIV 40.1387 --- --— PRLVMl . 19.583 ~-- ——— RPLIVZ 10.4367 -—- --- LogloT 5385.1728 --- --- The price elasticity of feed demand for crops is -.186 and -.26 with the variables evaluated at their 1951-74 and 1972-74 mean values. ‘Jhile still inelastic, the feed demand component is considerably more (elastic than food—industrial demand for crops (see Table 5.2). Seed Demand Equation (5.6) shows the demand for seed as a function of: own thrice, real seed price; the price of output, real price of crops; Dirices of other variable inputs, the real prices of labor and supplies; 71 and quantities of fixed inputs, harvested acres. (5.6) CROPS(S) = f(RPSEED, RPCROPS, RPSUP, RPLABOR, HARACR) - + + CROPS(S) = seed use of crops. RPSEED = real price of seed. RPCROPS = real price of crops. RPSUP = real price of supplies. RPLABOR = real price of hired farm labor. HARACR = harvested acres. The seed demand equation proved to be a difficult equation to estimate. Harvested acres did not prove to be a dominant variable as expected,a1though the coefficient is positive and has a t value of 3.03. Amajor problem is related to the signs and statistical sig- nificance of seed and crop prices. The problemprobably relates to the following factors: farmers may not be very price responsive to the price of seed; the price of seed and price of crops are posi- tively correlated (see equation (5.7) the seed supply equation) thus (making it difficult to distinguish between the "price of output" ef- fects of crop price in the seed demand equation and the relationship of crop price as an input price in the seed supply equation; andthereis heterogeneity in seed demand for individual crops, making the aggre- gate seed quantity a poor measure of seed use. The use of simultaneous equation estimators did not improve timestatistical significance of the seed and crop price coefficients. ‘Irlfact, three stage least squares (3SLS) estimation of the equation 72 with the same variables as shown in Table 5.6 results in a "switch- ing of the signs" of both prices and thus is theoretically unac- ceptable. The equation shown in Table 5.6, estimated via ZSLS, is used in the model as it was the best of the several specifications estimated. Table 5.6. Empirical Results for Equation (5.6), Seed Demand for Crops, CROPS(5) fir Explanatory Regression Selected Elasticities Variable Coefficient t Statistic 1951-74 1972-74 Constant ~256.35372 --- --— --— RPSEED -.365018 -.68 -.ll -.12 RPCROP .344871 .83 —-- --- RPSUP 1.63481 1.60 --- -—- RPLABOR 2.5682 3.81 --- ——- HARACR .65568 3.03 -—- -—— 2 . R .87, D.W. - 1.85, t.01 - 2.878, t.05 - 2.101, t.10 — 1.734 The price elasticity of demand for seed with the variables at their 1951-74 and 1922-74 means are -.11 and -.12, respectively. Seed Supply Equation (5.7) gives the supply of seed equation. In a manner Sindlar to the feed supply equation, the real price of seed is the Inormalized variable. The equation shows the supply price of seed as a function of the quantity supplied and the price of the major input, the price of crops. The ZSLS estimated results of equation (5.7) are presented in Table 5.7. 73 (5.7) RPSEED = f(CROPS(S), RPCROP) + + RPSEED = real price of seed for calendar year t. CROPS(5) = seed use in crop year t. RPCROP = real price of crops in crop year t. Table 5.7. Empiricallkamiltsfkanquation (5.7), Seed Supply, RPSEED Selected Flexibilities Explanatory Regression Variable Coefficient t Statistic 1951-74 1972-74 Constant -58.29288 --- --- --- CROPS(5) .29978 5.61 .97 .93 RPCROP .546898 8.21 —-- --- 2 _ - = = R - .85, D.W. — 1.83, t.01 - 2.831, c.05 2.080, c.10 1.721 The seed supply flexibility as shown in Table 5.7 is nearly unitary with the variables evaluated at both their 1951-74 and 1972- 74 mean values. Solved Seed Demand The partial reduced form equation (5.8) for seed demand is shown in Table 5.8. It is obtained via substitution of the seed supply equa- tion into the seed demand equation to eliminate the real price of the seed variable. It is interesting to note that the coefficient of real crop price is positive. This indicates that the positive influence 011 the demand for seed (as the price of output variable) outweighs tlie negative influence of crop price (entering via the positive relation bEitween crop price and seed price in the supply of seed equation). 74 Table 5.8. Empirical Results for Equation (5.8), Solved Seed Demand for Crops, CROPS(5) Selected Elasticities Variable Coefficient 1951-74 1972-74 Constant -211.8945 --- ' --- RPCROP .10932 .041 .035 RPSUP 1.4736 —-- —-- RPLABOR 2.3149 . --- --- HARACR .59102 -—- --- Livestock Demand Equation (5.9) shows the per capita demand for livestock as a function of the real price of livestock, real per capita disposable income,and the base ten logarithm of time. (5.9) LIV(5)/POP = f(RPLIV, RDI/POP, LoglOT) ' - + LIV(5) = U.S. livestock consumption in calendar year t, POP = total U.S. population on July 1, year t. RDI = real disposable income. T = time (year - 1949). As was the case in the industrial-food demand for crops, a "price (of substitute" variable was not included in the demand for livestock (equation. While the equation estimated via ZSLS (see Table 5.9) has I>roper signs and reasonable statistical significance for own price 21nd disposable income, it was necessary to include a time trend vari- alile to achieve these results. The time variable indicates a reduc- lzion in per capita demand for livestock caused by factors other than livestock price and income. 75 Table 5.9. Empirical Results for Equation (5.9), Demand for Livestock, LIV(5)/POP Explanatory Regression Selected Elasticities Variable Coefficient t Statistic 1951-74 1972-74 Constant 110.19175 --- --- --- RPLIV -.089916 -2.25 -.10 —.15 RDI/POP .006682 2.82 .17 .21 LoglOT - -19.l997 -4.64 --- --- 2 R = .80, D.W. — 1.69, t.01 - 2.845, t.05 - 2.086, t.10 — 1.725 The livestock demand equation used in the model (Table 5.9) has fairly low price elasticity with variables evaluated at recent and entire sample period mean values, although both income and price elas- ticities are higher for the recent period. Livestock Supply Equation (5.10) shows the quantity of livestock supplied as a function of current and lagged real livestock prices, the real price of feed (an important input) and a time trend. (5.10) LIV(2) = f(RPLIV, RPLVMl, RPFEED, LoglOT) + + - LJV(2) = quantity of livestock supplied in calendar year t- I?RLIV = real price of livestock in the current calendar year. IiPLVMl = RPLIV for the previous calendar year. IIPFEED = real price of feed in the current calendar year. T' = time (year - 1949). 76 The empirical results from ZSLS estimation of equation (5.10) are shown in Table 5.10. The inclusion of the time trend variable was necessary to get the desired signs and reasonable statistical significance of the other variables. It is hypothesized that the time trend represents improved technology in the livestock industry over time. The productivity index did not work as well in the equa- tion as did loglOT. This is probably because most of the variables reflected in the productivity index are the result of weather and technology associated with crop production. Table 5.10. Empirical Results for Equation (5.10) , Livestock Supply, LIV(2) Explanatory Regression Selected Elasticities Variable Coefficient t Statistic 1951-74 1972-74 Constant 7857.864 --- --- --- ‘ RPLIV 40.3453 3.09 .248 .31a LVMl 28.10999 1.84 .41 .53 RPFEED -23.4773 -l.56 --- --- LoglOT 5196.8917 8.16 --- --- 2 R = .86, D.W. - 1.44, t.01 2.861, t.05 2.093, t.10 - 1.729 aLong run elasticity The short run elasticity of Supply (.31 for the recent years' data) indicates a reasonable capacity to change output of livestock and livestock products in a short period of time. This elasticity is based on a coefficient with high statistical significance. The real price of livestock lagged two periods was dropped from the equa- tion because of low statistical significance. 77 Demand for Other Non-Durable Inputs Equation (5.11) shows the formulation of derived demand for other nondurable inputs. This category of inputs includes items such as pesticides, utilities, and ginning. The own price variable used is the index of pricestxxkifor farm supplies. The equation shows quan- tity demanded as a function of own real price, the price of output, the fixed stock of a substitute durable input, and a time trend. (5.11) OTHER = f (RPSUP, RPTOUT, SMMV4, LoglOT) - + + OTHER = quantity of other nondurable inputs. RPSUP = real price of supplies. RPTOUT = real price of total output. SMMV4 = stock (quantity measure) of machinery and motor vehicles. T = time (year—1949). The empirical results for equation (5.11) are shown below in Table 5.11. Other specifications of the equation included real fertilizer price and the real price of hired labor. The fertilizer price coefficient was not significantly different from zero. The price of labor is highly correlated with time and had a highly significant positive coefficient (in the absence of the time variable) when included in the equation. The author did not feel that labor and other nondurable inputs were substitutes to the extent indicated by the equation. Instead, it seems more reasonable to include the time trend to represent the increased use of intermediate inputs such as pesticides and herbicides over time as new technologies have evolved. 78 Table 5.11. Empirical Results for Equation (5.11), Derived Demand for Other Nondurable Inputs, OTHER Explanatory , Regression Selected Elasticities Variable _ Coefficient t Statistic 1951-74 1972-74 Constant 10530.4645 -—- -—- —-— RPSUP -74.9380 -7.36 -2.29 -1.34 RPTOUT 22.5522 6.48 —-— -—— SMMV4 -.0841 -3.27 --- -—- LoglOT 2831.3732 5.05 --— ——- 2 _ = = = R — .98, D.W. 1.62, t.01 2.861, t.05 2.093, t.10 1.729 Table 5.11 gives price elasticities of the derived demand with variables at their mean values for the whole data period and the recent three year period. Both elasticities are greaterthancnuatnn:with variables at recent levels, the elasticity is much lower. The nega- tive coefficient on the stock of machinery and motor vehicles (SMMV4) indicateszisubstitute relationship between other nondurable inputs and services from durables. When entered as separate variables, the prices of livestock and crops had different signs and low'statisticalsignificance. This prompted the usage of RPTOUT which is aweighted average of RPCROP and RPLIV. Immand for Hired Farm Labor Equation (5.12) shows the formulation of the derived demand for hired farm labor. The quantity demanded is shown to be a function of own real price, the real price of output, and the quantities of fixed inputs (family labor and the stock of machinery and motor vehicles). 79 (5.1?) LABOR = f(RPLABOR, RPTOUT, FAMILY, SNMV4) _ + _ _ LABOR = quantity of hired farm labor. RPTOUT real price of total output. FAMILY number of operator and family workers. SMMV4 = stock (quantity measure) of machinery and motor vehicles. The empirical results for the labor demand equation are shown in Table 5.12. Alternative specifications of the equation include using the prices of crops and livestock as separate variables and in- clusion of acres harvested. When the prices of livestock and crops are entered as separate variables rather than combined in RPTOUT, they both have significant positive coefficients.. Other coeffici— " values remain about the same. The equation including ents and "t RPTOUT was chosen because of slightly larger t values and a slightly smaller standard error of estimate (98.8 vs. 104.6). Cropland acres harvested had a very significant negative coeffi- cient when included in the equation. This would imply a substitute relationship between hired labor and land which is contrary to the author's expectations. Thus, the variable is excluded from the equation. Table 5.12. Empirical Results for Equation (5.12), Demand for Hired Farm Labor, LABOR Explanatory Regression Selected Elasticities Variable Coefficient t Statistic 1951—74 1972-74 Constant 10049.2 --- --- -—— RPLABOR -44.0789 -7.60 -l.11 -l.66 RPTOUT 15.4469 5.71 --- --- FVOELY -412.466 -4.94 --- --— SPOW4 . -.039701 -3.73 --- ’ --- R2 = 95 D W = 95 t = 2 861 t = 2 093 t = l 729 ' ’ ° ° ° ’ .01 ° ’ .05 ° ’ .10 ' 80 Table 5.12 shows the price elasticity of demand for hired labor to be -1.11 with variables at their 1951 to 1974 mean values and -l.66 with variables at the 1972-74 mean values. The Durbin-Watson sta- tistic is at the lower end of the inconclusive region (dL at the 1 percent level is .80). This would indicate a problem with first order auto-correlation. Using the Cochran and Orcutt procedure for correcting first order auto-correlation results in a substantially different coefficient for the stock of machinery and motor vehicles and somewhat different coefficients for output prices and family workers. The sign of the coefficient of machinery stock changes and is not statistically significant (it = . 14) . The size of the coefficient of output price is reduced from 15.45 to 8.93 (t value from 5.7 to 3.4),and the family workers coefficient is changed from -412.45 to -287.87 (t value from 4.9 to 3.4). The equation without the correc- tion for auto-correlation was chosen for use in the model largely because of the author's belief in a substitute relationship between machinery and labor over the relevant range. Demand for Fertilizer Equation (5.13) shows the formulation of the derived demand for fertilizer. The equation shows the quantity of fertilizer demanded as a function of real fertilizer price, current and lagged real price of crops, and the real price of supplies. (5.13) FERT = f(RPFERT, RPCROP, RPCRMl, RPSUP) - + + 81 FERT = Quantity of fertilizer used in calendar year t. RPFERT = real price of fertilizer for calendar year t. RPCROP real price of crops in crop year t. RPCRMl real price of crops in crop year t-l. RPSUP = real price of supplies in calendar year t. The empirical results of equation (5.13) are shown in Table 5.13. It is interesting to note that lagged price of crops has a coefficient about twice as large as current crop prices. It might be reiterated here that crop price in year t is the price for the crop year beginning the previous fall and extending until harvest. The results indicate that there is some positive response of fertilizer demand to cr0p pricechangesudthin theyearlnnzthat a larger determinant of fertilizer demand is the crop price for the previous year. Table 5.13. Empirical Results for Equation (5.13), Demand for Fertilizer, FERT Explanatory Regression Selected Elasticities Variable Coefficient t Statistic 1951-74 1972-74 - Constant 9451.26 --- -—- —-_ RPFERT ' -21.770l -3.45 -1.28 -.63 RPCROP 21.2711 5.21 —-- --- RPCRMI 40.0646 3.88 —-- --— RPSUP -113.6409 -6.37' --- --- 2 ‘ _ _ I! = .95, D.W. — 1.89, t.01 — 2.861, t.05 — 2.093, t.10 — 1.729 The empirical results show other nondurable inputs (OTHER) and fertilizer to be highly complementary inputs (indicated by the nega- tive coefficient for real price of supplies). The coefficient probably 82 overstates the "economic" relationship and partially reflects the correlation between a technologically related decline in RPSUP and increased fertilizer usage. The results indicate a large differ- ence between the price elasticity of demand for fertilizer with the variables at their means for recent years and over the whole data period. In recent years, there has been a movement into the inelastic portion of the linear fertilizer demand function. One of the variables hypothesized to be an important input, har- vested acres, did not have a statistically significant coefficient and thus was excluded from the equation. Petroleumg Fuel, and Oil Expense The cash expenditures for petroleum,fuel, and oilannaestimated using the price of motor supplies and the "quantity measure" stock of machinery. The equation is not intended to be a structural equa- tion giving the demand for petroleum, fuel, and oil. The data are not adequate to support that effort. The specification chosen is based on information in Agricultural Handbook 365, VOlume 3. The handbook suggests that numbers of motor vehicles, average fuel consumption, and the price of petroleum products are used in data calculations. The model does not contain data on the numbers of maChines; thus the stock measure of machinery and motor vehicles was used in the equation. 83 Table 5.14. Empirical Results for Equation (5.14), Expenditure for Petroleum, Fuel, and Oil, PETRO Explanatory Regression variable Coefficient t Statistic Constant -403.6385 --- PMOTSUP 16.9919 18.70 , 8M4 .0051055 1.51 2- - = = R — .947, t.01 - 2.831, t.05 2.080, t.10 1.721 PMOTSUP = price index of motor supplies. 8M4 = stock "quantity” measure of machinery and motor vehicles. The coefficient of 3M4 is not statistically significant but is of the expected sign and therefore is left in the equation. Investment Demand for Machinery and Motor Vehicles Equation (5.15) shows the gross investment demand equation for machinery and motor vehicles.. Gross investment is shown to be a function of real amortized price (the own price variable), the begin- ning of year stock of machinery and motor vehicles, the estimated scrap- page of machinery and motor vehicles, the real net cash flow (represents price of output and internal availability of funds) and the acres per farm (represent the technology being used and changes in operator labor). (5.15) ‘EXPMMV = f(RAMORTM, SM4, DM4, RNCF, ACRFRM) - - + + + EXPMMV' = gross investment (quantity) in machinery and motor vehicles. ‘KAMORTM = real amortized price of machinery and motor vehicles. 84 SM4 = stock (quantity) of machinery and motor vehicles. DM4 B approximated scrappage and output decay of machinery and motor vehicles. RNCF = real net cash flow. ACRFRM = harvested acres per farm. Table 5.15 presents the empirical results for equation (5.15). Several other specifications of the investment demand equation were estimated. Among these were separate usage of the machinery price and interest rate. Separate usage resulted in positive coefficients for both variables, a result contrary to the theoretical expectation. When combined into the single variable RAMORTM (via amortization of the machinery and motor vehicles price index adjusted for investment credit and deflated by the CPI), the coefficient becomes negative but not statistically significant at normally acceptable levels (t value is -1.04). The real price of farm output was substituted in the equation for real net cash flow:hlanalternativeformulation. The variable had a positive significant coefficient, but the equation with real net 1 cash flow was chosen because of a higher R2. The stock and scrappage measures and harvested acres per farm liave significant coefficients of expected sign. The scrappage (DM4) 'variable hasarcoefficientsubstantially less than one (.219),giving EMDmB indication that replacement investment is not automatically equal tC> scrappage (assuming DM4 is a reasonably good estimate of scrappage). 85 Table 5.15. Empirical Results for Equation (5.15), Investment Demand for Machinery and Motor Vehicles, EXPMMV Explanatory Regression Selected Elasticities Variable Coefficient t Statistic 1951-74 1972-74 Constant 7679.71 --— --- —-_ RAMORTM —266.187 -1.04 -.5558: -.5545: —.5545 -.5541 -.2476C -.2821C SM4 -.118741 -4.41 4—- -—- DM4 .218561 2.11 --- --- RNCF .0600804 2.43 --- —-- ACRFRM 38.8175 2.25 --- --- R2 = .78, D.W. = 1.93, t.01 = 2.878, c.05 = 2.101, t.10 = 1.734 is elasticity with respect to RAMORTM. b is elasticity with respect to RAPMMV. C is elasticity with respect to RBK. Elasticities have been calculated with respect to real amortized price, real price of machinery and motor vehicles adjusted for invest- ment credit, and the interest rate. With variables at their mean values for the 1951-74 period, the elasticities are -.5558, -.5545,' and -.2476 with respect to RAMORTM, RAPMMV, and RBK, respectively. 'Demand for Machinery and Motor Vehicle Maintenancg_and Repair Equation (5.16) shows the demand equation for repairs and main— tenance of machinery and motor vehicles. The quantity demanded is .a function of the amortized price of new machines, the beginning of Year stock, scrappage during the year, and the real price of hired farm labor. 86 (5.16) RREPM .. f(RAMORTM, SM4, DM4, RPLABOR) + + - - RREPM = quantity of repairs and maintenance of machinery and motor vehicles. RAMORTM 3 real amortized price of machinery and motor vehicles. SM4 = beginning of year stock (quantity measure) of machinery and motor vehicles- DM4 = estimated scrappage and output decay of machinery and motor vehicles, RPLABOR = real price of hired farm labor. Empirical results for equation (5.16) are shown in Table 5.16. Table 5.16. Empirical Results for Equation (5.16), Machinery and Motor Vehicle Maintenance andRepairIknived Demand, RREPM Explanatory Regression Variable Coefficient t Statistic Constant 1962.75 --- RAMORTM 137.283 2.16 SM4 .0130701 1.98 DM4 -.0743533 -2.74 RPLABOR -21.0810 -5.03 2 _ R = .90, D.W. — 1.54, t.01 = 2.861, t.05 - 2.093, t.10 — 1.729 The equation does not contain an "own price" variable per se in that the cost of maintenance and repair usually involves a combin- ation of parts, resources (such as oil), and labor. RAMORTM repre- sents the price of a substitute; that is, new machines substitute fiat repairing old machines. The stock of machines was expected to luive a positive coefficient,as one would expect more repairs with a greater number of machines. Scrappage (DM4) was expected to have a 87 negative coefficient as repairs should go down as more old machines are scrapped. The real price of labor is included to represent a major cost of repairs and maintenance. In addition, RPLABOR prob- ably picks up the effects of increasing prices of parts. In an alternative specification,the real price of machinery and motor ve— hicles (used to deflate maintenance expenses) was included in the equation as a separate variable (it also enters in an altered form via RAMORTM) to reflect the price of parts, but the coefficient was notstatisticalLysignificant. In the absence of RPLABOR,inclusion of RPMMV has a significant negative coefficient, appropriate for an Jown price variable, but the coefficient of RAMORTM became negative and wastxnzsignificant. The real price of total output was also included in an alternative specification. The coefficient of RPTOUT had the wrong sign (negativeL,was not statistically significant, and thus is not included in the equation. Investment Demand for and Repairs of Farm Buildings An extensive set of data on capital expenditures and repairs of farm buildings is published by the USDA.l/ The building categories include 1) farm operator dwellings and 2) other and land improvements. (kitegory two includes service buildings, other structures, fences, Willdmills, wells, dams and ponds, terraces, drainage ditches and tiles lines, other soil conservation facilities, and dwellings not 1/ ‘r Charrently in Farm Income Statistics,formerly called Farm Income Situation. 88 occupied by the farm operators. Data are published for both cate- gories for expenditures, repairs and maintenance, and depreciation. Scott and Heady (1967) used the data on building expenditures (or transformations of the data) to estimate regional demand for farm buildings. This prompted a series of comments, replies, and rejoinders, by Grove (1969, 1970), O'Dell (1969), and Scott and Heady (1969, 1971) over the inappropriate use of the data. The basic point was the use of farm income to explain expenditures when the data for expenditures wereestimated from changes in income. This method of estimating the data is apparently no longer used.l/ For a reasonably detailed, but possibly out of date, review of the building data series, see Bhatia (1971). The frequently refer- enced Agricultural Handbook 365, Volume 3, published by tiKEUSDA appears to be of little value in determining the procedures used to estimate these USDA data series. The July, 1974 Farm Income Situation has significant revisions of the expenditures on buildings and land improvements back to 1959. This revision reverses a downward trend in expenditures and leaves a seemingly large jump in the data between 1958 and 1959. To the «extent of the author's knowledge, there is no published explanation ()f this large data revision. Itis the author's understanding that the expenditures for faznn operator dwellings data series is moved from year to year ac- corniing to changes in total building expenditures. In addition, ll ' * Personal discussion with David L. Kincannon, USDA, ERS. 89 repairs and maintenance are estimated as a percentage of capital ex- penditures for both farm operator dwellings and other buildings and improvements to land. One final point is that expenditures for service buildings, other structures, and land improvements are published as a total. Pre- sumably there is more faith in the sum of the two components than the individual components (see Bhatia, 1971, p. 493). Thus, although the data are fairly complete, the "real" informa- tion is not as extensive as a casual look at the data would suggest. The following sections give the equations used in the model to pro- ject the above data series. In sOme instances, the procedures are ad hoc but seem, to the author, apprOpriate for the data Investment Demand for Service Buildings, Other Structures, and Land Improvements Equation (5.17) shows the formulation of the investment demand equation for service buildings, other structures, and land improve- ments. The equation includes variables representing own price, begin- ning stock, scrappage, and the price of output. In addition, harvested acres per farm represents several factors. The quantity of operator .labor (a Substitute variable) is reflected in the denominator. If there are fewer farms, the ratio increases. The ratio also is an :hndicator of changing technology over time. As farm size has in- creased, so has the availability of new types of buildings and other structures 0 90 (5.17) EXPBLD = f(RANORTB, 8B4, DB4, RPTOUT, ACRFRM) - - + + + EXPBLD = gross investment (quantity) in service buildings, other struc- tures, and land improvements. RAMORTB = real amortized price of buildings. 8B4 = stock (quantity) of buildings. DB4 = approximated scrappage and output decay of buildings. RPTOUT = real price of total output. ACRFRM = harvested acres per farm. Table 5.17 gives the empirical results from estimating equation (5.17). Table 5.17. . Empirical Results for Equation (5.17) , Investment Demand for ServiceBuildings,OtherStructures,amdlmprovementstx>Land, EXPBLD Explanatory Regression Selected Elasticities Variable Coefficient t Statistic 1951-74 1972-74 Constant 1651.32 --- ~ --- --- RAMORTB ' -198.859 -3.14 -1.31: -1.41: -l.31 -1.40 -.558° -.737c SB4 -.0639205 -l.83 --- --- DB4 .0819004 1.59 --— --- RPTOUT 5.07174 2.52 --- --- .ACRFRM 14.5637 2.18 —-- --- R2 - .89, D.W. = 1.36, c.01 = 2.878, t.05 = 2.101, c.10 = 1.734 a is elasticity with respect to RAMORTB. is elasticity with respect to RAPBLD. c is elasticity with respect to RFLB. As indicated earlier, the nature of the data used as a basis for (“1&3 equation is in question. Although the author felt there was enougfli information in the data to estimate a structural equation, one 91 should view the results cautiously. The data period 1951 to 1974 was used to estimate the equation. The low Durbin Watsonl/ statistic may result in part from the previously mentioned jump in the data between 1958 and 1959. The stock and scrappage variables are not significant at high levels but have been included in the equation because of their theo- retical importance, correct sign, and the lack of other variables to measure the concepts they represent. The elasticities with respect to own price factors of RAMORTB, RAPBLD, and RFBL with variables at their 1972-74 mean values are -1.41, -1.40, and -.737, respectively. Repairs and Maintenance of Service Buildings, Other Structures, and Land Improvements As indicated earlier, it is the author's understanding that the historical data series on repairs and maintenance are estimated from the expenditure series. Thus, no attempt is made here to present structural equations. Equation (5.18% giving current dollar expenditures for repairs’ and maintenance as a function of current expenditures for service Insildings, other structures, and land improvements, and a dummy vari- able is shown in Table 5.18. The equation was estimated using 1955 to 1974 data. Several of the: equations intended to reproduce USDA procedures for estimating l 'J The Durbin Watson statistic is in the indeterminate range (du = 1.01, dL = 1.79) at the .05 level. 92 Table 5.18. Empirical Results for Equation (5.18), Repairs and Maintenance of Service Buildings, Other Structures, and Land Improvements, REI’BLD Explanatory Regression Variable Coefficient t Statistic Constant 272.836 --- CE(2) .206476 20.37 DUMZ 17.2526 4.67 2 — — = = R = .9856, D.W. - 2.22, t.01 - 2.878, c.05 2.101, t.10 1.734 CE(2) = current dollar expenditures for service buildings, other structures and land improvements. DUMZ = a binary variable, 0 for years before 1973, 1 for 1973 and later years. data fit much better fordata after 1955 possibly indicating consisten- cy in procedures after that time. Equation (5.18) includes a dummy variable for recent years under the hypothesis that there may have been a shift in the data series, possibly based on new benchnarks. The use of the dummy variable was prompted by a pattern of residuals“ clearly indicating that something had changed (the dummy variable is quite significant as the t ratio is 4.67). Expenditures for Farm Operator Dwellings Although it was indicated to the author that expenditures for faann operator dwellings are estimated as a proportion of total expen- ditures for all building, a clear identity equation was not to be fOUJId from the published data. The equation used in the model, equation 93 (5.19),includes a time trend and a dummy variable for 1974. Table 5.19 shows this equation. Table 5.19. Empirical Results for Equation (5.19), Expenditures for Farm Operator Dwellings, CE(3) Explanatory Regression Variables Coefficient t Statistic Constant 425.952 --- CE(2) + CE(3) .145323 2.82 DUM 182.727 2.68 T -10.2616 —3.12 2 —o R = .7346, D.W. - 2.22, c.01 — 2.845, c.05 - 2.086, tolo — 1.725 CE(3) = current dollar expenditures for farm operator dwellings. CE(2) = current dollar expenditures for service buildings, other structures, and land improvements. DUM = a binary variable, 1974 and later years equal to one, zero otherwise. T = time (year - 1949). The data period used for this equation is 1955 to 1974. Resi- duals of an equation including 1950 to 1954 data indicated thatzadiffer- ent method of estimating the historical data was used for the earlier period. The time trend was clearly indicated by the pattern of residuals from an equation excluding time. This may be the result of nxyving to a different coefficient for total expenditures over time. It: addition, there was a large jump in dwelling expenditures in 1974. 1t.:1s assumed that this is a result of shifting to a new benchmark. 94 In the model, the coefficient for DUM is simply added to the constant for projections. An additional source of error in the equation is the inclusion of land improvements in expenditures (explanatory variable) which,to the extent of the author's knowledge,are not included in the actual identity that produces this USDA data series. Repairs and Maintenance of Farm Operator Dwellings The equation for repairs and maintenance of farm operator dwell- ings is shown in Table 5.20. It is estimated on data for the 1955 to 1974 period. The equation chosen for use includes a time trend variable which is statistically quite significant. With an R2 of .9951, the equation is nearly an identity. The residuals for recent years did not differ substantially from other years, indicat- ing that while different procedures may be being used for estimating expenditures,the procedures for estimating repairs are probably still the same. Table 5.20. Empirical Results for Equation (5.20), Repairs and Maintenance of Farm Operator Dwellings, REPDWL Explanatory Regression Variable Coefficient t Statistic Constant 2.68799 --- CE(3) .565384 56.70 T -1.77588 -l3.08 2 R =.9951, D.W. - 1.84, t.01 = 2.898, t.05 - 2.110, t.10 — 1.74 95 REPDWL = current dollar repairs and maintenance of farm operator dwellings, CE(3) = current dollar expenditures for farm operator dwellings. T = time (year - 1949). Real Estate Price and Transfers There have been a large number of studies of the farm real es- tate market. Among these are Tweeten and Martin, Herdt and Cochrane, Reinsel, Melichar (1973), and Nelson. While a fair amount of theorizing is done in the above studies and others, empirically a large number of sets of variables will produce reasonable equations with high Rz's (at least equations with some rationalization). The author has made no attempt in this study to further the theory as it relates to farm real estate prices, although it is the author's opinion that previous authors have not properly specified and/or examined quantity supplied and quantity demanded variables. Real Estate Price The equations estimated here for real estate price (probably a demand equation) include variables standard in the above studies, except :in the study by Reinsel, and some ad hoc experimenting. Table 5.21 presents the results of estimating three real estate Irrice equations, one with nominal price as the dependent variable ar1d two with real (deflated) real estate price as the dependent vari- al>Jue. The basic independent variables considered are net cash flow (Iresal net cash flow), the consumer price index (used as a deflator 96 and as an independent variable), and the number of farms. The index of farm real estate value per acre is the value reported as of March lst. All independent variables are values for the previous year. The author has some faith in the coefficients of the income variable and general price level indicator (CPI). The meaning of the number of farms, however, is not clear. The negative coefficient might indi— cate that as there are fewer farmers, there is less competition for land and thus price is lower. It might mean that there is a simul— taneous relationship and that as real estate price goes up, pe0ple sell out, resulting in fewer farms. Alternatively, the relationship might be purely spurious, only indicating that land price has gone up over time while the number of farms has gone down. The coefficients for net and real cash flow were statistically significant in all equations. New cash flow rather than net farm in- come was used in the equations because it would represent, in addition to profitability, the availability of internal funds (there is no other cost of funds variable in the equation). There is frequently ad hoc theorizing that land is purchased as a hedge against inflation; there is also pOpular Opinion that land prices increase faster than the rate of inflation. This later hypo- thesis is tested in two ways. In equation (5.21a), the real estate Ibrice is not deflated, and the CPI is included as a separate variable. TIWJe test that the coefficient is 1 versus the alternative that it is tucnzlq yields a t statistic of 1.41, which gives about a .10 level 97 mfiumm mo mopeds u m2m can CH nmmcwa mdoaumscw Hammw mma.H u 0H s .emo.~ u no u .mqw.~ u as u Aem.auv Am.mV Aam.sv am.H mews. Hst.aH- smaomm. amqoouoo. meam.mm aHac3m unnumm Hmwm .mH m 1-14-111 9 a _ Gammnmm H cameos mumumm Hmwm .na 1 z 1. 1 .cn1wmm1 1 H scmnums>cH “massage .01 1 1 . .1 1 . . .rugnnu _ . 1 mcamomm ptmcanum: .ma 1 . H 1 x 1 x . Rumba“ acseumo>cH acmcacumz .qa m a 1r. K sommu1 cannon tueuo .ma _ a x K “WM-WW UCNEOQ MOO-QJ .NH w I'l-1I|Iiivllliv|ll4vlll11|lI-AvluI' H 1 K ./Hl1orrA|Ilhw II.Vl'IIATI"AVIIIIIII—llll|'lvlllu "' - nrquwa ”meHfiUerw .Hfl 1” a. - 1 . 1 1 1.1 . I A-vpfiu 1 . 1 >uwucon zoo mm> . u M u x ran 1 H . a as ea . ..11 u H _ ocaa . 1.1.51 . . 1 . La 1,333.53 .11 1 v: . 1 a a _ . a.) u “ "dl'III-dl‘llll‘I'al'A7--- ‘|"—"I| I a A \ . . QCHLHHM _ . . 1 1 Ilillllvlllfllll1vlll1ll|elll£l|livlll .1 PHOODm xUOumw5.H.H .m . . — _ _ 0‘l'v . -- 1 1 . _ 1 _ T T T 1 . Dun-um“. 1 1 u r .zuquUUH mQOHU .\ 1 1 .- M u u a H x 1 Amvmacmu_ 1 n ” zaacum poem .0 . _ 1 . .1 ‘ m 111. . H u 1 _ > 7 m Input“ N 1 _ n 1 ncmfima boom .m _ . 2.. n _ u 1 u 7 1 m ”A . ALVVJGLWUW . 1 w H m .IHQQDW CNN..." .Q _ 1 ”r. 1 . - J ,1 1 1 _ 1 1 . Anvunrutu . _ “ vrmfima comm .m 1 m u _ n A _ non1 . II- 3.... l..l.l1 . 1 C ..a M J Anvmduw ” 1 fl r r111:1;pwunnrnanvaaurnnarnnsrvlnr 1 1 p m u mean no u amen-msvcmnnoow .n . . «1 1fii|| I‘lllt-||*||| ll . l l. 1 1 1 . 1 1 1 1 1 - -7 . II. . _ _ . _ _ . H . 1- ./ I 1 1 _ 1 parrsm aouu .H m . _ 1 1 newunwccmun cofiumnvu ‘ . u h H O“ 1‘ V .0 p» .Jpv~“ u MW 0 O“ .l m .D_.~ 103 system; and 6) demand for categories of durables and demand for re- pairs and maintenance of the same are not simultaneous. Equation one, the crop supply equation, would become part of the simultaneous system if current crop price were included in the equation. If the prices of variable inputs (e.g.,PFERT, PLABOR, PSUP) were endogenized, the demand for inputs (plus the added supply func- tions) would be simultaneous with the supply-utilization equations. The real estate equations would be simultaneous if acres sold were to be included in the demand equation. Britros (1976) presented a "statistical theory" and estimated equations for capital expenditures and repairs. He argues that re- pairs and expenditures are simultaneous. However, the theory out- lined in Chapter IV'doesnot derive quantities of repairs as a func- tion of the quantity of expenditures and vice versa, as Britros includes in his equations. The theory indicates that quantities are functions of prices, and thus the demand equations are not simultan- eous unless input supply equations are included in the model. _Bgduced Form Equations Equations 2 through 10 in Table 5.24 can be solved for a gibven level of population. That is, the sub-matrix can be inverted. Flaom this and an appropriately reduced matrix B, one can get reduced fCDInlequations. Here the system will be described as, A? =.§i’ where orrly'the simultaneous equations ( 2 through 10) are included. Talale 5.24 presents A?¥§_for population at the 1975 level. 104 Aavuau hacsfia Aqv>~a Amemmomu 141macmu Amemacac Anvmdomu afl1ma0xu aayments. The historical data for the 1960 to 1974 period were the IJSDA estimates (Farm Income Statistics, Table 5D) multiplied by the trumber of farms. Data for the 1950 to 1959 period were estimated by tassing an equation relating off-farm income to personal income of the féarhnpopulation from nonfarm sources (Farm Income Statistics, Table 511). The equation was estimated via OLS and is shown here: 122 (5.41) OFFY = -254.8324 + 1.2452*FRMPOPY, R2 = .9964 (60.15) OFFY = total off-farm income. FRMPOPY = personal income of the farm population from nonfarm sources. The source of off-farm income in the model is an equation pro- jecting a time trend for deflated per farm off-farm income. The CPI is used as the deflator. This equation is as follows: (5.42) ROFFY/FARMS = 731.0804 + 207.8911*T, R2 . (19.67) .9462 ROFFY = real off-farm income. FARMS = namber of farms, T = time (year - 1949). Demand for Loan Funds The model projects the demand for loan funds (net of repayments) as a residual source of funds. A critical weakness of this approach is that the sum of errors in other components, as well as errors of omission, directly affectstfluaestimated net flow of loan funds. This approach to estimating debt financing requirements is the basic reason for emphasizing cash flows in the Sources and Uses of Funds Statement. If the purpose of the SAUF statement is to estimate the net flow of loan funds, then it is not necessary to include flows that do not require financing or are self-financing, The Sources and Uses of Funds Statement does not indicate that the net flow of loan funds is derived from real estate or nonreal (estate lenders. However, the Balance Sheet does include a breakdown 123 of debt by real estate and nonreal estate. The breakdown used in the model is made 0“.EQ real information. The procedure in the model is to maintain a constant ratio of real estate to nonreal estate debt,beginning with January 1, 1975. Handling of Intersector Flows Generally, intersector flows are not measured, and thus the historical data are based on assumption or, in the case of real es- tate purchases, from discontinuing proprietors,based on a one-time survey. Estimating equations from such data has been avoided to the extent possible. In the following paragraphs, the assumptions made and method of handling each item with respect to intersector flows are des- cribed. Real estate is the most important individual item. Purchases from discontinuing proprietors are estimated as the value of voluntary real estate transfers less debt owed on the transfers (11.1%) times the proportion of sellers who do not remain active in farming (l-.O95). The information on percentage of debt (11.1%) and proportion of sellers who remain active in farming (.095) is based on a one-time survey. It should be noted that this is the only place where debt of discontinuing proprietors is taken into account . Implicitly, it is assumed that discontinuing proprietors have no 'Other debt. One might note the implications of handling debt of discontin— uing proprietors in this manner. The net flow of loan funds given 124 as a source of funds understates the net flow of continuing propri- etors by the amount of debt attributed to discontinuing proprietors. One might include the entire value of real estate purchases from discontinuing proprietors as a use of funds and add debt of discon- tinuing proprietors to the net flow of loan funds. The implied debt of discontinuing proprietors in 1975 was approximately 1.2 billion dollars. The net flow of real estate loan funds was 4.6 billion in 1975. Thus, the net increase in real estate debt of continuing proprietors is 26 percent greater than the change for the sector. The above explanation is included largely to clarify the mean- ing of the accounts. The accounts herein are largely sector accounts, not accounts for continuing proprietors. As one can see by the impli- cations of the above, these might differ in presentation of data. The other flow related to real estate is net purchases of real estate from the farming sector. This flow is estimated by assuming that the change in acres in the farming sector (SRS land in farms data, see Agricultural Statistics) is sold to nonfarm sectors at the average value per acre for the year. Discontinuing proprietors' inventories of crops and livestock, as well as their holdings of machinery and motor vehicles,are not taken into account. Financial assets and household equipment and furnishings owned by discontinuing proprietors are assumed to be equal to the per farm stock at the beginning of the year. It is assumed that these assets leave the sector with discontinuing proprietors,and this withdrawal. is estimated by multiplying the per farm stock by the change in number 125 of farms. This assumption implies that investment by continuing pro- prietors in financial assets is the change in nominal value plus the amount removed from the sector by discontinuing proprietors. Purchases of household equipment and furnishings have been ad- justed to reflect assets removed by discontinuing proprietors (see Baker, 1973). Adjustments in cash receipts, off-farm income, personal taxes, and so forth are not made for discontinuing proprietors. The adjustments are not made because a prOprietorirlthe sector for part of the year is counted as being in the sector for the entire year. It may be more accurate with this view to make adjustments described above based on year end stock of assets, or the average of beginning and ending. The view taken here is that the adjustments being made are based on little real knowledge and should not be given the legitimization implied by more complicated calculations. All of the intersector flows are token adjustments, not based on real observations. Any correspondence to the actual flows is a result either of making realis- tic assumptions or luck (more probably the latter). Calculating Stocks of Durable Assets and Depreciation There are four categories of depreciable durable assets in the tnodel. These are: machinery and motor vehicles, service buildings -and other structures, farm operator dwellings, and household equip- tnent and furnishings. The following will describe the methods used t1) calculate stock levels and depreciation for each category. 126 Machinery and Motor Vehicles. There are two basic stock vari- ables for machinery and motor vehicles in the model. One is based on a value concept and is used to calculate the Balance Sheet value of machinery and motor vehicles and depreciation for use on the Income Statement. The other stock variable is based on a produc- tive capacity concept and is used in the investment demand, crop supply, and machinery repair equations. The productive capacity concept and construction of the variables are explained in greater detail in Baker (1978). With respect to the Balance Sheet value of machinery and motor vehicles (and related depreciation), the model is not totally consis- tent with the published USDA series. The inconsistencies arise in several areas. The USDA data are based on calculations using investment series for individual components (e.g. tractors, trucks, and autos). The model has investment as the sum of the components and therefore is not broken down by item. Thus, only one rate of depreciation is used for the aggregate. In addition, the lack of component (iata for investment in machinery and motor vehicles requires making :adjuStments for household use of automobiles and trucks as a constant (Martion of total machinery and motor vehicles rather than of the com- 1/ ponents o" 1 -/.According to USDA, Agricultural Handbook 365, Volume 3, p. 10, the farm business portion is 40 percent of automobile and 78 percent of truck expenditures. 127 The stock data in the model-aremaintained internally on a con- stant dollar basis. Depreciation and the stock are converted to a current dollar basis via multiplication by a price index. This is similar to the USDA procedure. A difference arises with respect to the constant dollar Balance Sheet value. To initialize the model for simulation runs, the current dollar value of the machinery stock is deflated by a price index for machinery and motor vehicles based on prices paid by farmers indices for machinery and motor vehicles. This gives a different constant dollar value than that published in the Balance Sheet of the Farming Sector. The depreciation rate is applied in a declining balance manner to the beginning—of—year stock. The mean value of the ratio of motor vehicle depreciation plus other machinery and equipment depre- ciation to the beginning of the year stock for the 1970 to 1975 period (.147) is used in the model as the depreciation rate. Service Buildings and Other Structures. In a manner similar to that of machinery and motor vehicles, there are two stock and depreciation series for service buildings and other structures. The stock based on a productive capacity concept is explained in Baker (1978). The rather stock, used on the Balance Sheet and to calculate depreciation. is calculated using a declining balance rate of 7.2 percent applied tc> the constant dollar stock. The constant dollar data are converted Ix) Current dollars via multiplication by a price index. The constant dollar stock is also adjusted by accidental damage deflated by the price index for buildings, A later section on real estate will give more detail. 128 Farm Operator Dwellings. Only one measure of depreciation of- farm operator dwellings is included in the model. This is a "value" concept used with a declining balance rate of 4.5 percent. As in the case of the other series, the basic internal data are in con- stant dollars with conversion to current dollars using a price in- dex for dwellings. Depreciation and stocks calculated in this man- ner are included in the Income Statement and Balance Sheet respec- tively. A later section on real estate will give more detail. Household Equipment and Furnishings. A stock measure for house- hold furnishings and equipment is required for the Balance Sheet. In addition, an estimate of replacement investment is required for use in conjunction with historical data on the stock of household equipment and furnishings to estimate expenditures. Baker (1978)givestim:explanation of how the expenditure series for household equipment and furnishings is constructed. A corres- ponding method is used in the model to calculate purchases for the Sources and Uses of Funds Statement and depreciation for use in cal— culating net capital formation. The ending current dollar stock of household equipment and furn- ishings is projected using a time trend for per farm household «equipment and furnishings. This equation is the following: (5.43) PFHEF = 3330.65 - 213.829*T + 10.5298*T2, R2 = .96 (-5.85) (9.05) PIfiHEF = per farm stock of household equipment and furnishings. T = time (year - 1949)- 129 The end of year total value of household equipment and furnish- ings is PFHEF multiplied. by the number of farms. The total end- ing value of household equipment and furnishings, deflated by the price index for household equipment and furnishings, is used in con- junction with depreciation (replacement expenditure) and discontinuing proprietor withdrawals to calculate implicit constant dollar pur- chases. Conversion to current dollars is made via multiplication by the price index. Investment Credit The handling of investment credit in the model is not ideal, but the ideal was not possible. Over the historical data period, investment credit was observed at 0 and 7 percent. Sometimes the credit was available for only a portion of the year Over the - period to be simulated, the investment credit will be at levels other than ()cnr7 percent. Investment credit cannot be treated as a continuous variable in regression equations because of the lack of observation at various levels. Treatment as a binary (0, 1) variable will not accommodate projection under alternative invest- ment credit levels. With the preferred methods lacking feasibility, the following ‘umthod of handling investment credit was chosen. It was assumed that :farmers view investment credit as a reduction in the price of the aisset being purchased. The prices of machinery and motor vehicles 811d buildings were reduced in years when inveStment credit applied by' the following method. The investment credit (e.g., 7 percent 130 of the price) is received in the form of a tax reduction on the cur- rent year's income. The reduction is effectively received in the following year. Thus, the prices in the current year were reduced by the present value of the tax benefits. The discount rate used was the bank rate of interest and the period discounted was one half of a year. Machinery and building prices in years for which the investment credit was available for only part of the year were reduced by a fractional amount of the discounted tax benefits. The model determines personal taxes via a rather crude method which does not allow for linkages between investment credit and re- duced taxes. The level of credit and fraction applied to the historical data is shown in the following table. Table 5.29. Investment Credit Data Year Level Fraction 1962 .07 l 1963 .07 l 1964 .07 1 1965 .07 l 1966 .07 0.5 1967 .07 0.5 1968 0.0 O 1969 .07 0.25 1970‘ 0.0 0 1971 .07 0.5 1972 .07 1 1973 .07 1 1974 .07 l .131 Price and Quantity of Total Output The prices and quantities of total output are indexes construc- ted in the following manner. Price of total output is the sum of crop and livestock prices using quantities utilized in the crop year and calendar year, respec- tively, as weights. P = (P1(Q1)/(Q1 + 02)) + (P2(Q2)/(Q1 + Q2))/B1967 where: P = price index for total output. "U ll 'crop price index. 1 Q1 = quantity of crops utilized = feed + seed + exports - imports + food and industrial use. P2 = livestock price index. Q2 = livestock use = consumption + exports — imports. B1967 = value of P in 1967, converts to 1967 = 100 index. The quantity of output index is simply livestock and crop out- put occurring in the calendar year summed and divided by the base year. Q = (Q1 + Q2)/Q1967 where: Q 8 index of total output- (21 - Crop output (e.g. crop output for crOp year 1951-52 is entered into the 1951 output index). (22 = livestock output. Q1967 = 01 + 02 in 1967. 132 Property Taxes Property taxes (nominal dollars) are projected in the model as a function of time and the beginning of year real estate value. The equation used is given as follows: (5.44) PROPTAX = 413.1961 + 3.5363(REAL) + 62.5678(TIME), R2 = .9815 (3.67) (7.59) PROPTAX_= nominal property tax. REAL = aggregate farm real estate value (nominal dollars). TIME = time (year - 1949). Disposable Income Aggregate disposable income is calculated from gross national product (GNP) in the model. The hypothesis that time would be a fac- tor; because of rising income and the increasing marginal tax rates, was rejected. Deflated per capita disposable income is used in the demand for livestock equation. It was reasoned that disposable income would be a more accurate measure of a consumer's budget constraint than using GNP. The equation used in the model is as follows: (5.45) D1 = -2.5384 + .6987 (GNP), R2 = .9992 (162.9) ‘D1 = nominal disposable income. GNP = nominal gross national product. Real Estate Value The model divides real estate into three components: value of flarhioperator dwellings, value of service buildings and other structures, 133 and value of land and improvements totflualand. 'The inclusion of the breakdown of real estate value into its component parts for the Balance Sheet in the model is not consistent with the Balance Sheet published in The Balance Sheet of the Farming Seetor. It is included in the model to maintain consistency between the Balance Sheet and the rest of the model. Also, the values printed in the Balance Sheet are required internally. Thus, the author felt that While the data may not be highly reliable, the implied values for these items should not be hidden in the model but printed for inspection. Farm Operator Dwellings The model contains an internal constant dollar stock of farm operator dwellings (SDCON). Depreciation in constant dollars is 4.5 percent of the sum of the beginning constant dollar stock and one half of the current year's purchases of dwellings in constant dollars. The constant dollar dwelling purchases are the current dol- lar purchases less accidental damage to dwelling (DMAGD) deflated by PDWL. The ending constant dollar value isbeginning value plus purchases less depreciation. The ending constant dollar value is inflated by PDWL to use in the Balance Sheet. Historical Data Series. The model is initialized for constant sand current dollar values of dwellings by using the value implied in tzhe USDA depreciation series. The depreciation of dwellings, as (zzilculated by USDA, is 4.5 percent of a constant dollar stock times 811 inflator. Thus,the implied current dollar stocks can be derived' 134 by dividing published depreciation data by .045. Division by the price index for dwellings yields the constant dollar value. To get the beginning of year implied values, one must subtract one half of the current year's expenditures. This procedure was applied to the data for the period 1964 to 1974. The procedure is not error free as, during some years in which the implied constant dollar stock increased,there was implied net disin- vestment (purchases minus depreciation). The current dollar stock of dwellings data constructed were used in deriving the historical series for land. Capital Gains. Nominal capital gains for farm operator dwell- ings are calculated as ending Balance Sheet nominal value minus beginning Balance Sheet nominal value minus purchases. plus capital. consumption. Service Buildings and Other Structures As with dwellings, the model contains a constant dollar stock of service buildings and other structures (SBCON). Depreciation is 7.2 percent of the sum of beginning value and one half of the current year's constant dollar purchases. The current dollar purchases are the estimated expenditures (in constant dollars) for service build- ings, other structures, and land improvements less 21.4 percent of the same less deflated accidental damage. The 21.4 percent is the -average ratio of expenditures for land improvements to expenditures for service buildings, other structures, and land improvements for the 135 1967 to 1975 data period.l/ The ending constant dollar value is the beginning value plus purchases minus depreciation. All current values are constant dollar values inflated by PBLD. Historical Data Series. The historical series is constructed in a manner similar to that for dwellings. Depreciationgl is divided by .072 to derive the implied current value stock. One half of cur- rent purchases (service buildings and other structures), less acci- dental damage to service buildings (DMAGB), is subtracted from the implied stock to get the beginning of year stock in current dollars. The historical series is used in deriving the value of land data. Capital Gains. Nominal capital gains on service buildings and other structures.are calculated from current dollar values as follows: ending Balance Sheet value less beginning Balance Sheet value minus purchases during the year plus capital consumption. Value of Land and Improvements The value of land in the model is given as a function of real estate price times the acres of land in farms. This near identity equation is given as follows: (5.46) LAND = 5021.9 + 1.32347*PREAL*LDFRMS, R2 = .9982 (29.87) ‘LAND = end-of-year land value (beginning of following year value). IPREAL = index of real estate value for March lst of the following year. -l! From unpublished USDA data. 2/ -— Farm Income Statistics, Table 18H. 136 LDFRMS = land in farms, current year Historical Data Series. The historical data used to estimate Equation (5.46) was constructed by subtracting from total real es- tate value the value of service buildings and other structures (BLDVLU) and the value of dwelling (DWLVLU). Derivations of BLDVLU and DWLVLU havebeen explained above. Capital Gains. Nominal capital gains of land are calculated as follows: ending Balance Sheet value less beginning Balance Sheet Value minus land improvements during the year-Ll plus value of real estate sold to nonfarm sectors, 8(4). Accidental Damage to Farm Buildings A component of capital'consumption is accidental damage to farm buildings (DAMG). The model contains an equation giving DAMG as a time projection. This equation is shown below: (5.47) DMAG = 145.274 + 4.31609*T, R2 = .4414 (4.17) The equation was estimated from data covering 1951 to 1974. Al- ternative specifications including the value of farm buildings (cover— ing the data period 1965 to 1974) did not improve the equation. Accidental damage is subtracted from 1) the stocks of farm operator dwellings and 2) service buildings and other structures. DMAG is divided between the two categories according to their proportions 1/ -— Estimated in the model as 21.4 percent of expenditures for service buildings, other structures, and land improvements the average for the 1965 to 1976 data period. 137 of capital expenditures. Thus the following equations are used: DMACB = .786*CE(2)*DMAG/.786*CE(2) + CE(3) _DMAGD = CE(3)*DMAG/(.786*CE(2) + CE(3)) DMACB -accidental damage to service buildings and other structures. DMAGD accidental damage to farm operator dwellings.' DMAG = total accidental damage. CE(2) = capital expenditure for service buildings, other structures, and land improvements. CE(3) = capital expenditures for farm operator dwellings, .786 = 1965-1975 average proportion of service building and other structures expenditures to expenditures for the same plus land improvements. Other Uses of Funds This category is the residual use of funds on the SAUF statement. It is the summation of many items. A few of these are (a + indicates the item adds to the use a - indicates it is actually a source of funds): net cash gifts and inheritances from farming sector participants to nonfarming sectors (+); purchases of machinery and motor vehicles from discontinuing proprietors (+); net investment in off—farm capital, financial assets such as stock and bonds (+); equity flows from nonfarm sector participants such as limited partnerships and farm corporations (—); net error in measurement of all other sources and uses (+ or -); and all other unmeasured cash flows (+ or -). Table 5.30 shows the data for other uses of funds as defined in the model. It is hypothesized that a large portion of this residual use of funds is in the category including investment in nonfarm capital. 138 Table 5.30 Historical Data for Other Uses of Funds and Alternative Income Measures Net Farm Off—Farm Disposable Year Other Uses Income Income Farm Incomea Billion Dollars 1961 5.0 13.3 9.3 21.1 1962 5.5 13.5 10.3 22.9 1963 5.5 13.4 10.9 22.6 1964 6.1 12.1 11.7 22.1 1965 8.0 14.8 12.7 25.8 1966 9.8 16.0 13.9 28.0 1967 6.2 14.2 14.5 26.7 1968 7.1 14.3 15.5 27.6 1969 10.1 16.3 16.6 30.1 1970 9.7 16.2 17.4 30.6 1971 8.2 16.9 18.7 32.4 1972 12.1 22.2 20.4 30.1 1973 23.1 39.0 23.5 58.1 1974 19.0 31.6 26.1 52.6 1975 . 12.3 129.1 28.5 53.3 1976 15.9 24.2 31.1 50.6 aNet farm income plus off-farm income minus personal taxes. 139 Penson (1977) has estimated a similar category for 1970 to 1975 referred to as: net additions to equity in life insurance reserves, individual retirement accounts, stocks and bonds, and other nonfarm capital. While the details of the procedures used by Penson to construct the data differ in some respects from those used in this Study, the data act as one basis for comparison. The data for other uses of funds shown in Table 5.30 peak in 1973 at 23.1 billion. Penson's data also peak in 1973 (at 19.6 billion). However, for a direct comparison,the data used in this study need to be adjusted for net rent to nonoperator landlords (subtract 5.7 billion). Making this adjustment (23.1 — 5.7 = 17.4) leaves the other uses of funds calculated here at a level less than Penson's figure. Penson has used not additions to household furnishings ignoring replacement purchases. Additionally, Penson adjusted for withdrawals of current income by discontinuing proprietors, internal sales of breeding livestock, capital purchases by nonoperator landlords, and debt acquired by nonOperator landlords. Consistent adjustment for household furnish- ings and equipment replacement purchases and internal sales of breed? ing livestock would bring the data in this study and Penson's closer together. The other two items would increase the divergence. Penson's study does not adjust for the items in the residual such as gifts, inheritances, or purchases of machinery and motor vehicles from discontinuing proprietors. While the data calculated here correspond reasonably well with Penson's,the latter may be in error when attributing all of the residual to investment in off-farm financial assets 0 140 If it is reasonable to assume that a substantial portion of the other uses of funds category is composed of off-farm financial assets (an assumption that has not received empirical testing),then one would expect income of farmers to be highly correlated with this use of funds. Table 5.30 also shows data for net farm income, off-farm income, and diaposable farm income (net farm income plus off—farm income minus taxes). It is interesting to note that the turning points in the residual use of funds follow net farm income much more closely than the other income measures. One can see data that 1975 will have a large error because of the small decline in net farm income (the other two measures increased) and the large drdp in other users of funds. Regression equations were run using the three income measures individually and net farm income and off-farm income in combina- tion. The data covered the 16 years from 1961 through 1976. Addition- ally, an equation using net farm and off-farm income was estimated omitting 1975. These equations are shown in the following table. The equation chosen for use in the model is number 5.47e in Table 5.31. The choice was based partially on the a priori belief that both farm related income and off-farm income are factors affecting the elements making up the dependent variable. Equations 5.47s through 5.47d have very large errors for 1975 (relative to errors in other years) as was expected. In addition. equations 5.47a and 5.47d, the equations including net farm income underestimate for 1976. (Equation 5.47e, estimated with the 1975 observation deleted, fits 1976 very closely (15.9 versus 15.83) 141 Table 5.31. Equations for Other Uses of Funds U(6) Disposable Net Farm Off-farm Farm 2 Durbin Equation Constant Income Income 'Income R Watson 5.47a -1861.22 .6297 .91 2.32 (11.61) 5.47b -681.92 .6202 .64 1.54 (5.0) 5.47c -2434.83 .3789 .84 1.77 (8.46) 5.47d -217S.56 .5801 .0722 .91 2.36 (6.19) (.66) 5.47e -3464.63 .5712 .1766 .97 1.74 (10.85) (2.73) and fits other recent years as well as the other equations. How- ever, the higher R2 should be largely attributed to the deletion of 1975, not to the fact that it fits the other observations better. It is the author's opinion that these results are very encour- aging in the respect that an apriori troublesome area,.. the-need to Specify an equation for a conglomerate residual use of funds, has been handled with a reasonably specified equation that has a good fit. Summary This chapter has presented the equations and relationships Imzquired for the endogenous portion of the model. These include structural equations, time trends, near identities, and accounting 142 relationships. Statistical results have been presented where appropriate for the individual equations. The following chapter (VI) will investigate the properties of structural equations as a SEC. CHAPTER VI Model Evaluation Introduction Evalutaion and validation of the model consisted largely of 1) equation by equation statistical and economic criteria and 2) ability of the model to track over the historical period. The statistical and economic criteria for each equation have been presented in Chapter V. While the criteria used in estimating and selecting the individual equations are considered essential, they do not guarantee that the model will fit all variables well. While this is true of all multiple equation models, it is especially true when some equations are simultaneous. For example, a simultaneous equations model that has very price inelastic demand and supply equa- tions might track quantities well but have wide fluctuations in prices. A recursive equation or block recursive set of equations with very influential explanatory variables being current endogenous variables could have a high R2 with actual values of the endogenous variables but track poorly with predicted values. This could also be true of a single equation model with lagged endogenous variables. The model has been developed primarily to project (under alterna- tive conditions) the Income, Balance Sheet, and SAUF statements for 143 144 the farming sector. The data on these statements are largely trans- formations of other variables. Thus, errors in tracking variables on financial statements would be directly traceable to errors in track- ing the underlying variables. However, errors in tracking underlying variables might cancel each other. In the author's opinion, the validity of a model rests largely upon its ability to track endoge- nous variables. A model that tracks variables well because of off- setting-errorssfiunflxlnot be accepted without reservation. The indi- vidual equations provide the mechanism for evaluating policies (levels of controllable exogenous variables) and scenario projections of uncontrollable exogenous variables. If the individual equations (do not behave well, the conclusions with respect to the impacts of alternative scenarios will be less reliable. It is possible for a model to track individual variables rea- sonably well and not track a transformation of these variables. This would not seem to be a highly likely occurrence nor as invalidating as failure to track underlying variables. 'The model here has a very large number of variables (counting all transformations). Clearly, presenting data relative to the tracking of all variables would be counter productive._ Thus, a decision as to which variables are most important had to be made. The set chosen includes the variables in the simultaneous equation set and the input quantities resulting from the recursive demand for variable inputs equations. These variables are a large portion of endogenous prices and quantities which are transformed into cash flows and other finan- cial variables. An additional factor considered was the amount of 145 computer programming required to include additional variables in the evaluation. Extension of the set of variables evaluated substan- tially beyond those included would involve significant increments of programming. The following pages describe the evaluation procedures and results. The first Section describes adjustments made to the model. The following section presents tabular and graphical results of simulating over the historical data period. Adjustments to the Model Initial efforts at running the model over the historical data period indicated relatively poor performance by the system of simul- taneous equations. It was determined that the poor performance re- sulted from ygry_low price elasticity of demand for crops in the system of equations (Baker, 1978). Two modifications were made to the model and are presented below. Endogenizing Crop Inventories The model was conceptualized from a traditional static theoret- ical framework. This factor, combined with an intuitive feeling that an inventory demand equation would be difficult to specify theoreti- cally as well as empirically, led to treating inventories exogenous- ly. Further impetus was provided by complications involved with gov- ernment-hehd crop inventories and in simulating policies related to inventories. However, it becomes obvious when evaluating the model's Performance and looking at the data that a portion of the adjustments 146 to shocks in the Crop sector (e.g., high or low production, increased exports, etc.) is absorbed via changes in inventories. Thus, the problem of projecting a consistent set of exogenous variable values (e.g., export levels and inventory levels) would arise. To aid in this problem, an inventory demand equation has been estimated to use at least for base projections. The equation is estimated as a func- tion of the change in the real price of crops and beginning Crop inventories. These is herein no theoretical derivation of this formu- lation of inventory demand, although the formualtion is intuitively appealing. Empirical estimates for equation (6.1) shown below are presented in Table 6.1. (6.1) CROPS(8) = f(CHCRP, CROBS(1)) CROPS(8) = ending inventory 6f crops. CROPS(l) = beginning inventory of crops. CHCRP = RPCROP - RPCRMl. RPCROP = real price of crops. RPCRMI = real price of crops lagged one period. Table 6.1 Empirical results for Equation (6.1), Demand for Crop Inventory, CROP(8) Selected L Regression t Elasticities xplanatory Variable : Coefficient : Statistic : 1951-74 : 1972-74 Constant 1318.34 --- --- --- CHCRP . —40.4695 -2.37 -.42 -.86 CROPS(I) .86665 5.52 --— --— 2 ' _ _ _ = .8553, D.W. - 2.3145, t.Ol — 3.055, t.05 - 2.179, t.10 = 1.782 147 The equation for inventory demand was estimated using the 1960 to 1974 data period. Much of the variation in the observations over this period occurs in the later portion. It is hoped that this vari- ance in the observations will give the equation some predictive power for handling projections. The entire data period was not included in the estimation of the equation. This decision was made as an attempt to avoid including more observations under a different "structure" than necessary. The structural change referred to is the change from a situation of stable prices with the market absorbing all it would at those prices (with the balance going into inventories) to a situation of inventories, prices, and other quantities demanded adjusting simul- taneously. Part of the basis for choosing the inventory demand equation was the own price elasticity. This was necessary for the model to track well.and is a result of the crop price inelasticity of the other crop demand equations in the model. The short-run price elasticity of demand for the equation is -.86 at recent price and inventory levels. The long run price elasticity is zero. Revised Feed Demand Equation In addition to endogenizing inventories, examination of the mode].'s tracking performance led to re-estimation of the feed demand equatilon. Upon re-examination, two changes were made in equation (5.3). 1‘18 own price elasticity of feed demand (with respect to feed prjfui) and the price elasticity of feed demand with respect to crop Price (derived from the partial reduced form equation, see Tables 148 5.3 and 5.5) were in the range of -.2 at current prices and quanti- ties. This elasticity is misleading, as the price elasticity of de- mand for crops of the system is -.0288 at recent price and quantity levels (Baker, 1978). This price elasticity is lower than the own price elasticity for any of the individual components. The reason for the apparent cOntradiction is the simultaneous relationship with the livestock sector. That is, the elasticity in the crop demand component is "wiped out" by the livestock sector. Basically, the relationship is that increases in feed price cause increased livestock price which then increases feed demand and thus offsets the crop sector adjust- ment. This description is technically incorrect as there is no sequential adjustment as described; it is simultaneous. Thus, it was felt that the price elasticity of feed demand was low relative to the elasticity with respect to the price of livestock. The second problem with the earlier specified demand for feed equation is the trend variable. While feed demand includes an in- creasing trend over time that cannot be explained with prices, it can be explained by the overall increases in the level of livestock production as reflected in livestock inventories. Livestock begin- ning inventory is included as a replacement for the time trend in re-specification of the equation. The equation now used in the model is shown below as equation (6u2). Empirical results follow in Table 6.2. (6.2) CROPS(6) = f(RPFEED, RPLIV, RPLVMl, LIV(1)) - + + + CROPS(6) RPFEED RPLIV RPLVMl LIVl 149 feed use of crops. real price of feed. real price of livestock. real price of livestock lagged one year. beginning livestock inventories. CROPS(6) Table 6.2 Empirical Results for Equation (6.2), Feed Demand, Explanatory Variable Selected Regression t Elasticities : Coefficient : Statistic : 1951-74 : 1972-74 Constant RPFEED RPLIV RPLVMl LIV(l) -6603.67~ -—— --- --- -47.9611 46.29 -.565 -.475 34.7116 5.01 --- --- 10.8519 .1.29 --- --- .859385 15.72 -—— --- 2 t.10 = 1.729. R = .9587, D.W. = 1.53, c = 2.861, c .01 Elasticities in equation (6.2) differ from the earlier feed de- mand equation (5.3) in two major respects. First the equation has a higher own price elasticity and second, the own price elasticity has increased relative to the current livestock price elasticity. Model Performance The model was evaluated using a computer program written by Rodney Kite (USDA, ERS, CED, Forecast Support Group). The solution algorithm uses the method of Gauss-Siedel. While the model herein 150 can be solved directly in a given year by matrix inversion, the pro- gram (called CASSP), written by Kite, includes nmmerous convenience features. These include: generation of evaluative statistics, ease of exogenizing of "turning off" selected equations, and easily changing from using actual values for lagged endogenous variables (one Period Forecasts) to using predicted values for lagged endoge- nous variables. After initial difficulty in achieving convergence in the solu- tion of the system of equations, the model solved quickly with a stringent convergence criterion. The early problem was solved by changing the order of the equations. (The problem was one of finding the solution to the equations rather than there being more than one solution or converging to unreasonable solutions). Ortega (1972) discusses the solution properties of the Gauss-Siedel solution algorithm further. The convergence criterion was .0001. When the estimated values for all endogenous variables changed by less than .0001 between itera- tions, the algorithm stopped. The criterion seemed quite strict in light of the fact that the value of the smallest variable was in the range of 100 and that several of the variables had values over 10,000. The solution algorithm converged to within this criterion in about 20 iterations. Evaluative Data The evaluative data are presented in Tables 6.3 through 6.17 and in the accompanying Figures 6.1 through 6.15. Data are presented for each year and summary statistics are presented. The model is 151 evaluated over the 1952 through 1974 period (the first year of data prepared for GASSP was lost due to lagged variables). The evaluation procedure required the model to use its own forecasts as values for the lagged endogenous variables. That is, the model was forced to "feed" on itself. This method of evaluation is appropriate for a model to be used for multi-period projections. A model to be used extensively for single period projections would be evaluated using actual values for lagged endogenous variables. The data generated for each year are: 1) actual and estimated values, 2) the error (estimated minus actual, thus positive values indicate overestimation and negative values underestimation), and 3) the error as a percentage of the actual value. The summary data generated are: l) the mean value of actual data, 2) the mean abso- lute value of the actual data, 3) the mean value of the estimates, 4) the mean absolute value of the estimates, 5) the mean error, 6) the mean absolute error, 7) the mean percentage error, 8) the mean absolute percentage error, 9) standard deviations of the actual and estimated data, 10) the mean and mean absolute percentage change in values from year to year, 11) the square of the mean error, and 12) the root mean squared error of the forecast. For the purpose of evaluating the simulation model, the follow- ing summary data are considered the most informative among the group listed above: 1) mean percentage error, 2) mean absolute percentage error, and 3) root mean squared error of the forecast. The mean per- centage error should be near zero. A large mean percentage error (positive or negative) indicates that bias could be a problem. The 152 mean absolute percentage error utilizes the absolute values of the errors and thus positive and negative values do not cancel out. Thus, the mean absolute percentage error is a reasonable measure of fit. An alternative measure of fit is the root mean squared error. This latter measure has the feature of weighting large errors more heavily than small errors. Crop Quantities The evaluative data for crop quantities are presented in Tables 6.3 through 6.7 and Figures 6.1 through 6.5. With the exception of crop inventories, the model tracks crop quantities quite well. The mean percentage errors range from .149 to .0311 percent. The mean .absolute percentage errors range from 1.73 to 3.40 percent. These would indicate a good fit and little problem with bias. The crop inventory equation, Table 6.7 and Figure 6.5, under- estimates the levels of inventory for most years. Crop inventory is a stock variable. This factor must be taken into account when evalua- ting the equation. If in a historical simulation tflua value of a stock variable is underestimated for a year, the year to year change (the flow) could be properly estimated,1nn:an error would still occur in the level of the stock. Over a large portion of the historical period for which the equation underestimates the level of crop inven- tories, the structure was one of price supports with excess supply going into government supported storage. This is not the structure which the equation is intended to represent. (The crop inventory equation was estimated using only 1960 through 1974 data.) Table 6.3. Year Actual Estimate Error 1952 15941.074 15857.299 -83.8 1953 16692.524 16876.944 184. 1954 16563.681 16690.117 126. 1955 16376.271 16000.868 -375. 1956 16967.540‘ 16465.110 -502. 1957 17058.997 16931.459 -128. 1953 16870.684 16965.224 94.5 1959 18808.225 18616.766 -191. 1960 18347.779 17762.721 -585. 1961 19819.048 18761.284 -1057. 1962 19466.150 18998.708 -467. 1963 19883.691 19723.482 -160. 1954 20621. 965 20805. 951 184. 1965 20102.505 20730.031 628. 1966 21309.446 22037.889 728. 1967 20815.419 20779.250 -36.2 1963 22061.919 22102.307 40.4 1969 22545.329 22838.406 293. 1970 22595.945 22413.647 -182. 1971 21922.509 22492.752 570. 1972 24337.877 25510.602 1173. 1973 24548.632 24369.837 -179. 1974 25786.700 25786.473 -.227 Mean .200*10§ .200*105 3.18 Absolute .200*10 .200810 347. PERFORMANCE STATISTICS 1952—1974 MEAN MEAN MEAN STD DEV z ABSOLUTE --------- _ _ - - EHéN§E_ %_CBA§GE ACTUAL .200*10§ .287*10€I 2.34 3.65 ESTIMATE_ LZOOflO _.307410 2.35 4.37 SQUARED MEAN ERROR 10.134 ROOT SQUARED ERROR OF FORECAST 486-774 153 Evaluative Data for Crop Production, CROPS(Z) ZError -.526 1.10 .763 -2.29 -2.96 -.748 .560 -l.02 -3.19 —5.34 -2.40 -.806 .892 3.12 3.42 -.174 .183 1.30 -.807 2.60 4.82 -.728 -.00095 -.0967 1.73 MEAN Z EREOB MEAN ABSOLUTE Z ERROR -.0967 1.73 154 £102 200 J CROPS(Z) 180 J 160 l 40 "1950 1855 1560 1965 1870 1875 TIME IN YERRS - Figure 6.1 Crop Production fl Actual A Estimated 155 Table 6.4. Evaluative Data for Quantity of Feed Use, CROPS(6) Year Actual Estimate Error ZError 1952 7328.040 6987.228 -341. -4.65 1953 7095.544 7415.186 320. 4.50 1954 7041.321 7870.730 829. 11.8 1955 7046.355 7766.410 720. 10.2 ' 1956 7509.104 7548.519 39.4 .525 1957 7501.713 7175.675 -326. -4.35 1953 7739.019 7427.856 -311. -4.02 1959 8391.235 8328.251 -63.0 -.751 1960 8726.246 8338.395 -388. -4.44 1961 8725.898 8445.588 -280. -3.21 1962 9008.591 8619.900 —389. -4.31 1963 8920.802 9055.986 135. 1.52 1964 8919.331 9269.850 351. 3.93 1955 8921.038 9046.030 125. 1.40 1966 9526.674 9388.390 -138. —1.45 1957 9699.773 9137.149 ~563. -5.80 1953 9655.009 9557.997 -97.0 -1.00 1969 10125.317 10315 874 191. 1.88 1970 10734.392 10567.176 -167. -1.56 1971 10580.202 10548.210 -32.0 -.302 1972 11254.162 11666.560 412. 3.66 1973 11390.783 11282.641 -108. -.949 1974 11333 587 11118.288 -215. -l.9O Mean .901*104 .899*10A -12.9 .0311 Absolute .9O1*104 .899*104 284. 3.40 PERFORMANCE STATISTICS 1952—1974 MEAN MEAN MEAN MEAN MEAN STD DEV z ABSOLUTE % ABSOLUTE — ------------- EHANQE- %-C§A§G§ ERROR z ERBOB _ _ ACTUAL .901*104 .143*102 2.07 2.72 ESII§A$E_ ,829519 _.139*10 2.23 3.97 .0311 3.40 SQUARED MEAN ERROR 165.902 ROOT SQUARED ERROR OF FORECAST 365-690 156 '770 .590 950 1955 1880 1885 1870 1875 TIME IN YEHRS Figure 6.2 Feed Use of Crops [] Actual A Estimated Table 6.5. Evaluative Data for Quantity of 157 Seed Use, CROPS(5) Year Actual Estimate Error ZError 1952 358. 371. 12.6 3.52 1953 372. 367. -4.62 -1.24 1954 359. 352. -6.41 -1.79 1955 351. 352. .300 .0853 1955 353. 342. -10.3 -2.93 1957 332. 339. 7.59 2.29 1953 340. 335. -5.68 -1.67 1959 333. 340. 6.49 1.95 1960 328. 341. 13.5 4.13 1961 340. 332. —7.64 -2.25 1952 322. 330. 7.43 2.31 1963 330. 333. 3.19 .968 1954 »332. 336. 3.59 1.08 1965 341. 341. -.432 -.127 1965 346. 344. -1.42 -.410 1957 378. 360. -18.3 -4.85 1953 376. 361. -15.1 74.02 1969 356. 359. 2:11 .593 1970 352. 361. 9.17 2.61 1971 365. 371. 5.18 1.42 1972 362. 366. 4.25 1.17 1973 395. 393. -1.29 -.327 1974 422. 426. 3.93 .931 Mean 354. 354. .348 .149 Absolute 354. 354. 6.55 1.85 PERFORMANCE STATISTICS 1952-1974 MEAN MEAN MEAN MEAN MEAN STD DEV z ABSOLUTE z ABSOLUTE - ------------- SHAN§E_ %-C§A§G§ E8303 3 ERBOB _ _ ACTUAL 354. 23.6 .783 3.30 ESTIMATE 354. 22.3 .674 2.07 .149 1.85 SQUARED MEAN ERROR .121 ROOT SQUARED ERROR OF FORECAST 8 446 158 O 31 0 N1 * D C)- v U3 Ho 0300. a_m 1 , O sl 0: u L) O (0.. m u u c: ‘\T ; V1 11 H .L on "o“'l'Mk D on 1 I 1 1 ”1950 1955 1980 1985 1970 TIME IN YERRS Figure 6.3 Seed Use [] Actual A Estimated .j 1975 159 Table 6.6. Evaluative Data for Quantity of Food-Industrial Use of Crops, CROPS(7) . Year Actual . Estimate Error ZError 1952 7233.715 7219.805 —13.9 -.l92 1953 7248.260 7224.736 -23.5 —.325 1954 7216.452 7209.154 -7.30 —.101 1955 7122.090 7277.384 155. 2.18 1956 7296.246 7332.540 36.3 .497 1957 7463.062 7377.044 -86.0 —1.15 1958 7247.112 7397.208 150. 2.07 1959 7623.267 7548.567 -74.7 -.980 1960 7590.886 7623.048 32.2 .424 1961 7660.573 7687.147 26.6 .347 1962 7927.586 7784.907 -143. —1.80 1963 8002.793 7876.759 -126. -l.57 1964 8014.879 7985.563 —29.3 -.366 1965 8083.936 8106.606 . 22.7 .280 1966 8301.153 8249.097 —52.1 -.627 1967 8391.016 8300.865 -90.2 -1.07 1968 8478.514 8410.253 -68.3 -.805 1969 8671.612 8517.840 —154. -1.77 1970 8428.669 8525.123 96.5 1.14 1971 8573.895 8585.939 12.0 .140 1972 8692.328 8774.887 ' 82.6 .950 1973 8670.206 8804.705 134. 1.55 1974 8755.528 8733.794 -21.7 -.248 Mean .794*10: .794*102 -6.12 -.O623 Absolute .794*10 .794*10 71.2 g .896 PERFORMANCE STATISTICS 1952-1974 MEAN MEAN MEAN MEAN MEAN STD DEV z ABSOLUTE z ABSOLUTE ______________ SEWER- 2-9191131? ERBOB .7: EREOR ACTUAL 2794*102 572. .713 1.69 ESTIMATE_‘;794f19 . -593; _ _ .872 .964 .0623 .896 SQUARED MEAN ERROR 37.479 ROOT SQUARED ERROR OF FORECAST 91.284 160 920 *4 21C? 800 1 '08085171 79° 720 ,880. 950 1855 1860‘ 1885 1870 1875 TIME IN YERRS ' Figure 6.4 Food-Ind t ' C [1 Actual “5 r131 Use of rODS. A Estimated 161 Table 6.71 Evaluative Data for Ending Inventories of Crops. CROPS(8) .Year Actual Estimate Error ZError 1952 7111.027 7394.398 283. 3.98 1953 8443.700 8628.615 185. 2.19 1954 9862.658 9365.58 -498. -5.05 1955 10945.065 9193.634 -1751. -16.0 1956 11826.535 9482.462 —2344. -19.8 1957 11666.927 9615.047 -2052. -17.6 1958 11602.228 9830.433 -1772. -15.3 1959 12699.706 10863 894 -1836. ~14.5 1960 12334.487 10250.605 -2084. —16.9 1961 12946.258 10051 225 —2895. —22.4 1962 12425.091 9599.996 -2825. —22.7 1963 12703.188 9678.090 -3025. 423.8 1964 12765.733 9633.663 -3132. -24.5 1965 12349.227 9679.946 -2669. —21.6 1966 11850.858 10138.031 -1712. -14.5 1967 10747.565 9661.208 -1086. —10.1 1968 10863.894 10003 964 -860. -7.92 1969 11496.014 10941.690 -554. -4.82 1970 11192.144 10517.982 -674. -6.02 1971 9882.090 9789.729 -92.4 -.935 1972 10400.529 10984.366 584. 5.61 1973 8827.587 9210.704 383. 4.34 1974 7992.749 8668.551 676. 8.46 Mean .110*10§ .970*10: -—.129*104 -10.4 Absolute .100*10 .970*10 .148*10 12.6 PERFORMANCE STATISTICS 1952—1974 MEAN MEAN MEAN MEAN MEAN STD DEV z ABSOLUTE z ABSOLUTE .. ............. SHANQE- %-C.P.1AI§.G§ ERBOB .7: ERBOB .. .. ACTUAL .110*105 .164*104 .668 6.62 ESTIMATE .970*10 802 2989.._. 5:71__ -10.4 12.6 SQUARED MEAN ERROR ROOT SQUARED ERROR 0F FORECAST 1673307.293 1820.086 162 14102 190 110 120 150 CROPS(8) 90 1 80 l 0 .7 1850 1855 1850 1855 1870 TIME IN YEHRS Figure 6.5 Ending Inventory of Crops [lActual (AEstimated j 1976 163 Real Price of CrOps Table 6.8 and Figure 6.6 present the evaluative data for the tracking of real price of crops. The model does not track crop price as well as hoped. The mean absolute percentage error is 6.98 percent. The error varies greatly from year to year, with 1959 and 1971 exceed— ing 20 percent. The graphical representation given in Figure 6.6 indicates that the model tracks the general level--high real price in the early fifties, declining until the early seventies, then high in 1973 and 1974. However, the model depicts much greater swings in crop price than actually occur. The mean absolute percentage year to year change of the estimated values is 11.8 percent compared with the actual 5.98 percent. The greater fluctuation in estimated prices might be expected when one considers the structure of the model. The model is simulated over the historical period without the impostiion of government policy variables. The structural equations are intended to represent unrestricted markets. The procedure intended for projections is to impose constraints on the model to simulate government policy. For example, a price floor for crops could be established in the following manner. First, the model could be solved with no restrictions. If this would yield a crop price below the minimum, the model could be re-solved deleting an equation (RPCROP would not be endogenous) and making crop inventories a residual demand. Errors in estimating crop price feed throughout the model, causes errors in other prices. These include feed, seed, and livestock prices. 164 Table 6.8. Evaluative Data for Real Price of Crops, RPCROP Year Actual Estimate Error ZError 1952 147. 154. 7.94‘ 5.42 1953 135. 132. -3.14 -2.32 1954 130. 118. —ll.5 -8.90 1955 129. 124. —5.06 —3.92 1956 124. 119. -4.29 -3.47 1957 119. 117. —l.22 -1.03 1958 112. 113. 1.01 .904 1959 110. 87.7 -22.7 -20.6 1960 109. 99.6 —9.27 -8.52 1961. 110. 103. -6.32 —5.76 1962 111. 114. 2.59 2.32 1963 112. 113. .832 .740 1964 113. 115. 1.99 1.76 1965 100. 115. 14.5 14.5 1966 106. 104. -1272 -l.62 1967 102. 115. 12.7 12.4 1968 96.3 107. 11.1 11.6 1969 89.1 83.9 —5.20 -5.84 1970 83.5 90.9 7.39 8.85 1971 88.5 107. 18.2 20.6 1972 86.4 77.6 -8.86 -10.3 1973 114. 118. 4.16 3.67 1974 141. 133. —8.01 -5.67 Mean 112. 111. —.209 .212 Absolute 112. 111. 7.39 6.98 PERFORMANCE STATISTICS 1952-1974 MEAN MEAN MEAN MEAN MEAN STD DEV Z ABSOLUTE Z ABSOLUTE -------------- SHANEE- z_chNc§ ERROR z EREOB ACTUAL 112. 17.1 .269 5.98 ESTIMATE_ 111. 17.0 .589 11.8 .212 6.98 SQUARED MEAN ERROR 2044 ROOT SQUARED ERROR OF FORECAST 9-697 165 180 J U U 950 1855 1850 1855 1870 1875 TIME IN YERRS Figure 6.6 _50 Real Price of Crops n Actual ' A Estimated 166 Livestock Quantities The model predicts the quantities of livestock supplied and demanded quite well. The evaluative data are presented in Tables 6.9 and 6.10 and in Figures 6.7 and 6.8. ‘The mean percentage errors (-.0773 and -.0941) indicate little problem with bias. The mean absolute percentage errors of 1.31 and 1.32 indicate that the model tracks these variables well. Livestock Price In a manner similar to that for crop price, the model is some- what erratic in tracking the real price of livestock. The evaluative data are presented in Table 6.11 and Figure 6.9. The estimated values follow actual values with respect to the overall level but diverge from actual values in periods of more stable prices. This is most easily seen by looking at Figure 6.9. Much of the error can be traced to the errors in predicting real price of crops. This can be found by exogenizing RPCROP (solving the model using actual values for RPCROP). When this is done, the model tracks RPLIV, as well as other variables, much better. The mean percentage error of -.422 percent indicates that bias is not a problem. The mean absolute percentage error is 6.05 percent, with the largest errors occurring in 1959 and 1971, the same.years in which RPCROP has large errors. Real Price of Total Output The real price of total output is derived as an identity, a sum- mation of crop and livestock prices weighted by the quantities 167 ZError -.184 .370 -2.75 -2.11 .534 2.52 .709 —.556 2.13 —.923 -1.13 -1.29 -1.43 3.25 1.29 -.385 .791 2.50 -2.08 -l.93 —.951 —.185 .0315 -.O773 1.31 MEAN Z ERBOB MEAN ABSOLUTE Z ERROR Table 6.9. Evaluative Data for LiveStock Production: LIV(2) Year Actual Estimate Error 1952 16585.807 16555.268 —30.5 1953 16067.145 16126.527 59.4 1954 17057.814 16588.741 -469. 1955 17079.639 16718.431 -361. 1956 16362.104 16449.505 87.4 1957 16166.284 16573.426 407. 1958 17281.119 17403.676 123. 1959 17745.503 17646.777 -98.7 1960 17142.029 17507.433 365. 1961 18048.665 17882.124 -167. 1962 18372.742 18165.203 -208. 1963 18640.356 18400.441 -240. 1964 18683.368 18415.866 -268. 1965 17681.789 18256.740 575. 1966 18520.020 18759.404 239. 1967 18898.767 18825.928 -72.8 1968 19040.498 19191.103 151. 1969 19017.839 19493.517 .476. 1970 20183.089 19763.074 -420. 1971 20319.949 19927.925 -392. 1972 20423.045 20228.748 -194. 1973 20490.683 20452.851 —37.8 1974 20226.232 20232.602 6.37 Mean .183*10§ .182*10§ -20.4 Absolute .183*10 .182*1O 237. PERFORMANCE STATISTICS 1952-1974 MEAN MEAN MEAN STD DEV z ABSOLUTE ......... _ - _ _ SHANEE_ %-CHAEG§ ACTUAL .183*10: .143*105 .846 2.62 ESTIMATE_ ;1§2119 _.13Z*10 .929 1.56 SQUARED MEAN ERROR 416.119 ROOT SQUARED ERROR OF FORECAST 300-509 168 2980 2000 A 1920 ull)‘ 1540 LIV(2) 50 680 rd _1500 550 1855 1850 ' 1855 1870 ' 1875 TIME IN YEHRS Figure 6.7 Livestock Production |] Actual A Estimated Table 6.10. Year Actual 1952 16001.411 1953 16202.073 1954 16588.629 1955 16789.126 1956 16665.190 1957 16431.980 1958 16633.647 1959 17361.776 1960 16958.350 1961 17696.995 1962 17826.975 1963 18133.813 1964 18456.086 1965 18044.941 1966 18452.085 ' 1957 19037.559 1953 18918.488 1969 18905.473 1970 19733.985 1971 19942.987 1972 20103.253 1973 19729.352 1974 19986.427 Mean .180*105 Absolute .180*10 ACTUAL ESTIMATE .180*10 .180*10 5 169 Estimate 15970.871 16261.455 16119.556 16427.918 16752.591 16839.122 16756.204 17263.050 17323.753 17530.453 17619.435 17893.898 18188.584 18619.891 18691.469 18964.720 19069.093 19381.152 19313.970 19550.964 19908.955 19691.520 19992.797 .l80*10: .180*10 Evaluative Data for Livestock Consumption, LIV(5) PERFORMANCE STATISTICS 1952-1974 MEAN STD DEV Z SQUARED MEAN ERROR ROOT SQUARED ERROR OF FORECAST CHANGE .l34*10 1.06 .132*10 1.03 Error ZError -30.5 -.191 59.4 .367 -469. -2.83 —361. -2.15 87.4 .524 407. 2.48 123. .737 -98.7 -.569 365. 2115 -l67. -.941 -208. -l.l6 -240. -l.32 —268. —1.45 575. 3.19 239. 1.30 -72.8 -.383 151. 2796 476. 2.52 -420. -2.13 -392. -l.97 -l94. -.966 —37.8 -.l92 6.37 .0319 -20.4 -.0941 237. 1.32 MEAN MEAN MEAN ABSOLUTE Z ABSOLUTE ACME: ERBOB Z 511505 .. - 1.86 1.29 -.0941 1.32 416.119 300.560 170 2000 J 810‘ 1540 1520 1760 l LIV(5) 80 ‘D-n 1 1600 _1520 950 1855 1870 1975 1850 1855 TIME IN YERRS Figure 6.8 Livestock Consumption fl Actual A Estimated 171 Table 6.11. Evaluative Data for Real Price of Livestock, RPLIV Year Actual Estimate Error ZError 1952 139. 133. -5.65 -4.07 1953 121. 110. -10.4 —8.57 1954 112. 118. 6.37 5.71 1955 105. 108. 2.27 2.16 1956 100. 96.0 -4.28 44.27 1957 105. 97.3 -7.17 -6.86 1958 114. 108. -5.82 —5.11 1959 106. 89.2 ~16.7 —15.7 1960 103. 99.0 —3.89 —3.78 1961 101. 98.2 -2.89 —2.86 1962 102. 107. 4.93 4.85 1963 96.4 102. 6.04 6.26 1964 91.7 103. 11.1 12.1 1965 99.7 94.1 -5.57 -5.59 1966 101. 104. 3.25 3.22 1967 100. 100. .351 .351 1968 99.8 106. 5.97 5.98 1969 106. 96.0 -10.2 —9.59 1970 101. 110. 9.29 9.18 1971 95.5 110. 14.0 14.7 1972 107. 102. —4.41 -4.13 1973 135. 130. -5.00 -3.71 1974 111. 110. -.433 -.392 Mean 107. 106. -.813 -.422 Absolute 107. 106. 6.34 6.05 PERFORMANCE STATISTICS 1952—1974 MEAN ‘ MEAN MEAN MEAN MEAN STD DEV Z ABSOLUTE Z ABSOLUTE _. ............. SHANQE- ACME}? ERBOB Z5, ERBOB .. - ACTUAL 107. 11.5 -l.08 7.00 ESTIMATE_ 106. 10.5 -.256 . 19.09 -.442 6.05 SQUARED MEAN ERROR .661 ROOT SQUARED ERROR OF FORECAST 7.802 172 (D :34 a) 831 D H S:- > H HN 4:. 11 o- I. m . V‘ - u n u 11 I 9.7 . ~" 10.. 11 I I ' m u no ‘ 1 1 1 fl “1550 1955 1950 1955 1970 1975 TIME IN YERRS Figure 6.9 Real Price of LivestOck [IActual AEstimated 173 utilized. Thus with errors in RPLIV following the pattern of errors in RPCROP, the deviations of RPTOUT from actual would therefore follow the same pattern. These results are shown in Table 6.14 and Figure 6.12. Demand for Variable Inputs The demand equations for hired farm labor (LABOR), fertilizer and lime (FERT), and other nondurable inputs (OTHER) are recursive to the equations giving the variables previously discussed. These input demand equations include RPTOUT or RPCROP as explanatory vari— ables. The equations with actual values for RPTOUT or RPCROP have the properties described in Chapter V. Thus, additional deviation in estimated values would depend upon the values of RPTOUT or RPCROP and their importance in the equations. Real Feed and Seed Prices. The evaluative data for real price of seed and feed are presented in Tables 6.12 and 6.13 and in Figures 6.10 and 6.11, respectively. The price dependent feed and seed supply equation are not re- cursive to the equations determining quantities utilized and crop price because of inclusion of current quantities utilized in the supply relationship. However, one can still say that RPCROP is a major determinant of feed and seed price because the coefficients of RPCROP are large in the structural equations. Thus, one would expect the resluts shown in the tables and figures for these vari- ables. ‘While the mean absolute percentage error for real seed price 174 Table 6.12. Evaluative Data for Real Price of Seed, RPSEED Year Actual Estimate Error ZError 1952 140. 137. -2.29 —1.64 1953 126. 124. —2.01 -l.59 1954 117. 112. —4.87 -4.17 1955 123. 115. ~8.4O -6.80 1956 107. 110. 2.66 2.49 1957 107. 108. .865 .810 1958 103. 104. 1.11 1.08 1959 97.4 91.6 -5.81 -5.96 1960 100. 98.5 -1.88 -1.88 1961 98.2 97.9 -2297 -.302 1962 100. 103. 2.49 2.48 1963 106. 103. -2.44 -2.31 1964 103. 105. 1.85 1.79 1965 106. 107. .790 .746 1966 101. 102. 1.00 .993 1967 100. 113. 12.6 12.6 1968 99.8 109. 8.94 8.96 1969 96.5 95.1 -1.47 -l.52 1970 96.3 99.7 3.37 3.50 1971 102. 111. 8.96 8.77 1972 108. 93.9 -13.9 -12.9 1973 125. 124. -l.45 -1.16 1974 146. 142. —3.26 —2.24 Mean 109. 109. —.l49 .0758 Absolute 109. 109. 4.03 3.77 PERFORMANCE STATISTICS 1952—1974 MEAN MEAN MEAN MEAN MEAN STD DEV Z ABSOLUTE Z ABSOLUTE .............. SHANEE- ZEEAEGE ERBOB Z: ERBOB .. ACTUAL 109. 13.8 .884 5.52 - ESTIMATE_ 109._ _ _12.9 .677 7.76 .0758 3.77 SQUARED MEAN ERROR .022 ROOT SQUARED ERROR OF FORECAST 5-788 175 “1:950. 1855 1850 1855 1870 T875 TIME IN YERRS Figure 6.10 Real Price of Seed fl Actual 4 Estimated 176 Table 6.13. Evaluative Data for Real Price of Feed, RPFEED Year Actual Estimate Error ZError 1952 148. ~151. 2.56 1.73 1953 134. 131. -2.97 -2.23 1954 132. 118. -l3.7 -10.4 1955 123. 121. -2.29 -1.85 1956 119. 115. —4.26 ~3.58 1957 112. 111. -.663 —.595 1958 107. 107. -.554 -.516 1959 107. 87.0 —19.5 ~18.3 1960 103. 96.5 ~6.ll -5.95 1961 . 103. 99.3 —3.39 -3.30 1962 103. 108. 5.68 5.53 1963 106. 108. 2.53 2.39 1964 103. 110. 6.66 6.44 1965 103. 108. 5.81 5.66 1966 104. 99.9 -4.01 —3.86 1967 100. 108. 8.01 8.01 1968 90.2 102. 12.0 13.3 1969 87.4 84.0 -3.41 -3.90 1970 86;8 90.4 3.60 4.14 1971 86.6 104. 17.1 19.7 1972 84.6 81.8 ~2.83 —3.35 1973 121. 115. -6.21 —5.14 1974 131. 127. -4.10 -3.12 IMean 108. 108. -.435 .0397 Absolute 108. 108. 6.00 5.79 PERFORMANCE STATISTICS 1952-1974 MEAN MEAN MEAN MEAN MEAN STD DEV Z ABSOLUTE ABSOLUTE - - - - - ......... SHANEE- %-CEAEG§ ERBOB 2 ERBOB _ _ ACTUAL 108. 17.0 .0566 5.16 ESTIMATE_ 108. 15.7 .0507 9.83 .0397 5.79 SQUARED MEAN ERROR .189 ROOT SQUARED ERROR OF FORECAST 8.096 177 180 l 160 I 140 l 100 l 0 “1950. 1855 1850 1855. 1870 1875 TIME IN YEHRS Figure 6.11 Real Price of Feed []Actual A Estimated 178 Table 6.14. Evaluative Data for Real Price of Total Output, RPTOUT Year Actual Estimate Error ZError 1952 143. 144. 1.44 *1.01 1953 128. 122. -6.38 -4.99 1954 120. 118. -2.15 -1.79 1955 117. 116. -.605 —.518 1956 112. 108. —3.89 -3.48 1957 112. 108. -4.03 -3.61 1958 113. 111. -2.32 —2.06 1959 108. 88.4 —19.7 -18.3, 1960 106. 99.4 —6.71 -6.33 1961 106. 101. -4.62 —4.37 1962 107. 111. 3.75 3.51 1963 105. 108. 3.50 3.34 1964 103. 109. 6.56 6.38 1965 100. 105. 5.18 5.18 1966 104. 104. .553 .533 1967 101. 108. 7.02 6.93 1968 97.9 107. 8.76 8.95 1969 97.0 89.4 -7.55 -7.79 1970 91.7 99.7 8.01 8.74 1971 91.8 108. 16.3 17.7 1972 95.8 88.5 —7.23 -7.55 1973 123. 123. .156 .127 ~ 1974 128. 124. -4 67 —3.64 Mean 109. 109. -.379 -.0853 Absolute 109. 109. 5.70 5.51 PERFORMANCE STATISTICS 1952-1974 MEAN MEAN ‘ MEAN MEAN MEAN STD DEV Z ABSOLUTE z ABSOLUTE .............. SHAN§E_ %_C§A§G§ ERBOB Z ERROR ACTUAL 109. _ 12.7 - 483 4.24 ' ’ ESTIMATE 109. 12.5 .0108 8.22 -.0853 5.51 SQUARED MEAN ERROR .144 ROOT SQUARED ERROR OF FORECAST 7-617 Do " 179 158 g31950 1855 1850 1855 1870 1875 TIME IN YERRS ' , Figure 6.12 ' Real Price of Total Output IIActual A Estimate 180 and real feed price are not as large as for real price of crops, they are similar. The largest deviations of estimates from actual also occur in the same years as for real price (fl? crops. Hired Farm Labor. The results of demand for hired farm labor are presented in Table 6.15 and Figure 6.13. The mean percentage error is .0708 percent and indicates little bias. The mean absolute percentage error is 3.23 percent, with the largest errors occurring when there are large errors in RPTOUT. Fertilizer and Lime. The results for quantity demanded of fer- tilizer and lime are presented in Table 6.16 and Figure 6.14. The real price of crops is a very important variable in this equation (both current and lagged real crop price is included). Thus, the model does very poorly in years that RPCROP is poorly estimated, as well as in the following year because of the lagged effect. Other Nondurable Inputs. The results for other nondurable inputs are presented in Table 6.17 and Figure 6.15. The model fol- lows the general trend upward and catches the downturn in 1974. However, there are fairly large errors in the years in which RPTOUT is poorly estimated. Further Model Evaluation The previous part of this chapter presented results from using the model to simulate over the sample period used to estimate the equations in the model. This included the years 1951 to 1974. The procedure is sometimes referred to as ex post forecasting. The data 181 Table 6.15. Evaluative Data for Quantity of Hired Farm Labor, LABOR Year Actual Estimate Error ZError 1952 4257.630 4307.340 49.7 1.17 1953 3945.900 3857.620 -88.3 «2.24 1954 3826.670 3883.132 56.5 1.48 1955 3829.510 3812.687 ~16.8 —.439 1956 3757.140 3795.854 38.7 1.03 1957 3722.730 3829.537 107. 2.87 1958 3773.530 3841.488 68.0 1.80 1959 ‘3626.390 3335.936 ~290. -8.01 1960 3725.680 3544.470 ~181. -4.86 1961 3776.320 3575.135 -201. -5.33 1962 3803.850 3749.558 ~54.3 -l.43 1963 3822.500 3755.062 —67.4 —l.76 1964 3842.680 3854.052 11.4 .296 1965 3816.280 3836.511 20.2 .530 1966 3618.280 3727.584 109. 3.02 1967 3417.000 3648.533 232. 6,78 1968 3346.300 3515.687 169. 5.06 1969 3225.210 3082.490 -143. —4.43 1970 3146.880 3183.831 37.0 1.17 1971 3010.450 3305.867 295. 9.81 1972 2995.070 2861.589 -133. -4.46 1973 3130.970 3221.534 90.6 2.89 1974 3156.740 3051.639 -105. -3.33 Mean 35925102 .359*10: .149 .0708 Absolute .359*10 .359*10 112. 3.23 PERFORMANCE STATISTICS 1952-1974 MEAN MEAN MEAN MEAN MEAN STD DEV Z ABSOLUTE Z ABSOLUTE .............. 935N913- Z_C§A§C§ ERBOB ZERROR ACTUAL .359*102‘ 340. —1.62 2.72 ESIIEAIE_ L329219 _ 345._ _ '21235 4.65 .0708 3.23 SQUARED MEAN ERROR .022 ROOT SQUARED ERROR OF FORECAST 144.111 182 490 810‘ LHBOR _240 1850 1855 1870 1875 TIME IN YEHRS Figure 6.13 950 1855 Quantity of Hired Labor “Actual AEstimated 183 Table 6.16. Evaluative Data for Quantity of Fertilizer, FERT Year Actual Estimate Error ZError 1952 1115.690 1094.195 -21.5 -1.93 1953 1102.970 - 1356.368 253. 23.0 1954 1124.830 814.772 -310. -27.6 1955 1123.840 513.734 —610. -54.3 1956 1127.640 944.244 —183. -l6.3 1957 1120.270 1101.244 -l9.0 -l.70 ’ 1958 1158.710 1347.019 188. 16.3 1959 1288.180 779.627 -509. -39.5 1960 1299.790 256.600 —lO40. -80.3 1961 1371.680 788.566 -583. -42.5 1962 1483.450 1354.238 -129. ~8.7l 1963 1655.680 1961.716 306. 18.5 1964 1837.990 2029.438 . 191. 10.4 1965 1928.410 2221.933 294. 15.2 1965 2189.210 2426.799 238. 10.9 1967 2429.000 2531.636 103. 4.23 1963 2592.740 3259.265 667. 25.7 1969 2655.190 3048.855 394. 14.8 1970 2723.490 2642.363 -81.1 -2.98 1971 2888.440 3331.169 443. 15.3 1972 2865.430 3406.117 541. 18.9 1973 2989-000 3150.011 161. 5.39 1974 3349.930 3080.763 —269. —8.04 Mean .189*10:: .189*10: .831 —4.57 Absolute .189*10 .189*10 328. 20.1 PERFORMANCE STATISTICS 1952-1974 MEAN MEAN MEAN MEAN MEAN STD DEV Z ABSOLUTE Z ABSOLUTE - - .. .. .. .‘ ........ SENSE- 4-0114595 58505 1 ERROR ACTUAL .l89*102 761. 4 5.37 5.60 ESTIMATE_ L1§9f10 . _.1_03_*1_0_ _15.4 35.0 -4.57 20.1 SQUARED MEAN ERROR .690 ROOT SQUARED ERROR OF FORECAST 418-784 184 '480 1 °1950 A 1'955 1T960 1565' 1570 1'975 TIME IN YERRS Figure 6.14 Quantity of Fertilizer n Actual A Estimated 185 Table 6.17. Evaluative Data for Quantity of Other Non-Durable Inputs, OTHER Year Actual Estimate Error ZError 1952 1810.360 1687.010 -123. —6.81 1953 1872.840 1558.881 -314. -16.8 1954 -1959 520 2017.727 58.2 2.97 1955 2162.380 2113.837 -48.5 -2.24 1956 2285.790 2216.088 -69.7 -3.05 1957 2344.690 2558.515 214. 9.12 1958 2600.920 2848.856 248. 9.53 1959 3049.560 2472.207 -577. -18.9 1960 3214.790 2955.160 -260. —8.08 1961 3285.930 3046.360 —240 -7.29 1962 3458.600 3488.429 29.8 .862 1963 3576.420 3701.223 125. 3.49 1964 3733.760 3897.844 164. 4.39 1965 3885.730 3963.098 77.4 1.99 1966 4032.150 4232.047 200. 4.96 1967 4348.000 4457.332 109. 2.51 1968 4549.970 4657.842 108. 2.37 1969 4596.690 4556.232 -40.5 —.880 1970 4672.960 5023.328 350. 7.50 1971 5032.380 5267.448 235. 4.67 1972 5297.040 4845.101 -452. -8.53 1973 5573.820 5611.262 37.4 .672 1974 4911.230 4820.149 -91.1 -1.85 Mean .3S8*102 .357*102 -11.3 —.843 Absolute .358*10 .357*10 181.3 5.63 PERFORMANCE STATISTICS 1952—1974 MEAN MEAN MEAN MEAN MEAN STD DEV z ABSOLUTE Z‘ ABSOLUTE _________ _ _ _ _ 9H4NgE_ %-CBAEGE ERROR ; ERROR ACTUAL .358*102 .117*10:- 4.52 5.63 ESTIMAIE_ L357f10 _.123*1O 5.39 9.49 - 843 5 63 SQUARED MEAN ERROR 127.351 ROOT SQUARED ERROR 0F FORECAST 238-011 186 650 OTHER ulO‘ 310 380.‘ 470_, 1 2.30 _150 950 1555 1570 1875 1550 1555 TIME IN YERRS Figure 6.15 Quantity of Other Non—Durable Inputs []Actua1 A Estimated 187 presented were from a dynamic simulation. Actual values were used for exogenous variables, but forecasted values were used for lagged endogenous variables after the first period. While this information is useful in model evaluation, it is also informative to compare re- sults from the model with actual data for a period of years past those used in estimating the equations (ex ante forecasts). The following sections present results from ex ante forecasts and indicate'modifi- cations made to the model as a result of this testing. Initial Ex Ante Simulation An ex ante simulation was performed over the years 1975 through 1977 (1976-77 crop year) using known, preliminary, or estimated data for exogenous variables (the type of data depended on availability). For many of the endogenous variables, actual or preliminary data were available for comparison. The commodity supply and utilization aggre- gate data employed in the model were updated for comparison. The description of the process of evaluating and altering the model is presented here in the sequence in which it occurred. Crops Component The initial ex ante simulation runs overestimated net farm in- come by a sizeable margin for 1975 and 1976. The errors were traced to overestimation of feed demand in 1975 and to underestimation of crop supply for 1976. The actual quantity of crops used for feed dropped significantly in 1975 in response to the drop in crop output and resulting feed price increase. However, the model did not simulate a similar large 188 decline. The adjustments during this period included a significant shift in the cattle industry to fewer grain-fed cattle. It was rationalized that the coefficients in the feed demand equation would reflect the impact on feed demand for crOps of changing the intensity of grain feeding of livestock in response to price changes but not the larger kind of shift that occurred when a much larger proportion of livestock was marketed with very little or no grain feeding. Based on this reasoning, adjustments were made to the feed demand equation for the 1975 to 1977 period. The adjustments were made to the feed price coefficient (increased elasticity) and to the coefficient of beginning livestock inventory (decreased). The magnitude of the ad- justment was based upon that required to forecast reasonably close estimates of the quantity of feed demanded. Actual crop production made a large jump in 1976 and 1977. After adjuSting the feed demand equation and getting close to actual crop prices, the crop supply equation forecasted lower than actual values. Simulation runs were made with the crop supply equation replaced by IF statements giving actual values of crop production for 1976 and 1977. Essentially, crop production was made an exogenous variable for these years. This procedure was maintained, in part, when making the projections presented in a later chapter. A final adjustment to the crop'component was made to the cash receipts "near" identity. As was indicated in Chapter V, the usage of a crop year price causes underestimation of Cash receipts when prices are rising and overestimation when prices fall. The falling prices of crops over the 1975-77 period would thus cause error. 189 The model was programmed to make an adjustment to the price of crops (only in the cash receipts equation) based upon the difference be- tween actual calendar and crop year prices in 1975, 1976, and 1977. Livestock Component The inventory of livestock falls sharply from the January 1, 1975 high. In the simulation runs, this fall had a significant de- pressing effect on livestock price when compared with simulations without the drop in inventory. However, the downward effects simu- lated for 1975 (after adjusting the crop component) were greater than actually occurred. Additionally, the simulated results gave higher prices in 1976 and 1977 than actually occurred. These results were improved when the decline in livestock inventory was "smoothed" to take place more evenly over the three years. Other Components The model underestimates the real estate price for 1975 and the following years. The variables in the equation for real estate price simply do not reflect values required to forecast the rise in real estate price that occurred between early 1975 and early 1976. The decision was made to treat this error as a permanent shift in the equation. The net flow of loan funds was overestimated for 1975 and under- estimated for 1977 by the model after adjustments. No changes were made in this component. However, the underestimation in 1977 is sub- . stantial and troublesome. The data used for comparison are from Agricultural Finance Outlook (November, 1977). The results 190 simulated by the model differ from the preliminary data in the cate- gory of personal consumption and other cash uses (the category title is from Agricultural Finance Outlook). In the model, two of the major uses of funds in this category (purchase of nondurable goods and ser- vices and other uses of cash) are estimated, in part using net farm income. It would appear that purchases Of nondurable goods and ser- vices of farm operator families and/or other uses of cash (which in- clude investment in off-farm financial assets) did not decline in 1977 as the relationships in the model suggest. Table 6.18 shows initial ex ante simulation results for a set of the important variables in the model. These are the results before adjustments were made to the model. Table 6.19 shows ex ante simulation results after the previous- ly described adjustments were made. Complete output from this run of the model is shown in Baker (1978). Summary This section has concentrated on results from simulating over the poSt-estimation period and on the subsequent modifications made to the model. The mOdifications were few relative to the number of equations in the model. The model gives logical results, and the magnitudes of most variables are within tolerable error. The interested reader may wish to consult published data to compare with the results in Baker (1978). 191 .GHQNHHN>N U02 fl .¢.Z e.m n.4N m.mN .<.z NEN mEN mmN o.oaN EENN N.ON o.NN o.mN N.¢N OEN EEN mON m.NmN oEmN o.NN o.m N.mN N.NN ooN NEN eNN m.mNN mEoN 1 1 1 1 1 1 1muaNNon coNNNNm1 1 1 1 1 1 1 . 111111 1 EmmAMNmmm 111111111 ummomuom m Hmnuo¢ unmomuom m Hmauo< ummomuom m Hmsuo< ammumuom m Hmnuu< ummw mfififih OBOUG Ehm 0 smog No aon umz H E u z >HNE momom manmfium> Hmpoz may ou mom: uamfiumsnv< mfiom nuw3 nowumasafim muq< xm Eoum muasmmm Hmsuo< pom vmummoouom .mH.c maan .mHanHm>m uoz u .<.z o.m n.4N N.NN. .<.z NNN mEN mON . 8.8mN EEEN N.m o.NN «.mm N.¢N NEN EEN moN N.NNN oEmN N.NN o.a N.mm N.NN mwN NEN moN m.mNN mEmN 1 1 11111 mumNNoo coNNNNm 111111111111111 mmmummmm 111111111 uwmumHOh m HN§UU< UENUQHOW m HNSUU¢ UMNUUHOM m HNSUU< ummomhom m HMSUU<. HM”? mflfiflh GEOUfi BMW 0 smog No zoNE 062 H m u z >HNE meson mNnmNum> mucmsumsfiv< oz nufi3 aoaumasafim ouq< xm aoum mufldmmm Hmauud vow woummuouom .mH.o mHan CHAPTER VII Financial Accounts and Other Performance Variables Financial Accounts The output of the model is summarized in several financial statements and related analytical ratios for the aggregate farming sector. Included are the Balance Sheet, Income Statement, and a cash basis Sources and Uses of Funds Statement. In addition to these statements, tables presenting capital gains, capital formation, and other data and analytical ratios are available as output from the model. Further statements such as the Capital Flows and Capital Finance accounts could be constructed from the internal data of the model. Strengths and weaknesses of these and other farm sector accounts and data have been discussed in the agricultural economics literature on information and data systems (see American Agricultural Economics Association Committee on Economic Statistics, Bonnen, Carlin and Handy, and Weeks and others). In addition, an ERS sponsored workshop on farm sector financial accounts was held in April 1977 with a pro- ceedings publication forthcoming. Discussions of data and financial accounts usually occur at two levels, often simultaneously. The first discussed here might be 192 193 called the macro level and is the most critical. It relates to the very nature of what and/or whom the accounts represent. The accounts are intendddto represent the financial position and operations of a sector. The method used to divide a country of individuals and/or institutions into sectors is important both in terms of what goes into the accounts and what uses can be made of the accounts. Sectoring can be based on a product or establishment concepts. The product concept restricts entries in the financial statements to those relating to a certain type of activity. This concept is implied when using terminology such as "farm production sector." An est- ablishment concept would focus on financial activities of a particular group of individuals or institutions, e.g., farm operator families. In addition, there is the question of scope of the sector. This may be posed terms of the product concept as the question: where does the product-begin and end? Traditionally, the concern has been related to a sector that begins and ends at the farm gate. Recently, interest has expanded to the food and fiber sector“which vincludes a much broader set of activities relating to farm inputs and products. These questions have significant impact upon the finan- cial accounts, and changes relating to these have implications for data collection, equation estimation, and presentations in financial statements. The second level, which might be called the micro level, relates to rather specific questions of how data are presented in the 194 financial accounts. These included questions of terminology (differ- ent data transformations are often called the same thing, and differ- ent financial statements are sometimes labeled with the same name), grossing versus netting of the data, and how to present data, that is, which statement(s) to use. A good pOrtion of the micro questions is argument over what transformation of the data should be made and what subsets of the data and transformations should be presented in financial statements. Other micro questions do go deeper in that they have implications for data collection, for example, the grossing or netting argument. In the opinion of the author, most of the micro questions are reasonably easy to answer, given answers to the macro questions and the intended use of the financial statements (or problems to be addressed). The financial statements used to present the output of the model will be discussed in the following paragraphs. Uses, limitations, and relationship to product and establishment concepts will be pointed out. In addition, differences between the statements used here and pub- lished USDA accounts will be noted. The Balance Sheet The Balance Sheet provides a major data limitation with respect to pursuing greater detail in a model in terms of either the estab- lishment or product concept. The Balance Sheet contains farming assets and liabilities of all farm proprietors (operators and non— operator landlords) and some of the assets of farm operator house- holds. Lacking is a breakdown of ownership between operators and 195 nonoperators, preventing a breakdown between these two "establish- ments." Included are some nonfarm assets; thus, there is not strict adherence to the product concept. The absence of complete data on nonfarm investments of farm Operator families is one of the major flaws in the product-establishment mixture adopted for this study (nearly complete coverage of farm operator families plus £33m assets and liabilities of non—operator landlords). A minor limitation is that imposed by not having the standard Balance Sheet breakdown of short, intermediate, and long term assets and liabilities. The Balance Sheet constructed for use in the model is shown in Table 7.1.The Balance Sheet in Table 7.1 differs from that pub- lished by the USDA in The Balance Sheet of the FarmingSector in that real estate is divided into the components of dwellings, service buildings and other structures, and land with improvements. This breakdown is based on the published real estate value and constructed stocks (value measure) of dwellings and service buildings (Baker, 1978). These data are probably particularly weak, especially for dwellings. The breakdown was included in the Balance Sheet because of a basic difference between buildings, which reflects investment in real capital, and land, the quantity of which does not change substantially. Data from the Balance Sheet can be used in combination with other data to compute capital gains, rate of return on equity,2nd leverage ratios such as liabilities/total assets. 196 Table 7.1. Balance Sheet of the U.S. Farm Production Sector, January 1 - - -Million DOllars- - - A. Assets A1. Real Estate $ A2. Dwellings $ A3. Service Buildings and Other Structures A4. Land and Improvements A5. Non-Real Estate $ A6. Livestock $ A7. Crops A8. Machinery and Motor Vehicles A9. Household Equipment and Furnishings A10. Financial Assets $ All. Deposits and Currency $ A12. U.S. Savings Bonds A13. Investments in COOperatives A14. Total Assets $ C. Claims C1. Liabilities I$ C2. Real Estate Debt C3. Non-Real Estate Debt $ C4. CCC Loans C5. Proprietor Equity $ C6. Total Claims $ RA. Financial Ratios (in Percent) RAl. Equity/Total Assets RAZ. Liabilities/Equity 197 Calculation of rate of return on equity, a measure of current earnings divided by equity, requires conceptual consistency between the earnings measure and the equity measure (discussed in greater detail in the following section on the Income Statement). Analytical ratios such as those mentioned above are useful in monitoring the performance of the sector, subject to accuracy of data and consistency of concepts, in a historical context and are useful in policy analysis in a projections context. subject to accuracy of the model, ability of the model to handle policy scenarios, and the accuracy of projections of exogenous variables, The Income Statement The Income Statement is prepared largely as a product concept for it is intended to reflect income derived from farming activities. The USDA does not publish an Income Statement, per se, but publishes income related data (receipts, expenses, capital consump- tion,and inventory changes) in Farm Income Statistics. It is the author's understanding that although nonoperator landlord expenses are estimated and published the data series is very weak. The subtraction of nonoperator landlord net expense gives the Income Statistics an establishment flavor (farm operators), but still maintains a product concept (only farm income and expenses are included). Use of the income calculation as an indicator of the economic well being of farm operators is apparently a popular use of the ac- count by policy makers but is severely limited by the product concept of the account and by aggregation. 198 The farming sector is increasingLyacquiring1a‘bi-modal distribu- tion of farms by size. Thus, income of the aggregate (or average) may not reflect the economic well-being of any significant subset, much lesscflfthevflufljasector. Aggregation also occurs across different types of farms. In a given year, there can be a great divergence between incomes of various farm types, a factor not reflected in the aggre- gate figure. In addition, the welfare of farmers is also affected by nonfarm activities such as off-farm income. A final point on the.measure- ment of farmer welfare is that it is the particular nature of the farming sector to have low current or operating income (typically measured by the Income Statement) and relatively large capital gains. Surely, capital gains should be considered when discussing the well-being of farmers. Still, as an indicator of current income of farm proprietors from farming, the Income Statement is useful. As a monitoring tool, the Income Statement gives outcomes for a sequence of years,and thus one can follow changes. In a projection mode, future income can be estimated and changes in expected future income resulting from vari- ous policies can be simulated. Table 7.2 shows the Income Statement used to present the cur— rent incOme data generated by the model. Inventory changes and de- preciation (capital consumption) are calculated in a manner similar to the USDA calculation in Farm Income Statistics . Inventory changes are , quantity changes times the average price during the year. Depreciation is calculated using a replacement cost concept. This usage of deprecia- tion causes the author some concern as it is writing off the "current 199 Table 7.2. Income Statement for the U.S. Farm Production Sector, Year Ending December 31, 1974 . __._-~___- -H ...- ...—.-.- - »— --- ---..-g.-._.a- H H- -.- Million Dollars R. Cash Receipts R1. Crop Marketings $ 51,271 R2. Livestock and Livestock Product Marketings 41,377 R3. Government Payments 530 R4. Other Farm Income 894 R5. Total Cash Receipts $ 94,072 E. Cash Expenses El. Hired Labor (Cash Outlay) $ 5,609 E2. Feed Purchased 14,901 E3. Livestock Purchased 5,131 E4. Fertilizer and Lime 5,822 E5. Seed Purchased 2,028 E6. Repairs and Operation of Capital Items 6,506 E7. Interest on Non—Real Estate Debt 2,729 E8. Interest on Real Estate Debt 3,044 E9. Property Taxes ' 3,043 E10. Other Operating Expenses 7,473 Ell. Total Cash Expenses $ 56,286 NETl. Net Cash Income $ 37,786 NCE. Non-Cash Expenses and Adjustments NCEl. Perquisites to Hired Labor $ 412 NCE2. Capital Consumption Allowances 10,624 NCE3. Change in Livestock Inventories 454 NCE4. Change in Crop Inventories -2,065 NCES. Total Non-Cash Expenses and Adjustments $ 12,325 NMI. Non-Money Income NMIl. Value of Farm Products Consumed $ 1,300 NMIZ. Rental Value of Farm Dwellings 4,831 NMI3. Total Non-Money Income $ 6,131 NETZ. Net Farm Income $ 31,592 200 value" of capital consumed in production (which includes capital gains) against current income. That is, some capital gains are written off against current income. A divergence from USDA data is that there is no estimate of net rent to nonoperator landlords. The decision to exclude net rent to nonoperator landlords was based on the following factors: the historical data series is weak and con- sists of a complex combination of cash and noncash items; there would be little theoretical basis for specifying structural equations to estimate this item for projections; and the income figure to use in calculating rate of return on equity should include returns from farm operations of farm proprietors whose farming related equity is calculated on the Balance Sheet."Thisincludes nonoperator landlords. Nominal Capital Gains Capital gains are an important portion of the year to year wealth changes of farm proprietors. Fairly consistent year to year increases occur in the value of the land. Buildings and machinery and motor vehicles have gains associated with increasing prices but also incur depreciation because of use and obsolescence. Crops and livestock have gains and losses because prices are not stable (or stable upward). The gains calculated will be nominal capital gains in that there will be no adjustments for inflation. This is occasionally misleading for there may be a positive nominal gain but a negative real gain. See Melichar and Sayre (1975) for more on calculations of capital gains in the farming sector. Other studies of capital gains of interest '1nC1Ude Hoover, Grove (1960), Boyne, and Bhatia. 201 One should keep in mind that there are major limitations in using capital gains to make welfare inferences. When summing gains across various assets, the total will not reflect the variance of net gains among individual owners of the assets. For example, the owner- ship of crops and livestock is obviously concentrated among crop and livestock farmers, respectively. Also, non—operator landlords probably own a proportionately large share of land and a small share of other assets. Capital gains calculated here are unrealized (thus do not directly provide spendable income) and are before taxes (when real— ized, the gain wOuld be taxed). One should not add capital gains and net farm income to get total income because of the substantial differences between the two. Financial assets and liabilities will not enter capital gain calculations. In the following section, the procedure for calculating capital gains will be explained. Land and Improvements The gain associated with land is the end-of-year value minus the beginning-of-year value adjusted for intersector exchange of real estate and expenditures for land improvements. The intersector flow adjustment is to add to the capital gain the value of net decreases in land in the farming sector (this would be negative if the flow were land into the sector). This adjustment iS based on very weak information. The land in farms data are not 202 particularly strong data,and there is no information as to the value of land entering or leaving the farming sector. It is assumed that acres leaving the sector have a value equal to the average value per acre 0 Non-Land Durable Assets In general, the capital gain associated with a nonland asset is the end of year value minus beginning of year value minus purchases during the year plus capital consumption plus withdrawals from the sector by discontinuing proprietors. These assets include machinery and motor vehicles, farm operator dwellings, service buildings and other structures, and household equipment and furnishings. Capital consumption is computed at current prices, similar to. the concept used for the Income Statement. Crop and Livestock The capital gain associated with crops and livestock is end of year value minus beginning of year value plus the changes in inven- tory charged on the Income Statement (change in quantity times aver- age price during the year). Table 7-3 shows the format for presenting implicit nominal cape ital gains in the model. Sources and Uses of Funds The third major financial statement included in the model out- put is a Sources and Uses of Funds Statement (SAUF) on a cash basis (see Table 7.4). This statement is presented,in part,because of the 203 Table 7.3. Implicit Nominal Capital Gains on Physical Assets for the U.S. Farming Sector, Year Ending December 31 -.— —--~---—. - -—-- --- o — . ———. ---_-. .... -._---.- — .....- ”t - -Million Dollars- - 001. Gain on Real Estate $ C02. Dwellings $ 003. Buildings and Other Structures 004. Land and Improvements 005. Gain on Other Physical Assets $ C06. Livestock $ 007. Crops C08. Machinery and Motor Vehicles C09. Household Equipment and Furnishings 0010. Total Nominal Capital Gains $ view taken of demand for debt capital. The view is that components of the SAUF statement are individual demands for cash and internal supply of cash. Many of these are conceptually a function of the rate of interest (few of the empirically estimated equations include the interest rate). If the rate of interest, the price of debt cap- ital, is exogenous-~the agricultural sector is a perfect competitor with other sectors for funds--then the demand for debt is a residual. Debt is determined by calculating uses and subtracting internal sources. If the rate of interest were endogenous to the model, one would need to include a supply equation for debt (quantity supplied as a function of the interest rate and other factors). This equa- tion would be solved along with other equations that back up the items on lines of the SAUF statement to solve for the level of debt, interest rate, and other sources and uses of funds. 204 Table 7.4. Cash Sources and Uses of Funds for the U.S. Farm Production Sector, Year Ending December 31 -._’ - -Million Dollars- - 8. Sources of Cash 81. Cash Receipts from Farm Operations $ 82. Off-Farm Income of Farm Operators S3. Net Flow of Loan Funds S4. Net Sale of Real Estate to Non-Farm Sector S5. Total Sources of Cash $ U. Uses of Cash U1. Cash Farm Operating Expenses $ U2. Capital Expenditures Machinery and Motor Vehicles $ Service Buildings and Land Improvements Farm Operator Dwellings Household Equipment and Furnishings ' Financial Assets Autos and Trucks for Family Use U3. Purchase of Real Estate from Discontinuing Proprietors U4. Purchase of Nondurable Goods and Services U5. Personal Tax and Non-Tax Payments U6. Other Uses of Cash (1) mmm '0) U7. Total Uses of Cash 205 The definition of cash flows used in this study includes some transactions thag in a strict sense,may not involve cash. For exam- ple, all debt flows are considered cash transactions,enun1though the transaction may not have actually gone through a cash account. Noncash flows can be included on the SAUF statement as both sources and uses. This type of SAUF statement has been suggested by Penson, Lins,and Irwin. A statement including noncash flows has not been presented here for the following reasons: 1) inclusion of noncash flows detracts from presentation of flows that require financing and the financing of these flows both from internal and external sources and 2) noncash flows,such as appreciation in assets are not controllable by individuals. Foremample,(nuacannotdis- invest in real estate appreciation per se; one can only sell land, a decision that would incur cash flows. Capital Formation in the Farming Sector The statement presented in Table 7.4 includes information re— quired to calculate gross capital formation, but it does not have the information required to calculate net capital formation. The infor- mation needed is capital consumption, change in crop and livestock inventories, and intersector flows of capital items. These data are all available in the model. Thus Table 7.5, Gross and Net Capital Formation in the U.S. Farming Sector, is available as part of the model output. This table presents "sector" accounts for farming businesses and for farm households. In addi- tion, a combined business and household account is presented for continuing farm proprietors. 206 Table 7.5. ~ -————.--~..—‘—.-~o—-.—..- ”.....—- .— CFl. Cross and Net Capital Formation in the U.S. Farming Sector for the Year Ending December 31 ---—~—.—. .- -—-- .—_--_ - —Million Dollars- — Farm Business Account Capital Expenditures CF15. CF2. CF3. CF40 CFS O CF6. CF7. CF8. CF12. CF13. CF14. Machinery and Motor Vehicles $ Service Buildings, Other Structure and Land Improvements Net Change in Farm Inventories Livestock $ Crops Farm Business Gross Capital Formation Capital Consumption CF9. Depreciation of Machinery and Motor Vehicles $ CF10. Depreciation of Service Buildings and Other Structures CF11. Accidental Damage Net Real Estate Transfers to Other Sectors Gross Capital Disappearance Farm Business Net Capital Formation Farm Household Account Gross Capital Formation CF16. CF17. CF18. CF19. CF20. CF21 O CF22. CF23. CF24. CF25. Automobile and Truck Purchases $ Purchases of Household Equip- ment and Furnishings Farm Operator Dwelling Expenditures Net Change in Financial Assets Gross Capital Disappearance Depreciation of Dwellings $ Accidental Damage, Dwellings Depreciation of Household Equip- ment and Furnishings Depreciation of Automobiles and Trucks Household Net Capital Formation {D-(D (D400) ‘(D-(I) 207 Table 7.5 (cont'd) .— --.. - -Million Dollars- - Business and Household Net Capital Formation of Continuing Proprietors CF26. Business and Household Net Capital Formation $ Plus the Following: CF27. Withdrawals by Discontinuing $ PrOprietors CF28. Household Equipment and Furnishings $ CF29. Financial Assets CF30. Machinery and Motor Vehicles CF31. Real Estate CF32. Net Capital Formation of Continuing Proprietors $ 208 Financial accounts for continuing proprietors differ from ac- counts for the "sector" in the implications of flows associated with proprietors entering and leaving the sector. The most significant difference is in capital formation (and savings) by continuing, proprietors versus capital formatiOn (savings) by the sector. As one can see from Table 7.5,capita1 formation of continuing proprie- tors is equal to capital formation for the sector plus withdrawals of capital by discontinuing proprietors. Other Data and Analytical Ratios The final table presenting financial data from the model is a table providing other data and a set of analytical ratios. Table 7.6 presents the format available for output in the simulation model. The analytical ratios presented in this table are identical in some instances and, in other instances, similar to the ratios sug- gested by Melichar (forthcoming) at the ERS sponsored workshop on farm sector financial accounts. Savings The issue of savings of the farming sector is one of the more confusing issues in the study of aggregate financial relationships. Penson (1977) and Simunek (1976) present quite different values for savings. Melichar (fortthming) discusses these differences conclud- ing that the major difference in the accounting used to derive sav- ings is that Simunek attempts to measure savings used to finance farm capital formation1while Penson attempts to measure total savings 209 Table 7.6. Other Data and Analytical Ratios, U.S. Farming Sector, Year Ending December 31 .... 01. Net Cash Flow from Operations $ 02. Total Capital Flow (Gross Capital Formation of Continuing Proprietors) $ 03. Internally Financed Capital Flow $ 04. Debt Financed Capital Flow $ 05. Net Cash Flow Minus Business Capital Consumption $ Relative Burden of Capital Flows RAB. Capital Flow/Net Cash Flow Z RA4. Real Estate Purchases from Discontinuing Proprietors/Net Cash Flow Z Relative Sources of Financing RAS. Debt Financing/Capital Flow Z RA6. Internal Financing/Capital Flow Z RA7. Internal Financing/Debt Financing 1 RAB. Debt Financing/Net Business Capital Formation Z Prospective Burden of Debt RA9. Debt Financing/Net Cash Flow Z RAlO. Debt Financing/(Cash Flow Minus Business Capital Consumption) Z N RAll. Debt Financing/Net Farm Income 210 (a large amount of savings apparently finances nonfarm1 capital for- mation). Melichar is critical of Penson's exclusion of Simunek's net transfers of real estate to nonfarm sectors as a source of capi- ‘tal and of the specific treatment by Simunek of the same item. Melichar suggests that only net sales of assets to nonfarm sectors by continuing proprietors represents a source of funds to be recorded in the capital finance account. In addition, Melichar main- tains that purchases of real estate from discontinuing proprietors are not capital formation andznnaproperly excluded from capital flows and finanCe accounts. These points bring into view and provide a basis for examining the issue of continuing proprietors versustfluasector. ‘Melichar argues for a "sector" concept in deriving gross and net capital formation; however, he supports a "continuing proprietor" concept of capital flows. Net sales of assets, such as, real estate to nonfarm sectors, do represent a source of funds to the sector. The entire portion (not only net sales by continuing proprietors) is a source of funds, and the entire portion is part of the sector's gross capital disappearance. When capital formation of continuing proprie- tors is calculated, purchases of assets from discontinuing proprie- tors, or implied gross investment by continuing proprietor, are in— cluded in gross capital formation. Melichar (forthcoming) argues for inclusion of net sales of assets by continuing proprietors as sources of funds in the capital finance account 1Nonfarm in the respect that the capital formation is not in the form of farming assets. The capital formation referred to here is under the ownership of farm proprietors. 211 rather than the total because, he reasons, ...in many cases, an intersector transfer involves no transaction whatsoever; that is a person simply ceases (or resumes) farming operations, which automatically moves his assets out of (or into) the farming sector. The effect is identical when a retiring farmer sells his real estate to the nonfarm sector and simultaneously also retires himself (and his cash and other assets and debts) into the nonfarm sector. The reasoning described by Melichar is not in error; however, the solution, to include only net sales by continuing proprietors, is in error. A reduction of acreage in the farming sector is a disinvest- ment in a capital item by the sector and provides a source of capi- ta1,regardless of the fact that some of the land may be sold by dis- continuing proprietors.l Nearly all of the capital flows have dif- ferences between the flow financed by the sector and the flow financed by continuing prOprietors. To adjust for this difference in only one item is inconsistent. The following gives an example of this for a use of funds in the capital finance account. If a farmer retires his assets (e.g., bank deposits) into the nonfarm sector, then investment (capital flow to be financed) in bank‘ deposits by the sector is the change in Balance Sheet values. But, investment by continuing proprietors has implicitly been the Balance Sheet change plus the amount withdrawn by the retiring farmer. This does not necessarily imply that the deposits were purchased from the 1Iwould not argue with Melichar's criticism of the method of estimating this source of funds but can offer no solution to the problem without empirical data. 212 retiring farmer; it only implies that gross investment by continuing pro- prietors was greater than for the sector by that amount. Where it is clear that assets are purchased directly from discontinuing proprie- tors,the use of funds can be described as such (e.g., purchases of real estate from discontinuing proprietors). The implication of the "sector—-continuing prOprietor" problem for savings calculations is somewhat subtle. The Simunek and Penson accounts differ in this respect. The Simunek account calculated sector savings used to finance sector farm capital formation,whi1e Penson's account calculates total savings of continuing proprietors. It would seem true that one could calculate savings by continuing proprietors to finance farm capital formation of continuing proprie- tors. However, total sector savings and total savings of continuing proprietors would be identical. The concept of savings used in this study begins with computa- tion of gross savings by continuing farm proprietors. Gross savings are defined as personal income of continuing prOprietors less per- sonal outlays by continuing proprietors. Personal income is defined as net cash income plus increases in inventories of crops and live- stock plus off-farm income. Personal outlays-are defined as per- sonal taxes and non-tax payments plus purchase of nondurable goods and services plus repairs and maintenance of farm operator dwellings. Net savings are defined as gross savings less business and household gross capital disappearance. Gross and net savings data in the output from the model are printed at the bottom of the table giving capital formation. 213 Values of Exogenous and Endogenous Variables The financial data presented in the previously discussed tables do not show the values of many of the variables directly. The data are shown indirectly or transformed in that the values were used to calculate data presented in the financial accounts and table. Thus, a table of values for exogenous and endogenous variables is provided as part of the model's output. These data are useful in determining what is really going on in the model. Variable names, as listed in the glossarx are the lables placed on the values in the table. Summary This chapter has described the financial statements and other reports generated by the model. This description was necessary for the following: to clarify the nature of the accounts to the reader, to present some of the reasoning behind the decision to present re- sults from the model in the formats chosen, to present some of the recent arguments relating to sector accounting, and to clarify some of the calculation procedures. CHAPTER VIII Projections under Alternative Scenarios Introduction The chapters to this point have described the purpose of the model, developed a theoretical model of system relationships, estimated parameters for the equations necessary to build a model of the system, evaluated the ability of the model to track historical data, and developed in detail the set of data desired as output from the model for evaluative purposes. The intent of this chapter is to present results from using the model to project into the future under alternative scenarios. First, the methods of projecting exogenous variables are explained. Secondly, alternative scenarios for some exogenous variables are developed. Projections for one alternative are then presented in detail. Finally, indications of the direction and magnitude of change in some important endogenous variables resulting from simu- lations under alternative scenarios are presented. The output from the model is extensive, and only summary tables for selected years are included in this chapter. Baker (1978) includes additional results from a simulation run of the model. 214 215 Projection of Exogenous Farm Input Prices The model is programmed to provide prices of farm inputs from either of two sources. First, the user of the model can provide the values for some or all years to be simulated. The alternative is to use equations relating the input prices in a particular year to the CPI in that year. These equations are given in the follow- ing paragraphs. Price of Motor Supplies Regressing PMOTSUP against CPI gives an equation with reasonably small errors up to 1974 when the error is very large. It is reasoned that this jump was caused by large increases in energy prices. The equation available in the model was estimated over the 1951 to 1974 data period with a dummy variable for 1974. The value of the dummy variable is added to the constant in the equation for pro- jections. This is assuming a one time permanent shift in the equation and a continuing constant relationship with the CPI. Equation (8.1) presents the estimated equation. (8.1) PMOTSUP = 46.5096 + .543268 * CPI + 34.2496 * DUM (21.46) (14.63) PMOTSUP = price of motor supplies CPI = consumer price index DUM = a binary variable with value 1 for 1974, 0 otherwise Price of Buildings The price index used as the price of service buildings and other structures can be determined in the model by equation (8.2). 216 The equation includes a shift variable for 1974, the last year of the data used in estimation. This is assumed to be a permanent shift upward in the function and is added to the constant term when used in the model. (8.2) PBLD = 2.79 + 1.033 * CPI + 35.28 R DUM, R2 = .9664 (16.27) (5.97) PBLD = price of buildings. CPI = consumer price index. DUM = a binary variable with value 1 for 1974, 0 otherwise. Price of Machinery and Motor Vehicles The price of machinery and motor vehicles was regressed against the consumer price index. The equation seemed to fit well without adjustments. Equation (8.3) presents the equation available in the model for projecting the price of machinery and motor vehicles, given values of CPI. (8.3) PMMV = 33.645 + 1.3122 R CPI, R2 = .9947 (64.35) PMMV = price of machinery and motor vehicles. Price of Dwelling The price index for dwellings equation did not track as well as the above equations. However, a shift variable is included for 1974 to start the equation at the 1974 level for simulating beyond 1974. The equation available in the model is equation (8.4). 217 (8.4) PDWL = 4.955 + 1.01 * CPI + 22.22 * DUM, R2 = .9527 (14.72) (3.50) PDWL = price index for dwellings. DUM = a binary variable, 1 for 1974, 0 otherwise. Implicit Price Index for Household Equipment and Furnishings The price index for household equipment and furnishings can Abe projected using equation (8.5). The equation is estimated from 1965 to 1974 data. This period was chosen because of increas— ing prices over tin: period. Earlier periods included year-to- year declines in price and thus would not be estimated well using the CPI. (8.5) PHEF = 37.093 + .6382 * CPI, R2 = .9932 (34.20) PHEF = implicit price of household equipment and furnishings Price of Fertilizer The price of fertilizer did not fit well when regressed against the CPI. However, it was felt that a default internal mechanism should be available to make projections. The equation provided in the model, shown here as equation (8.6), will project the price of fertilizer based on the percentage change in CPI. That is, the 1974 price of fertilizer is increased by the percentage increase in the CPI since 1974. 218 (8.6) PFERTt (CPIt/CPI ) * PFERT 1974 1974 1.469*CPIt PFERT = price of fertilizer Price of Supplies The price of farm supplies, used as the own price variable for other nondurable inputs, can be projected on the same basis as the price of fertilizer. The equation available in the model is shown here as equation (8.7). (8.7) PSUPt A 1974) PSUP1974 1.011*CPIt (CPIt/CPI PSUP = price of farm supplies. Scenario Development The projections in this chapter have been made using alter- native time paths for some of the exogeneous variables. A set of the exogenous variables If; treated in the same manner for all projection. The following sections first present the methods and equations for projecting tine latter set. Additionally, pro- jections for exogenous variables projected under alternative time paths are explained. Base Scenario for Exogenous Variables The set of exogenous variables which are treated in a similar manner for all scenarios includes: number of farms, harvested acres, government payments, investment credit, U. S. pOpulation, prices of all farm inputs, interest rates, livestock imports, livestock exports, and crop imports. 219 All exogenous variables for which data are readily available are given their actual values for 1975 and 1976. Additional years are projected as described in the following paragraphs. Number of Farms. The number of farms is projected to decline at the rate of .01 million farms per year. The rate of decline per year in the number of farms has fallen over time to .014 million farms in 1974. However, the number of farms declined by .022 million farms in 1975. Harvested Acres. Historically, the acres of cropland har- vested have been affected a great deal by government programs. This is likely to occur again in the 1978-79 crop year. However, the acres of Cropland harvested are projected for these scenarios at the 1975 level of 333 million (a high level). Government Payments. Government payments to farmers have }varied greatly over history. However, it is assumed for the scenarios here that there are no effective government price and income policies. Thus, government payments will be low. The value used is 807 million, the 1975 level for all years. Investment Credit. Investment credit is projected at a con- stant rate of 10 percent, the 1975 level. U. S. Population. Projections of U. S. population are based upon a modification of projections indicated in Volume 1 of 220 1972 OBERS Projections: Regional Activity in the U.S. Series E Population. The equations used are: POP = 214 + 2 * (YEAR - 1975), 1975 5 YEAR < 1980 POP = 224 2.2*(YEAR — 1980), 1980 S YEAR < 1990 POP = 246 + l.8*(YEAR - 1990), YEAR 2 1990 Farm Input Prices. The prices of farm inputs are projected for years after 1975 using the equations presented earlier. They are essentially "driven" by the rate of inflation. Livestock Imports. Livestock imports are projected to grow at 1 percent per year from the 1974 level. The average annual com- pound rate of growth between 1950 and 1974 is 5.87 percent. How- ever, growth has not been steady since 1963, (see appendix table A.3). Livestock Exports. Livestock exports have been low and relatively unimportant over the historical data period. Livestock exports are projected at the 1975 level as a constant. Crop Imports. The level of crop imports has varied within a reasonably small range over the historical data period. Imports of crops are projected as a constant at the 1975 level. Interest Rates. The two interest rate variables in the model, the interest rate charged by banks (RBK) and the interest rate charged: by Federal Land Banks (RFLB), have been projected at 7.8 and 8.8 percent respectively. 221 Alternative Scenarios for Exogenous Variables The exogenous variables for which simulations under alterna- tive scenarios have been made are the rate of inflation, crop exports, productivity, real GNP, and livestock inventory growth. The various levels projected reflect, in part, a priori specifica- tion and, in part, feedback based on results of projections. Figure 8.1 shows the levels of exogenous variables and labels each scenario with a number. The scenario numbers will be used later to refer to the set of conditions indicated in the figure. Inflation rate. The rate of inflation, the annual compound rate of change in the CPI, is projected at two levels, 4.0 and 6.0 perCent. While these rates seem low in light of recent rates of inflation, the author feels they are reasonable in a long run con- text. Over the period from 1955 to 1976, the average annual rate of inflation has been 3.6 percent. Over that period, a rate greater than 6 percent was not reached until 1973. The 4 percent level of inflation will be used as the "most likely" level. Crop Exports. The level of crop exports is an important exogenous variable. The three levels of growth in crop exports are 2, 3, and 4 percent from the expected level of exports- in the 1977-78 crop year (see appendix A, Table A.3 for the starting level). VThese levels of exports are within the range of growth that would result from very slow growth in wheat exports, reasonable growth in feed grain exports, and fairly rapid growth in soybean exports. 222 muonssz oaumcmom was moanmfium> msocmwoxm mo mHm>mA H.w mudwfim whoa aw m=Hm> nwfls m Eoum :uBouw unmouma 039 I. \m a 2. m N m . 0.8 Q mm. H/ n .o m N m N mE. m mo.~ o.q m mE. m N N unmoumm huouco>cH noneaz xooumm>fig suzouo coNuUSpoum suBouu mama oauwcoom Ga Lusouu mzu Hmmm mono uuomxm mono nowumHmcH 223 Crop PrOduction. The original intent of this project was to project alternative crOp production growth via different rates of increase in the productivity index, an important variable in the ' crOp supply equation. However, the poor performance of the crop supply equation for 1976 and 1977 has led to treating crop production as an exogenous variable. As will be seen later, the starting level of crop production (1978) has a significant impact upon income in the first few years of the projections. Thus, two starting levels have heen chosen. One level for the 1978-79 crop year is the average of 1976-77 and 1977-78 crop years. Production of crops is projected at two rates of growth from this level. The rate of growth used in the "most likely" scenario is 2 percent. This is the rate of growh over the most recent 25 years. A lower rate Of growth (1.63 percent) is also used. The lower rate cor- respOnds to the rate of growth over the period since 1940. Real Gross National Product. The level of gross national product is projected in real terms. Thus, the growth in nominal gross national product depends upon the rate of inflation, as well as the growth of real GNP. The base rate of growth in real GNP is 3.0 percent, based in part upon the 1972 OBERS Projection. However, an alternative rate of 1 percent was also simulated. The level chosen for the most likely scenario is 3 percent. The historical growth between 1950 and 1975 was 3.25 percent. .Recent increases in energy prices and other phenomena have caused negative 224 growth in 1974 and 1975. However, growth in real GNP was large in 1976. The level of growth in real GNP is important over the long run, as it has significant influence in the demand for livestock equatin. Growth in Livestock Inventory. The inventory of livestock has grown at a fairly steady rate (about 1 percent) over the 195041974 period. Thus, the rate of 1 percent initially was chosen for use in the model. However, this resulted in greater growth in inventory than production. When projected over a long period, a .75 percent compound rate of increase in inventory yields a more reasonable inventory relative to the level of production. The growth in inventory is an important variable in the model developed here, since it is one of the livestock demand components. Inventories are Often assumed constant in equilibrium models such as the National Interregional Agricultural Projetions (NIRAP) system developed in the Economic Projections and Analytical Systems program area of Economic Research Service. The reasoning is that in equilibrium there wou1d be no change in inventories; thus, if one is interested in long-run projection, inventories can be ignored. Projections from scenario seven will give an indication of the differ- ence in outcomes between the base growth in livestock inventory and zero growth. 225 Summary of Scenarios The scenarios projected here should be partially interpreted as sensitivity analysis. Most of the exogenous variables are treated in a similar way for all scenarios, while five variables considered by the author to be very important are projected under alternative growth rates. The scenarios were developed by starting with one growth rate for each of the five variables, then deviations of these growth rates were considered. The initial growth rates were based upon historical growth rates or trends. It is important to project crop production and exports in a consistent manner. Thus a higher level of crop exports is projected at the higher level of production. Scenarios one, two, and three represent different levels of crop production at 4 percent inflation, 3 percent growth in real GNP, and .75 percent growth in livestock inventory. Crop export growth is 3 percent at the lower two levels of crop production and 4 percent at the higher level. Scenario 1 has the middle level of crop production, scenario 2 has lower crop production, and scenario 3 has higher crop production. Scenarios 4 through 7 are alterations of scenario 1. Scenario 4 has lower growth in real GNP, scenario 5 has lower growth in crop exports, scenario 6 has higher inflation, and scenario 7 has lower growth in livestock inventory. 226 Results from Projecting Alternative Scenarios The model is capable of producing six pages of output for each year. It quickly becomes obvious that each year of every projection cannot be studied in even a cursory manner. Thus one is limited to selecting important variables and a few years to examine in greater detail. Examination of output from the model can be approached in numerous dimensions. These might include: intensive analysis of a scenario for a particular year in the future; examination of effects of changes in exogenous variables on projections for a particular year; analysis of the projected path of a variable or set of var- iables under one or more scenarios; or cOmparison 6f historical and projected values. It is very difficult to examine the time paths of all of the variables for numerous years and numerous scenarios. However, one should be careful not to focus on'a very narrow set of variables, for there is a great deal of interaction in the model within years and over time. Thus, unreasonable values in some components affect other components. In the following sections, results from alternative scenarios projected to year 2025 are presented. While any projections should be viewed wtih extreme caution, projecting nearly 50 years into the future with a model estimated on half that many years data should probably be described as pure folly. However, the discussion of the projection will concentrate on the 1980's. The results from much longer projections are presented for several reasons. 227 The very distant projections give some indication of the stability of the model.' They show that the model does not exhibit wide cyclical behavior. In addition, the results show that the model does not project prices and quantities outside of an intuitively' acceptable range under reasonable scenarios. An exception to this would be the case of projecting a scenario that is reasonable over a portion of the projection but eventually fails because the growth rate in one (or more) variable(s) "swamps" the others. An additional reason for presenting the projections to 2025 is that trends that are somewhat subtle over the period to 1990 became explicit when the projections are extended another 35 years. The task attempted in this section is to examine the values over time of a significant portion of the variables in the model under a "most likely" scenario. Some attempts will be made to compare these values with historical data, compare with other projections, discuss key equations or assumptions affecting the variables, and discuss the impacts of alternative projections of exogenous variables. Time Paths of Exogenous Variables It was useful to the author in selecting growth rates for exogenous variables to see the values implied by the growth rates. These are presented for selected variables in Table 8.1. 228 Table 8.1. Values of Selected Exogenous Variables, Selected Years, Scenario One Year var‘ 1955a 19758 1980b 1985b 1990b 2000 2025b iable Million Persons POP 165 214 224 235 246 264 309 Index (1967=100)- CPI 80.2 161.2 204.6 248.9 302. 448.3 1195.2 Billion Dollars GNP 399.3 1516 2311 3259 4597 9145 51044 RGNP 498 940 1130 1309 1518 2040 4271 Million Farms FARMS . 2.654 2.808 2.758 2.708 2. 2.558 2.308 Index (1967=100) PFERT 101.3 217 301 ,366 445 659 1756 PSUP 94.5 160 207 252 306 453 1208 PBLD . 87.0 206 249 295 351 501 1273 PMMV 69.9 1178 235 293 364 555 1535 PLABOR 61.0 192 283 362 458 717 2048 Million Acres LDFRMS 1215 1082 1052 1023 994 938 802 a Actual. Projected. 229 It is interesting to note the size of the CPI by year 2025 at a 4 percent inflation rate. It is 7.4 times greater than in 1975. At a 6 percent rate of inflation (scenario six) the CPI reaches a value of 2982 by year 2025, about 2.5 times the value at the 4 percent rate of inflation. It is also interesting to note the effect on nominal GNP of 4 percent inflation and 3 percent real growth. Real GNP increases by 4.5 times between 1975 and 2025, and nominal GNP increases by over 30 times the 1975 value. Choosing different growth rates for the CPI has very significant impacts on nominal values in the model. However, the impact on real values is not certain a priori. Choosing different rates for variables such as imports, exports, inventories, or real GNP clearly will affect nominal and real values of many variables in the model. Commodity Supply, Utilization, and Prices Some of the more important endogenous variables in the model are commodity supply and utilization quantities and commOdity prices. These variables have impacts either directly or indirectly on nearly all money flows in the model. This section presents results for real and nominal prices of livestock, crops, and real estate. In addition, projections for the supply and utilization of crops and livestock are presented. 230 Price of Crops. Projections of nominal and real prices of crops are presented under two scenarios. Results from scenario one are presented in Table 8.2, and results from scenario three are pre- sented in Table 8.3. These scenarios differ largely in regard to ' the level of crop production beginning in the 1978-79 crop year.l/ Both Scenarios have a 2 percent growth rate in crop production. The beginning level for scenario one assumes that 1978-79 production will not grow from the high production levels of the three recent crop years but will be between the 1976-77 and 1977-78 crop year production levels. The interested reader might consult appendix table A.3 to examine the historical data for crop pro- duction. Scenario one shows an increase in nominal price of crops between 1978 and 1979 (175 to 188), while scenario three shows a decline from 175 to 160.' Thus, scenario one has a 17.5 percent greater price for 1979. A later section will examine the impact of this on net farm income. Under scenario one, the real price of crops is projected to increase slowly, but for practical purposes the projection shows a real price nearly equal to the real price in 1967 throughout the 1980's. l/ . —- Exports grow at 4 percent for scenario three and 3 percent for scenario one. However this has little effect relative to to ‘ production level for the first few years of the projection. 231 Scenario three projects a decline in real price of crops through 1983 and significant gains in real price between 1985 and 1990. The 4 percent growth in export ”catches" up with the growth in production between 1990 and 2000, thus causing signifi- cant gains in the real price of crops. A similar phenomon occurs with scenario one. However, it takes longer for this to occur. Price of Livestock. Projections of real and nominal prices of livestock are presented for scenarios one and three in Tables 8.2 and 8.3 respectively. Both scenarios projected gains in both nominal and real prices of livestock. This is the reflection of an inelastic supply equation and shifts in the demand for livestock caused by population growth, growth in real income, and growth in livestock inventory. It might be noted that while there is a time trend in the livestock demand equation, the growth in livestock price is not caused by this factor, since the time trend is negative. Scenario one projects a higher level of livestock prices than does scenario three. This is a result of the relationships between the livestock and crops components of the model. Essentially, the projection of higher crop prices results in a shift to the left in the supply of livestock equation and thus gives a higher price of live- stock. It should be noted that the adjustments are actually simul- taneous, not recursive as the description might imply. 232 Table 8.2. Projections of Selected Farm Prices, Scenario One Real Estate8 Crops Livestock Year PREAL RPREAL PCROP RPCROP PLIV RPLIV -— -Index 1967 = 100 -- 1955b, 55 68 103 129 84 105 1975b 7 213 132 224 139 172 107 1977b 314C 156c 197 108 175 96 1977 314 156 195 107 179 98 1978 323 155 175 92 196 104 1979 351 163 188 95 217 110 1980 373 168 198 97 226 110 1981 399 173 209 98 242 114 1982 424 178 218 - 99 256 116 1983 451 183 228 99 272 118 1984 480 188 238 99 288 120 1985 512 193 248 100 306 123 1990 705 223 305 101 413 137 2000 1354 295 446 99 745 166 2025 8719 727 2006 168 3598 301 a Price as of February 1 and March 1 of the following year. b Actual data. .c Preliminary. 233 Table 8.3. Projections of Selected Farm Prices, Scenario Three Real Estateb Crops Livestock Year PREAL RPREAL PCROP RPCROP PLIV RPLIV Index 1967 = 100 -- 1955a 55 68 103 129 84 105 19758 213 132 224 129 172 107 19773 314C 156C 197 108 175 96 1977 314 156 195 107 179 98 1978 326 156 175 92 196 104 1979 V339 157 160 81 209 106 1980 354 158 153 75 216 106 1981 374 162 152 72 230 108 1982 397 166 156 71 243 110 1983 423 171 164 71 260 112 1984 453 176 174 73 276 115 1985 486 183 188 75 294 118 1990 704 222 285 94 410 135 2000 1529 334 665 148 790 176 2025 14760 1232 7046 590 4559 381 3 Actual data. b Price as of February 1 or March 1 of the following year. C Preliminary. 234 Price of Real Estate. Projections of the index of average value per acre of farm real estate are presented in Tables 8.2 and 8.3 for scenarios one and three respectively. The real estate price equation was particularly troublesome for the ex ante simulation. There were substantial increases in the real estate price index over the 1975 through 1977 period that the equation was unable to forecast. These increases were treated as exogenous shifts in the equation. Under scenario two, the model forecasts a small increase, 2.9 percent, in the nominal price index between 1977 and 1978 (actually between February 1978 and February 1979). This is a result of low income projected for 1978. A larger increase, 9.7 percent, is projected for the next year. The rate of increase between 1980 and 1990 is 6.6 and 2.9 for the nominal and real price indices of average value per acre of farm real estate, respectively. These rates of increase are small in comparison to the recent history. However, the lower growth in real estate prices may be consistent with the rate of inflation assumed for scenarios one and three (4 percent). Livestock Quantities. The projections for livestock supply and utilization quantities are shown in Table 8.4 for scenario one. The actual data for 1976 are shown as a basis of comparison. Appendix Table A.4 shows data since 1950, if the reader desires greater historical perspective. Livestock production is projected to grow at a .99 percent annual rate between 1980 and 1990, while 235 Table 8.4. Projections of Supply-Utilization of Livestock, Scenario One Beginning Inventory Production Imports Exports Consumption Year LIV(l) LIV(2) LIV(3) LIV(4) LIV(5) 19768 21250 19390 789 239 19268 1977 21062 19770 805 293 20853 1978 20491 20265 813 293 20975 1979 20301 20701 821 293 21077 1980 20453 20928 829 293 21311 1981 20607 21109 838 293 21499 1982 20761 21325 846 293 21722 '1983 20917 21530 854 293 21934 1984 21074 21746 863 293 22157 1985' 21232 21961 872 293' 22380 1990 22040 23093 916 293 23551 2000 23750 25470 1012 293 26010 2025 28629 33676 1298 293 34466 a Actual data, the units are the summation of quantities of commOdities in year t weighted by the farm prices of the commodities in a base year (1967). 236 livestock consumption increases at a nearly identical rate. The effects of growing inventories and growing imports very nearly off-set each other. Per capita consumption of livestock and live- stock products is projected to increase slightly (less than 1 percent in total) between 1980 and 1990. Crop Quantities. The projections, under scenario one, of crop supply and utilization quantities are shown in Table 8.5. As indicated earlier,tfld£:scenario assumes a small decline in crop production for the 1978-79 crop year. Figure 8.2 shows historical data for crop production and projections under the three alternatives for which results are present in this chapter. Crop inventories are projected to decline in 1979 and later years. This is largely caused by increasing real price of crops (the inventory demand equation has a negative coefficient on change in real price of crops). The formulation of the inventory equation has zero long run price elasticity of demand. This formulation of the equation seemed reasonable, since there would be little cause to hold grain per se other than as a "pipe-line" quantity. It is possible that the inventory quantities projected for years 2000 and 2025 are too low relative to the level produced and utilized. However, the projections through 1990 appear to maintain a reason— able relationship between inventory and production. Feed use of crops is projected to increase at a 1.7 percent compound rate between 1980 and 1990. This is approximately 1.7 times the rate of increase in livestock production. The food 237 Table 8.5. Projections of Supply-Utilization of Crops, Scenario One Beginning Food- Inventories Production Imports Exports Seed Feed Industrial Year CROPS(l) CROPS(2) CROPS(3) CROPS(4) CROPS(5) CROPS(6) CROPS(7) 1976a 8320 26557 708 7157 510 10227 8937 1977b 9308 26402 1106 6912' 457 10468 9063 1978b 9916 17367 1106 6935 459 11299 9189 1979' 10508 26885 1106 7143 463 11309 9280 1980 10304 27427 1106 7357 467 11445 9377 1981 10187 27971 1106 7578 470 11632 9483 1982 10101 28531 1106 7805 474 11821 9592 1983 10045 29101 1106 8040 477 12026 9704 1984 10006 29683 1106 8281 481 12237 9818 1985 9978 30277 1106 8529 484 12455 9936 1990 9904 33428 1106 9888 497 13601 10560 2000 9912 40749 1106 13288 517 16151 11911 2025 8712 66853 1106 27872 553 22751 16923 a Actual data. See Appendix A for description of the data. Production and exports are "actual" values estimated using prelimary data 70 60 50 40 ' 30 20 l() THOUSAND T UNITS WP- 11— *1 Historical .o L 5 -I . ...-.--... 1950 1960 1970 238 e. 0.1 4:69 .1 .I . O l 06‘» ,-/). I, 6:00 "y/Oi‘e’ .v /-o , ' i 9"» x" . 499' ,/ ‘ f/c" / ' g /, ,,”§§° l /' x .o ,/‘ /’<@53 /‘ /$Ce’ ,/” /, : /'/"5/’ J . ,9. I, o . f ~ 1 ' Projected 1 1 1 l 1 1 .15 1.-1”uw_.l.111”__mmt wnl-- .11.,-_M-,. i 1980 1990 2000 2010 2020 YEAR Figure 8.2 Historical Data and Alternative Projections of CrOp Production 239 and industrial use of crops is projected to increase at the rate of 1.2 percent per year. The per capita food and industrial use increases at the rate of .25 percent per year. Projections for Indicators of Farmer Welfare A major use of the financial accounts for the farming sector is to provide indicators of farmer welfare or well-being. The most popular figure used is total net income from farming (Fagm' Income Statistics, Table 2H). The net farm income projected in this study as the residual on the Income Statement is similar to the data in Farm Income Statistics, with the exception that net rent to nonoperator landlords is §9£_subtracted out. A second major source of wealth increase in the farming sector is capital gains. Very large capital gains are a relatively recent phenomenon. While it is true that there have been relatively few declines in real estate value since the early 1930's, the nominal capital gains on physical assets in the 1970's have dwarfed all previous gains. Table 8.6 presents historical data for nominal and deflated net farm income (as defined in this study) and capital gains (close to the definition used in this study). The data on nominal capital gains are from Melichar and Sayre. The reader can see from Table 8.5 that nominal capital gains have been relatively large in the 1970’s. Nominal capital gains in the first seven years of the 1970's have been more than 1.5 times the sum of nominal capital gains in the previous 30 years since 240 Table 8.6. Historical Data for Capital Gains and Net Farm Income Capital Gains8 11_ Net Farm Income Year CPI Nominal Deflatedb Nominal Realb Index Billion Dollars 1967=100 . 1950 72.1 16.4 22.7 14.8 20.5 1955 80.2 3.2 4.0 12.4 15.5 1960 88.7 1.2 1.4 12.6 14.2 1961 89.6 7.7 8.6 13.3 14.8 1962 90.6 7.1 7.8 13.6 15.0 1963 91.7 6.7 7.3 13.4 14.6 1964 92.9 8.8 9.5 12.2 13.1 1965 94.5 14.2 15.2 14.8 15.7 1966 97.2 12.2 12.6 16.0 16.5 1967 100. 10.0 10.0 14.2 14.2 1968 104.2 11.7 11.2 14.3 14.9 1969 109.8 10.0 9.1 16.4 14.9 1970 116.3 9.4 8.1 16.3 14.0 1971 121.3 21.1 17.4 16.8 13.8 1972 125.3 40.4 32.2 22.2 17.7 1973 133.1 79.8 60.0 39.0 29.3 1974 147.7 841.0 27.8 31.2 21.2 1975 161.2 60.9 37.8 28.9 17.9 1976 170.5 78.4 46.0 24.1 14.1 a Gain on physical assets. b The nominal value deflated by the CPI (1967=100). 241 Table 8.7. Projections of New Farm Income, Alternative Scenarios Scenario One Scenario Two Scenario Three Year Nominal Real Nominal Real Nominal Real - - - - - - - - — -Billion Dollars ------ — — 19768 24.1 14.1 24.1 14.1 24.1 14.1 1976 25.0 14.7 25.0 14.7 25.0 14.7 1977 23.3 12.8 23.3 12.8 23.3 12.8 1978 21.6 11.4 21.6 11.4 23.4 12.4 1979 27.9 14.2 27.7 14.1 22.8 11.6 1980 31.0 15.1 31.3 15.3 22.0 10.7 1981 35.4 16.6 36.7 17.2 23.6 11.1 1982 39.1 17.7 42.0 19.0 26.0 11.8 1983 43.0 18.7 47.8 20.8 29.4 12.8 1984 47.2 19.7 54.4 22.8 33.9 14.2 1985 52.0 20.9 61.8 24.8 39.5 15.9 1990 80.5 26.6 107.9 35.6 79.9 26.4 2000 172.3 38.5 273.0 60.9 256.9 57.3 2025 1534.0 128.4 2936.7 245.7 4375.8 366.2 aActual data. 242 the Balance Sheet of the Farming Sector was begun. The nominal capital gains in 1973 alone were 90 percent of the gains over the ten years 1960 through 1969. The reader may wish to refer to Table 8.6 to provide a basis for comparing projections. The deflated data are provided to facilitate comparisons over the fairly wide historical and projected periods. Net Farm Income. Projections of real and nominal net farm income are presented in Table 8.7. Results from three scenarios are shown. Scenario one is the base scenario. Scenario two has a lower growth rate in crOp production. Scenario three has the same growth rate in production as the base scenario (but from a higher 1979 level) and a higher growth in exports. The critical nature of the assumption concerning the level of crop produciton in the 1978-79 crOp year can be seen in Table 8.7. Scenario three projects net farm income in 1980 to be $22 billion. Scenario one projects 1980 net farm income at $31 billion, that is, 40 percent higher. Scenario one projects nominal net farm income to match the historical high (1973) by 1983. However, real (deflated) net farm income under scenario one does not reach the 1973 level in the 1980's. While real net farm income is not projected at the high level of 1973, the projections for the 1980's compare favorably with the 1960's (compare results in Table 8.7 and historical data in Table 8.6). 243 Table 8.8 Projections of Capital Gains, Alternative Scenarios Scenario One Scenario Two Scenario Three Year Nominal Deflated Nominal Deflated Nominal Deflated ---------- Billion Dollars — — - — - — w - - - - 19763 78.4 46.0 78.4 46.0 78.4 46.0 1976 63.2 37.1 63.2 37.1 63.2 37.1 1977 51.3 28.2 51.3 28.2 51.3 28.2 1978 19.3 10.2 19.3 10.2 24.4 12.9 1979 51.1 25.9 50.6 25.7 24.7 12.5 1980 38.9 19.0 41.3 20.1 27.1 13.2 1981 45.4 21.3 49.4 23.2 36.5 17.1 1982 44.8 20.3 50.5 22.9 40.8 18.5 1983 47.7 20.7 54.7 23.8 46.2 20.1 1984 50.5 21.1 58.8 24.6 51.7 21.6 1985 54.1 21.7 63.6 25.5 57.1 22.9 1990 72.8 24.0 85.6 28.3 84.0 27.7 2000 134.8 30.1 171.3 38.2 175.3 39.1 2025 843.7 70.6 1232.7 103.2 1938.9 162.3 aActual data. 244 The projection of crop production at the longer term growth rate (scenario two) results in a 34 percent higher net farm income by 1990. The relatively small difference in crOp production growth, 1.63 percent versus 2.0 percent, points out the critical nature of the crop production assumption on net farm income projections. Ggpital Gains. Projections of nominal and deflated capital gains on physical assets are presented in Table 8.8. Results from scenarios one, two, and three are shown. One of the major variables affecting the level of capital gains is the real estate price. Projection of PREAL for scenarios one and three have been presented and discussed in an earlier section. Scenario one indicates a significant decline in the level of nominal capital gains for 1978. This may be unreasonably low, since the real estate price may be underestimated. The projections from scenario one show a jump in capital gains in 1979, a reflection of higher crop and livestock price projections. The level of nom- inal capital gains projected fOr the 1980's are not quite as high as for 1973, 1975, and 1976. However, the values projected are much nearer the high years than the level of the 1960's. The deflated capital gains projected for the 1980's are approximately one half of the 1971 to 1976 average but are over twice the level of the 1960's. A later section presents projections of capital gains under a higher rate of inflation. 245 The capital gains projected under the high crop production, low income,scenario three show significantly lower capital gains for 1979 through the early 1980's. This is a reflection of the lower rate of increase in the price of real estate. However, in the later 1980's, scenario three has capital gains that exceed scenario one. This is partially a result of improved crop prices. Scenario three has higher crop export growth, in addition to higher crop production, which also contributes to capital gains. Scenario two (lower crop production) projects higher capital gains in the 1980's than does scenario one. However, the relative difference in nominal capital gain projections of the two scenarios (18 percent greater in 1990) is much less than the relative differ- ence in net farm income (34 percent). Consumption and Other Uses of Cash. Projections for consumption of nondurable goods and services by farm Operator households, gross capital formation of farm operator households, and other uses of cash under scenario one are presented in Table 8.9. The data for other uses of cash are presented here for two reasOns. First, a major component of this use of funds is hypothesized to be invest- ment in off—farm financial assets. Secondly, the calculation of other uses of cash historically is based on an assumed (esti- mate based on one observation) consumption relationship. Thus, the two data series should not be viewed independently. Historical data for 1971-76 are presented in Table 8.9 to provide perspective for the projections. 246 Table 8.9. Historical Data and Projections of Farm Operator Household Expenditures and Other Uses of Funds, Scenario One Consumption of Gross Farm House- non-durable Other Uses of hold Capital Goods & Services Cash Formation Year Nominal Real Nominal Real Nominal Real --------- Billion Dollars - - - - — - - - — 19718 14.4 11.9 8.2 6.8 5.0 4.1 19723 15.7 12.5 12.1 9.7 6.1 4.9 19738 20.0 15.0 23.1 17.4 5.6 4.2 19748 23.1 15.6 19.0 12.9 5.6 3.8 19758 25.3 15.7 12.3 7.6 6.5 4.0 19763 24.1 14.1 15.9 9.3 6.1 3.6 1976 23.5 13.8 16.2 9.5 6.2 3.6 1977 23.4 12.9 15.7 8.6 6.4 3.5 1978 24.0 12.7 15.2 8.0 7.0 3.7 1979 25.5 12.9 19.2 9.7 7.3 3.7 1980 27.6 13.5 21.4 10.4 7.6 3.7 1981 30.6 14.4 24.4 11.5 8.0 3.8 1982 33.3 15.1 27.0 12.2 8.4 3.8 1983 36.1 15.7 29.7 12.9 8.7 3.8 1984 39.1 16.4 32.7 13.7 9.0 3.8 1985 42.2 16.9 36.0 14.5 9.3 3.7 1990 61.2 20.2 55.7 18.4 11.5 3.8 2000 119.5 26.7 117.9 26.3 16.3 3.6 2025 774.2 64.8 953.3 79.8 40.0 3.3 a . . . Historical data calculated uSIng assumptions consistent with those used in the model. 247 The projections show both nominal and real consumption growing in the 1980's. However, the model forecasts declines in real comsumption in 1976, 1977, and 1978. This decline is a reflec- tion of declining net farm income over the period and the assump- tion of consumption as a constant percentage of nominal dis— posable farm income. It is probably unrealistic to project declining real levels of consumption over this period.' This statement is based partially on the knowledge that the model underestimates the sum of consumption and other uses of cash for 1977.}! The resulting error in projecting the net flow of loan funds has been discussed in Chapter VI and is discussed further in this chapter. Gross farm household capital formation is the sum of expendi- ture for household automobiles and trucks, household equipment and furnishings, farm operator dwellings, and net change in financial assets. The nominal expenditure for these items is projected to grow at approximately the rate of inflation in the 1980's since the projected deflated expenditure remains about 3.9 billion dollars over this period. Farm Input Projections Projections of selected farm inputs under scenario one are shown in Table 8.10. The nondurable inputs included are: hired labor (LABOR), fertilizer and lime (FERT), and other nondurable inputs 1/ - Calculated based on preliminary data in Agricultural Finance Outlook. 248 .pmuomnoum U .mump Hmsuo< n .mmowua usacfl m>wuomammu mLu kn Usumammp mumHHop coHHHNE mum mufias msH m aam.a Noe.z aNa.H mea.a HEN.H mma Nmm ems aeo.aa ONE.NH aEe.oH Hma.a oew.w oam.m ONe.oH amm HHN.e ENE.a mem.a mNm.a mom.a Eaa.a eEN.N axe aNe.ma oaH.Ee mEm.Ne amo.ae mea.oe ENN.Nm emm.am azm HNN.N NNN.H meE.a mae.H NEm.a HOE.H own assess Nam.m meN.m Hmm.a NNE.a Ema.a NNw.e mom.a seesaw NON.m aoo.m Noa.a NNE.a Nma.a Ham.e NeH.N mmmeo elm.a oom.N oom.N cow.N oom.N omm.m aNH.H News oom.a oom.a moo.a mow.a oao.N Ema.m omw.m mommm umz Hmufiamu umz Houwmmo sz Hmmm uoz Hmuwamu umz Hmufiamo mmouu one OHHmcmom .mwsw>mm van .ummmcmuH wumumm Hmmm .GOflumBuom Houflamu mo msOHuomhoum .¢H.w manme 259 Table 8.15. Projections of Financing of Capital Flow? Alternative Scenarios Scenario One Scenario Two Scenario Three Year Debt Internal Debt Internal Debt Internal ---------- Percent - - - — - - — - - - - — - 1976 33.7 ' 66.3 33.7 66.3 ' 33.7 66.3 1977 28.3 71.7 28.3 71.7 28.3 71.7 1978 29.1 70.9 29.1 70.9 28.7 71.3 1979 26.0 74.0 26.0 74.0 28.9 71.1 1980 27.0 73.0 26.7 73.3 28.6 71.4 1981 629.6 70.4 29.1 70.9 28.1 71.9 1982 31.5 68.5 31.0 69.0 28.4 71.6 1983 32.4 67.6 32.2. 67.8 28.6 71.4 1984 33.1 66.9 33.2. 66.8 29.2 70.8 1985 33.9 66.1 34.4 65.6 30.2 69.8. 1990 38.5 61.5 41.1 58.9 37.3 62.7 2000 44.2 55.8 49.5 50.5 48.2 51.8 2025 63.8 36.2 71.3 28.7 73.1 26.9 a The table shows the percentage of capital flow (gross capital formation) of continuing proprietors) financed by internal and external (debt) sources. Capital flow here includes purchases of real estate from discontinuing proprietors. 260 Table 8.14 also shows net capital formation and savings of con- tinuing prOprietors. These calculations take into account inter- sector transfers andother uses of funds respectively. An earlier section indicated a projection of increasing leverage. An indica- tion of financial strength of the farming sector (those remaining in the sector) is the projection of large net capital formation and savings by continuing prOprietors. It should be noted that the savings figure assumes that the other uses of funds category is applied (100 percent) to uses which would contribute to equity, that is not to consumption, gifts, and so forth. 1 Table 8.15 shows projections for the percentage of gross capital formation of continuing proprietors (including intersector flows) that is financed via internal and external sources. Results from scenarios one, two, and three are shown. The results are as one would expect,given the increasing leverage projection. Namely,the percentage of capital flow financed with debt is projected to increase over time. Under scenario two (higher income),the percentage financed by debt is lower than scenario one for years 1980 through 1983 and is higher in later years. Under scenario three (lower income in the 1980's) the percentage of debt financing is higher than scenario one in the early 1980's but lower in the later 1980's. Selected Analytical Ratios. The projected increase in leverage leads one to some possible concern over the ability of the farming sector to handle the increasing debt load without inflows of outside 261 Table 8.16. Projections of Selected Analytical Ratios, Scenario One . _—._. __._.. Prospective Burden of Debt 7 Year M‘MPA(9) if RA(10) TNWM8A(11)“_1 RA(4) RA(l) - - - -I ----- Percent - - - - l — - - - e - - 1977 29.6 ‘ 54.4 40.3 44.1 83.4 1978 34.1 68.5 46.9 48.5 82.4 1979 26.0 44.6 34.3 42.6 82.3 1980 26.4 43.8 33.9 41.5 81.8 1981 ‘ 27.7 44.0 34.7 39.8 81.4 1982 28.8 44.6 35.4 38.8 80.7 1983 28.7 43.5 34.8 38.0 80.1 1984 28.4 42.1 33.9 37.2 79.5 1985 28.1 41.0 33.3 - 36.3 78.9 1990 28.4 38.5 32.2 ' 33.4 75.7 2000 27.0 33.7 29.3 30.9 70.0 2025 22.9 : 25.0 23.7 28.0 60.2 RA(l) = (equity/total assets) * 100. RA(4) = (real estate purchases from discontinuing proprietors/net cash flow) f 100. RA(9) = (debt financing/net cash flow) * 100. RA(10) = (dig; financing/net cash flow - business capital consumption) RA(ll) = (debt financing/net farm income)/100. 262 equity capital. Table 8.16 presents projections of several analy- tical ratios that might give some perspective to the size of the net, flow of loans relative to several indicators of ability to repay loans. These ratios are: debt financing/net cash flow; debt financ- ing/net cash flow minus business capital consumption; and debt financ- ing/net farm income. The fourth ratio in the table is a relative measure of the equity flow leaving the sector, real estate purchases from discontinuing prOprietors/net cash flow. The fifth ratio is equity/total assets to again show the projection of increased lever- age. The three ratios indicating the prospective burden of debt all show the same pattern. In 1978, when low income is projected, the ratios are projected to be high. A decline is projected for 1979 (a jump in income is projected for 1979). Increases in the ratio are then projected until 1983 when a declining pattern begins. The lack of increase in these ratios throughout the projections and the decline in real estate purchases from discontinuingpro- prietors relative to net cash flow would seem to lessen the concern over the future financial stability of the farming sector. Comparisons of Results from Alternative Scenarios The discussion and presentation of results from the projections have, up to this point, concentrated largely on scenario one, with some discussion of scenarios two and three. This section will present projections of selected variables for scenarios four, five, six, and seven. Referring back to Figure 8.1 will show that each of these 263 scenarios differsihxnnscenario one in the projection for one exogenous variable. Scenario four has a lower growth in real gross national product. Scenario five has a lower growth incuxn>exports. Scenario six has a higher rate of inflation.and scenario seven has a lower (zero) growth in livestock inventory. Tables 8.17 through 8.21 will provide the data referred to in the following discussion. Most of the tables contain data for a par- ticular variable for several projections in order to facilitate compari- son of alternative projections. However, the discussion follows a different pattern. The effect of each scenario on the time path of several important variables will be the focus of discussion. Low Growth in Real Gross National Product. Scenario four is identical to scenario one except that the rate of growth in real GNP is 1 percent rather than 3 percent. ,It was shown earlier that scenario one projects (through 1990) a nearly constant real price of crops (at about 100) and a rising real price of livestock (see Table 8.2). The results from scenario four indicate a rapid decline in RPCROP and very little increase in RPLIV (Table 8.17). The livestock component is affected directly in terms of the income elasticity of livestock demand. The crops component is affected via food-industrial demand and reduced feed demand result- ing from adjustments in the livestock component. The income elas- ticity of crOp demand is greater than that of livestock demand (the calculations take into account the simultaneous effects). This is demonstrated in Baker (1978). 264 The projected effects of lower real income growth on nominal net farm income and capital gains are shown in Tables 8.19 and 8.20 respectively. Under scenario,one net farm income increases in the 1980's; with lower real GNP growth,net farm income decreases. In 1985,the projection under scenario four is only 60 percent of scenario one. Capital gains are reduced under the low real GNP assumption,but not nearly to the extent that net farm income is. The 1985 projection is 74 percent of the scenario one value. Under scenario four, the net flow of loan funds is lower than under scenario one (Table 8.21). The projection for 1985 is 86 percent of the value under greater real income growth. Lower gXOwth in Exports. Scenario five provides a projection assuming a lower growth in crop exports (2 percent versus 3 percent for scenario one). The largest impact of lower exports is on the crops component. The real price of crOps falls throughout the projection (Table 8.17). However, there is an impact on the live- stock sector in that the real price of livestock is only 94 percent of the scenario one value in 1990. The RPCROP is 65 percent of the scenario one value in 1990. Net farm income and nominal capital gains are both lower under scenario five than under scenario one. The values of net farm inCome and nominal capital gains projected for 1985 are 80 and 85 percent respectively of the scenario one values. Under lower exports,the projected net flow of loan funds is also lower (Table 8.21). The 1985 value is 94 percent of the scenario one value. Thus, the net flow Of loan funds falls less than net farm income. 265 Higher Inflation. Scenario six includes a 6 percent rate of inflation versus 4 percent in scenario one. Table 8.18 shows the projection of nominal real estate price, nominal and real price of crops, nominal and real price of livestock, and the debt/equity ratio. ’The projected real pricescflflivestock and crops do not differ greatly from those projected under scenario one. However, nominal real estate price grows at 9.7 percent between 1980 and 1990 versus 6.6 percent for scenario one. These convert to growth rates in real terms of 3.5 and 2.5 percent for scenarios six and one respect- ively. The resulting increase in equity is large enough to signi- ficantly reduce the rate of increase in the leverage ratio. The 1990 value projected under high inflation is 24.7 percent while the 4 percent inflation SCenario projects a value of 32.1 percent (see Tables 8.18 and 8.13). Under 6 percent inflation, net farm income (Table 8.19) is projected to grow at a rate of 12.4 percent between 1980 and 1990, compared with 10 percent under scenario one. These nominal rates convert to 6.0 and 5.8 percent rates of increase in real net farm income for scenarios six and one respectively. Thus real net farm income is projected to rise at a slightly faster rate under higher inflation. Nominal capital gains are projected at a much higher level under the higher inflation level (Table 8.20). This is clearly a reflection of the previously discussed high rate of increase in the nominal real estate price. Table 8.17. Scenario Four 266 Scenario Five Projections of RPCROP and RPLIV, Alternative Scenarios Scenario Seven Year .11 RPCROP RPLIV RPCROP RPLIV RPCROP RPLIV A -... fir...“ - — — - - Index 1967 = 100 ----- 1978 91 , 102 , 92 104 92 104 1979 92 107 94 110 94 107 1980 91 105 94 110 93 108 1981 89 106 93 113 91 110 1982 86 106 91 114 88 111 1983 83 107 89 116 85 113 1984 80 107 86 118 81 114 1985 76 107 83 119 76 116 1990 56 109 66 129 55 124 267 Table 8.18. Projections of Selected Variables, Scenario Six Year PREAL PCROP RPCROP PLIV RPLIV D/E I - - - -Index 1967 = 100 ------- Percent 1978 332 178 92 200 104 20.7 1979 371 195 95 225 110 20.3 1980 404 210 97 239 110 20.3 1981. 444 225 98 261 114 20.6 1982 485 241 99 281 116 21.0 1983 531 257 99 305 118 21.4 1984 582 273 100 330 121 21.8 1985 638 290 100 357 123 22.2 1990 1016 392 101 530 .137 24.7 2000 2668 695 100 1155 166 29.4 2025 38780 ' 5015 168 8978 301 40.8 268 Table 8.19. Comparison of Net Farm Income Projections under Alternative Scenarios Scenario Scenario Scenario Scenario Scenario Year One Four Five Six Seven -------- Billion Dollars- - - - - - — - - - - 1978 ,21°6 20.4 21.6 22.4 21.6 1979 27.9 25.2 27.4 29.7 26.3 1980 31.0 26.3 29.6 34.0 28.8 1981 35.4 28.3 32.8 39.9 31.3 1982 39.1 29.2 35.0 45.2 33.1 1983 43.0 29.9 37.1 50.9 34.5 1984 47.2 30.5 39.2 57.3 36.0 1985 52.0 31.1 41.5 64.5 37.5 1990 80.5 28.8 50.9 109.3 42.8 2000 172.3 Not Real- Not Real- 279.8 Not Real- istic istic istic 2025 1534.0 Not Real- Not Real- 3660.4 Not Real- istic istic istic, 269 Table 8.20. Comparison of Nominal Capital Gains Projections under Alternative Scenarios Year Scenario Scenario Scenario. Scenario Scenario One ‘ Four Five Six Seven -------------- Billion Dollars- — - - - - - - 1978 19.3 14.6 19.3 35.4 19.3 1979 51.1 45.7 49.2 70.1 45.8 1980 938.9 31.5 35.8 60.4 36.6 1981 45.4 36.9 41.1 70.2 38.7 1982 44.8 34.9 39.6 72.8 38.0 1983 47.7 36.6 41.5 79.5 39.0 1984 50.5 38.0 43.4 86.6 41.0 1985 54.1 ‘40.2 46.0 95.0 43.3 1990 72.8 48.5 57.7 146.8 55.3 2000 134.8 Not Real— Not Real- 368.5 Not Real- istic istic istic 2025 843.7 Not Real- Not Real- 4760.6 Not Real- istic istic istic 270 Table 8.21. Comparison of Net Flow of Loan Funds under Alternative Scenarios Year Scenario Scenario Scenario Scenario Scenario One Four Five Six Seven ------------ Billion Dollars - - — - - - — - - 1978 10.1 10.2 10.1 9.9 10.1 1979 9.5 9.6 9.6 9.4. 9.4 1980 10.5 10.3 10.6 10.7 10.2 1981 12.3 11.8 12.2 12.9 11.8 1982 13.8 13.0 13.7 15.0 13.2 1983 14.9 13.6 14.5 16.6 13.9 1984 16.0 14.2 15.3 18.3. 14.6 1985 17.3 14.8 16.2 20.3 15.3 1990 25.9 19.1 22.1 34.5 _ 20.3 .2000 50.5 Not Real- Not Real- 91.2 _ Not Real— istic istic istic 2025 364.1 Not Real- Not Real- 1449.4 Not Real- istic istic istic 271 The projected net flow of loan funds (Table 8.21) is higher under scenario six than.under scenario one. This would be a result of many factors,but an important one would be the increased requirement for purchases of real estate from discontinuing proprietors (caused by the high real estate price). The nominal capital gains projected are more closely aligned in value with the gains in recent years, when inflation has exceeded 6 percent, than are the projections under 4 percent inflation. Zero (kowth in Livestock Inventory. The assumption of no growth in livestock inventory (scenario seven) has a large impact on the livestock and cr0p sectors. The major source of impact on the live- stock component is via the direct livestock demand reduction. The crops sector is hurt by the reduced feed demand (from the simultan- eous adjustments) and by the reduction in feed demand resulting from the livestock inventory being included in the feed demand equation (with a positive coefficient). The reduced real livestock and crop prices that result under this scenario lead the author to believe that ommission of the livestock inventory from the model would be a serious error. Summary This chapter began with an explanation of the methods of pro- jecting exogenous variables. Seven scenarios with minor variations in selected exogenous variables were developed. 272 Results from the projections were then discussed. Major findings include: (1) farm prices and associated income flows are sensitive to projections of crop production; (2) under moderate rates of inflation,leverage in the farming sector increases; (3) leverage increases less rapidly under higher inflation; (4) measures of the prospective burden of debt do not increase,even though lever- _age does; and (5) farm prices and incomes are sensitive to changes in real GNP, crOp export growth, and level of livestock inventory. CHAPTER IX Summary and Conclusions Introduction The purpose of the research reported here was to develop and test an aggregate farming sector economic projection model emphasizing financial aspects of the sector. The specific objectives were to: 1) Develop a theoretical model of the aggregate U.S. farming sector to provide a conceptual framework for estimation of an empirical model 2) Identify structural relationships among variables within the U.S. farming sector and the effects of variables exogenous to the sector through empirical investigation 3) Construct an operation aggregate economic projections model of the U.S. farming sector capable of making longerun pro- jections of alternative futures to provide input into public and private decision making The purpose of this chapter is to review the overall research effort, summarize some of the significant findings, discuss limitations of the model, and suggest areas in which further work would be useful. 273 274 The System to be Modeled An important dimension to developing an economic model is to identify the exact system to be modeled. This includes identification of the boundaries of the system, identifying variables as endogenous or exogenous, and determining performance variables. For the model developed here,it was determined that the system to be modeled would include all economic activities of farm operator families and the farming activities of nonOperator landlords. This defini- tion is a mixture of the product and establiShment concepts but was determined to be an optimal choice,given the data constraints. The performance variables of interest were determined to be best described as those appearing on financial statements of the aggregate farming sector. This includes flow variables such as those on the Income Statement and Sources and Uses of Funds Statement as well as stock vari- ables which appear on the Balance Sheet. Many of the financial variables such as noncash flows, intersector flows, or numerous financial ratios can be calculated using the basic data underlying the previously mentioned financial statements. While the major performance variables of interest relate to financial statements, these variables are largely transformationSIxfunder- lying prices and quantities resulting from economic activities. Thus .to construct a simulation model to project the financial variables of interest,the underlying economic structure needed to be modeled and then the equations transforming the economic variables had to be formulated. 275 The Simulation Model Modeling a system involves at least two levels of abstraction. First, a theoretical representation of the system must be constructed. This representation is based on "standard" economic theory in this study. ' Static, nonstochastic concepts of supply and demand with only minor modifications provided the basis for Specification of equations to be included in the model. A theoretical model is the first level of abstrac- tion from the real system. Secondly, an empirical model must be constructed. This is most appropriately viewed as parameterization of the theoretical model. The estimates of parameters for this model were derived using statistical techniques where possible. Some parameters were based upoh single obser- vations. Other parameters were based upon untested assumptions. The empirical model is the second level of abstraction. A third level of abstraction involves going from the empirical model to the computer model. The model deve10ped in this study is a discrete time approximation to continuous time processes. Other types of abstraction associated with computer models would involve numerical techniques, such as numerical integration or numerical solution to non- linear simultaneous equations. The researcher and policy maker should be careful not to forget that any model (and results) is at best an approximation to the "real" system and not, in fact, the real system. 276 Empirical Results The approach to modeling the underlying economic structure was an aggregate economic approach. The decision was made to stop short of aggregating the commodities produced in the farming sector into a single variable. The aggregation was made to two products:(1) aggregate crops and (2) aggregate livestock and livestock products. In an Open economy with storable commodities, production in a given year is not likely to equal consumption. Thus, there was a need for historical data providing aggregate supply and utilization of craps and livestock. Because of the simultaneous nature of the economic structure to be modeled.the data needed to be of the nature that addition and subtraction,such as required in the supply and utilization identities,would hold (e.g., the crops identity is: beginning inventory + production + imports = exports + feed + seed + food-industrial use + ending inventory). Data of the type required for this study were not available. Thus the author con- structed the data by using secondary supply and utilization data for individual commodities weighted by 1967 farm level prices for each commodity and then summed across the individual commodities. Therefore, the crOp and livestock aggregates in a particular year are the sum of individual commodity quantities in that year weighted by prices from the base year (1967). Based on these data, equations were estimated for: livestock supply, livestock demand, crap supply, feed demand, seed demand, food- industrial crOp demand, crop inventory demand, feed supply, and seed supply. These nine equations plus the two identity equations for crop and livestock supply and utilization are the structural equations for 277 the following eleven variables: livestock production, livestock consumption, crop production, feed use, seed use, food—industrial use of crops, inventory of crops, price of livestock, price of crops, price of feed, and price of seed. The theoretical model specified a set of equations for factors of production that were recursive to the above equations.- This included: three categories of nondurable inputs, based largely on categories re- quired for the Income Statement: (1) hired labor, (2) fertilizer and lime, and (3) other nondurable inputs. The third category was composed of gross investment demand and repair and maintenance demand for: machinery and motor vehicles; service buildings, other structures, and land improvements; and equations for real estate price and quantity transferred. A number of additional equations were estimated. They might be categorized as follows: equations intended to reproduce USDA procedures,l/ "near" identity equations converting prices and quantities into levels of receipts or expenses, time trend equations for relatively unimportant variables, and equations to project exogenous input prices (as a function of the CPI). Other components of the model, including intersector flows and farm Operator household activities, were specified. These were, in general, based on very little empirical data. -l/ Numerous data series published by the USDA have little empirical basis. Often the procedure is to estimate one data series based upon a benchmark and movements in another data series.’ Where this procedure was known to be used, an attempt was made to reproduce the procedure, since the author felt the data were not sufficient to support estimation of structural equations. 278 The theoretical specification of the investment demand equa- tion required data for stocks and scrappage of durables based upon a productive capacity concept. It was anticipated that published data based upon a value concept would not adequately represent the variables required to estimate the desired structural equations. Data were con- structed to represent the desired concepts. The construction was based in part on concepts which were not empirically tested. However, the USDA data based on the value concept have little, if any, greater empirical content. When used in regression equations to estimate the desired structural relationships,the coefficients of the constructed stock and scrappage variables were of reasonable statistical signifi- cance and of theoretically appropriate sign. These results were improve- ments over equations estimated (but not reported here) using the value concept measures of stocks and depreciation,as well as improvements over results from previous research using the value concept data. Evaluation of the Model The first method of valuating the model would be to evaluate each equation individually. Statistical criteria such as "goodness of fit" measures and the statistical significance of individual coefficients are useful in this regard. Further criteria such as correspondence of coefficient signs to the theoretical expectations and elasticities are used when evaluating individual equations. These criteria are presented in earlier chapters for the equations estimated. When there are simultaneous (or recursive) equations and/or lagged endogenous variables in a model,the single equation criteria 279 are not sufficientl/to support conclusions about the models ability to forecast. Further evidence considereduseful relates to the model's ability,as a set of equations, to forecastlxnfliwithin the sample period and beyond the sample period. Forecasting within the sample period can mean (1) using actual values for lagged endogenous variables or (2) using forecasted values for lagged endogenous variables (after the initial period). Mbdels that are to be used for one period forecasts are normally evaluated using actual values for lagged endogenous variables. The model developed here. is to be used for projecting longer periods and thus was evaluated using forecasted values for lagged endogenous variables. The results from . simulating a set of the important structural equations over the entire sample period have been presented in Chapter VI. The results indicate that the model tracks quantities reasonably well. The model also tracked the overall level of prices well but had fairly large errors in some years. An ex post simulation, forecasting beyond the sample period using actual values of exogenous variables, was performed for the years 1975, 1976, and 1977. The model forecasted significantly greater crop prices than occurred in 1975 and 1976. These errors were traced to overestimation of feed demand for crops in 1975 and underestimation of cr0p supply in 1976. In addition,the real estate price equation underestimated significantly. Adjustments were made for these factors when making the projections reported in Chapter VIII. 1/ In general, however, the single equation criteria are normally considered necessary. 280 A significant (and disturbing) error is forecasted by the model for the 1977 net flow of loan funds (underestimated). Based on prelim— inary USDA data,it was determined that the model error came in the areas of consumption of nondurable goods and services and other uses of cash. It appears that these uses of funds did not decrease in 1977 as the relationships in the model suggest. It should be noted that these rela- tionships are based on little empirical data. Both categories are functions of current net farm income. Thus it appears that neither con- sumption nor investments in off-farm capital, to the extent that other uses of cash reflect investment in off—farm capital, have fallen sig— nificantly in response to falling net farm income. Projections A number of scenarios was generated based on alternative assumptions with respect to important exogenous variables. Exogenous variables in this category included the rate of inflation, the rate of growth in crOp exports, the level and rate of growth in crop produc- tion, the rate of growth in real gross national product, and the rate of growth in livestock inventory. The projections for the later 1970's and early 1980’s provedtx>be sensitive to the level of crOp production. The model appears to reflect the intuitively appealing behavior that "boom or bust" in the farming sector can be induced via events in the export market or in domestic cr0p production that are within the reasonable range of outcomes. 281 The base projections indicate an increasing real net farm income (from a fairly low value in 1978). This was largely the result of increasing real livestock prices (the real price of crops was nearly constant). Nominal capital gains were projected to increase from 1980. with a very small increase in deflated nominal capital gains. The net flow of loan funds to the sector is projected to increase throughout the 1980’s The increase in liabilities is sufficient to re- sult in an increasing debt/equity ratio. While the increase is fairly dramatic, several ratios (of the net flow of loan funds to cash flow or income indicators) intended to be indicators of the prospective bur- den of debt are projected to improve over the 19803. Net capital formation of the sector (business and household) is projected to berumu:zero if one subtracts an estimate of net real estate transfers to other sectors. However, net capital formation of continu- ing proprietors, which includes purchases of equity capital from discontinuing proprietors, is projected at a fairly high and growing level. Projections under the assumption of low growth in real gross market product result in a nearly constant projection of the real price of livestock and a rapidly falling real price of craps. This scenario results in significantly lower projected net farm income and capital gains than in the base scenario. Projections under the assumption of no growth in livestock inventory result in a reduced rate of increase in real price of live- stock and a declining real price of crops. As in the case of low growth in real GNP, net farm income and capital gains are signifi- cantly reduced when compared. with the base scenario. 282 Projecting the rate of inflation at a higher level (6 percent versus 4 percent) has interesting results. While there is very little change in the projected real prices of farm output and only_ a small increase in real net farm income, there is a large increase in deflated capital gains. This is a result of the specification of the real estate price equation. The increased rate of gain in equity from capital gains significantly reduces the increase in the debt/equity (leverage) ratio. Focus of Research The focus of the research project reported here has largely been oriented toward putting together data and estimating a set of equations to des- cribe the economic and financial activities of the aggregate farming sector. In the process of doing this, the day-to-day concerns of how historical data series are constructed, "t" statistics, variance- covariance matrices, predicted versus actual resuts, etc..have swamped the author's thinking. While there has been significant thinking about the usefulness of the proposed effort andintermittent reflection on the product-establishment nature of accounts and other items of more general nature, there has not been a really concentrated and produc- tive effort toward determining what information can be extracted from the accounts. Additional Research While there are many "detail" problems with the model that could use additional work, very significant work.needs to be done toward analysis of financial accounts. 283 Initial thoughts would be to explore further the usefulness and/or meaning of savings and capital formation and relate changes in these over the historical data to changes anticipated\nuiprojections. In addition, it seems that if the accounts, especially net farm income and capital gains, are to be used to make welfare inferences that capital formation and savings should be calculated for continuing farm prOprietors rather than for the sector (at least for welfare inferences if not for all purposes). This leads to the need for a more complete analysis of the difference between accounts prepared for continuing prOprietors and for the sector. Policy Analysis The usefulness of a model for direct policy analysis depends in part on the model's ability to handle policy scenarios (e.g., price floorsanulceilings, storage policies and related payments to farmers, or monetary policy). Many policies require programming of the model beyond the minimum required to get basic projections. Little of this programming has been completed to date but needs to be done to improve the usefulness of the model. Model Improvements The model could be improved greatly through the incorporation of additional data or better information on intersector flows and household activities. The relationships used in the model are based largely on empirically untested assumptions in these areas. Other improvements could be made in the model. Some of these are described in the following sections. 284 Livestock Capital Stock 'The model currently incorporates no information concerning the capital portion of the livestock inventory. It is the impression of the author that while a great deal of data is not available, there are some data available such that reasonable assumptions could be incorporated into the model. CrOp Price Currently,cr0p price is the unweighted season average. Errors in the crap identity and estimates of crop price coefficients might be improved if the crop price were weighted by quantities marketed. Use of calendar year rather than crop year price should be explored. Supply-Utilization Aggregates The implications of different price weights for the supply utilization data have not been explored. In addition, the incorporation of additional data on important excluded commodities such as citrus and noncitrus fruits might improve the data. Productive Durables Construction of "quantity" stocks and scrappage of durable assets for use in demand equations would benefit from incorporation of empir- ical information on length of life and survival rates. Lack of real investment data for some categories of durables prevented estimation of structural equations. In addition, empirical data on scrappage of durables would allow greater detail in estimation of investment equations. 285 Prices of Farm Inputs As is typical of the approach of agricultural economists, the model has assumed that input prices are exogenous (supply perfectly elastic). Clearly,the supply function5(fl=fertilizerznuifarnImachinerylunua some simultaneous effects,answitnessedtnrincreased machinery prices resulting from excess demand in recent years and the lack of substitutes or substitute uses for fertilizer. Incorporation of supply equations for farm inputs would improve the model, provided that:n1additional.large set of exogenous variables doesrun:accompany endogenizing input prices. Further Model Evaluation The dynamic properties of the model have not been explored to the extent desired. Interim and dynamic multipliers for the model should ~ be derived. The basic linear nature of the model should make it feasible to derive these analytically. Related analysis would be to investi- gate mathematically the stability of the model. This would involve investigation of the eigenvalues of the model. APPENDICES APPENDIX A Construction of Aggregate Commodity Supply Utilization Data Construction of Commodity Aggregates The view taken in this study is that it is reasonable to aggre- gate output of the agricultural sector into two categories, crops and livestock. These categories are based upon supply and demand simi- larities and demand for input considerations. The breakdown into two categories, while only one step away from viewing the sector as having one output, offers more flexibility and provides greater intuitive appeal than the single output approach. On the demand for output side, there are numerous and signifi- cant differences between crops and livestock. With respect to livestock, imports and exports are relatively insignificant,and demand is derived fairly directly from final consumer demand. Crop demand, on the other hand, is quite different. Feed demand, export demand, and direct con- sumer and industrial demand are all important. In consideration of the workability of this breakdown in a simulation model, there are signifi-- cant possibilities in terms of policies that are often applied to the cr0p sector alone (e.g., storage, price, and export policies). On the output side, the effect of technology and weather appear to have greater direct effect upon the crop sector; the effects are often indirect in the livestock sector (e.g., through feed price). In addition, policies such as acreage controls or scenarios 286 287 on variables such as fertilizer prices can be applied in a realistic and directmanner. With respect to the demand for inputs, there are some inputs such as seed and fertilizer which are directly related to crops, and there are others such as feed demand which are directly related to livestock. To develop a fairly detailed model of the supply and demand for farm commodities in which prices and quantities are simultaneously determined, one needs a set of quantity data for which the standard identities hold (i.e., for crops: production + beginning inventory + imports = exports + feed use + seed use + food and industrial use + ending inventory). The author was not able to find published aggregate data in this detail. Thus, the effort to construct the desired data is described in the following sections. The basic approach used to aggregate the various farm commodities was to weight the supply-utilization quantities (from published sources) of important commodities by 1967 farm level prices. Thus, the common denominator used to aggregate is 1967 farm value (million dollars). Baker (1978) contains the individual commodity data used in the construction of aggregate data for the calendar years 1950 to 1977. LiVestock and Livestock Products The individual commodity data for livestock and livestock products did not take changes of inventory into account. At these sources, production was basically livestock slaughter, eggs produced, and miflrand milk products produced. Consumption was calculated by ~subtracting net exports from production. 288 The beginning and ending stocks for the livestock aggregate werebased upon the index of livestock and poultry farms (Agricultural Statistics.l975, Table 501).l967 = 100, and the 1967 balance sheet value of livestock on farms (Balance Sheet of the Farming Sector.l976, Table 40). Inventories were calculated by multiplying the 1967 value of livestock by the index of livestock on farms. The change in inventory (ending - beginning) was added to the summation of the individual commodity produc- tion. weighted by 1967 farm prices, to get livestock production for use in the model as the quantity-supplied variable. Thus, the aggregate production data includes livestock produced and held on farms,as well as livestock and products marketed. Table A.l shows the livestock commodities included, price weights used, and source of data. Crops Nearly all of the individual crop data were derived from sources with complete supply-utilization data on a crop year basis. Exceptions are noted below. Corn and sorghum silage production was assumed to equal feed use in each year. Data were not available on inventory changes. Com— plete data were not available for vegetables; thus, production was assumed to equal food use. Minor errors are present in the soybeans and soybean meal data because of differences in bushels of soybeans crushed in the Supply and disappearance of soybeans (Agricultural Statistics.I 1975, Table 172). and the supply and disppearance of soybean oil and meal (Agricultural Statistics,l975, Table 173). 289 Table A.2 shows the price weights used, commodities included, and the sources of data used in construction of the crop aggregates. ‘Qggg The aggregate supply and utilization data constructed and used in the development of the model are shOwn in Tables A.3 and A.4. The data have been updated and extended as far as possible beyond 1974, the last year used in the estimation. 290 mac manna .Nmma muuoexm was muuoaaH owm.man6e .memfl.muaumaumum Hatsuasuauw< coauuseonm mama. .Non wwwm moo manmy .Numa mum magma .mmquWMwumfiumum Housuasofiuww. coauUSpoum owmm. .u3>H .mna moxusa omm magma .Nmma omm manna .mmma.moaumfiumum HousuHSUHnwm. sowuosvoum Omna. .u3>a .mnH cmxofico mnma How usoEoHQQSm mam mmm .oz .sfiuoaasm Hmoflumflumum «am: muuoexm «mm manme .Nnma «ma mHan .mnma muuanH qu memH .Nmma mes magma .memeqmuaamaumum Hmtsuesoauwm, coaauseoue Heme. .us>a .moe .Icouuss \H was Lama mnma you usmEmHmadm can mmm .oz :Huwaasm Hmowumfiumum H .mnH \mamm> use «mom mMUMDOm «949 m MUHmm mHHZD wHHQozzOo mmumwmwww< xuouwm>flq mo sofluosuumsoo s“ new: mmousom mama use .muswfimx ooaum .mmauwuanoo A.4 manna 291 .umm% some ca xooumm>aa mo mmmao sumo mom ammunoouma wswmmouv man hp soamfl>aw mw> munwwmsm>wa ou vmuum>soo mums muanmB mmmoumo .munwfims mmmoumo CH whoa sumo Hmcawwuo uuomxo use usomEH \w muuoaxm muHonH msomeP .cowumSuHmIwamm ase6 massacre sea: Nam magma .meme.mwaumaumam Hmnsuflsuauwa, soauuseoum ea.q seas .u36 can see: mmoMDOm oz.xem~m3v coaamsuam 066:2 eema.muaumaumum Hmusuenuaewa 66: 666m om.H Hmamsn 866:3 mmumsom om om magma .agma om magma .Nnma Om maan .mmma vmmm wma manna .moma mma mHQMH .Nmma Nna manna .mmma.moaumaumum amusuasoauwfl muuoexm .mxooum \Mm.ow msou Hmmz smmpmom mmumsom 3mm .amwSm uomn can mcmu \m .NnIHH momma .mammn unwamz moamm mamaamm \w .mammn xUOum awaamm mamscm mm.H mmsau manna amsamx umnu manmsoaumama mnu wsam: owns ma Amwam> moa> aov mammp xuoum aosamw ou manna Hmsamx anw noamam>soo .mammn Hmsamx \m _.wmuMum moaam mnu mo unmoaua mom.m~ wsam: ca wmuasmma mafia .Hmnmsn awn aao mo mussom m.oa use Home mo mwssoe m.n< mo soauaasmmm man so mums mma usoEumSncm mnH .Aao.va mcmwnhom mom ouaaa Show mnu ou unmam>asvw woaam wmnaaaoo m new on mam: vmumum mmonu 80am noumsncm oam3 mmoaam Hams use Hao ammnxom \M .mammn unwam>asvo .nwsom \w HNN magma .Nama ama manna .mmma moaumaumum ama:uadoaam< mama HH< o.mo msou moanmumwm> ma magma .mama mom mamamammsm mam mma .oz uaoamm moaBoCoum Hmasuasowams mam . 0; thus, the confidence intervals calculated (or significance level) indicate a "better" fit than actually exists. A widely used test for first order autoregression is the Durbin-Watson test. The test statistic (d) is reported for equations 303 in the model. For testing the hypothesis that p = 0 versus the alterna- tive that p > 0, the decision rules are as follows: 1) Reject if d < dL- 2) Do not reject if d > dU.- 3) The test is inconclusive if dL < d < dU- The values of dL (lower limit) and Du (upper limit) are given in the table provided by Durbin and Watson and are reproduced in numerous econometrics and statistics textbooks. The problem of autoregressive disturbances is complicated by the inclusion of lagged endogenous variables in the regression equation. This, in itself, can be shown theoretically to induce autoregressive disturbances. The Durbin-Watson statistic does not provide an un- biased estimate when endogenous variables are included as explanatory variables (either lagged endogenous or current and/or lagged endoge- nous variables in the case of simultaneous equations). However, be- 'cause of popular usage, the values of the Durbin-Watson statistic have been reported in this study for equations that are simultaneous, as well as for those with lagged endogenous variables. The reader should be aware that they are not reported to provide a legitimate . statistical test for first order autoregression. The often used Cochrane—Orcutt iterative procedure for correct- ing first order autoregression does not provide unbiased estimates of the correlation coefficient when there are lagged endogenous vari- ables. In addition, the procedure does not necessarily "purge" first order autoregression but only adjusts the data until one cannot tell if there is first order autoregression. 304 In general, the equations in the modelhave been estimated without correction for first order autocorrelation. No structural equations estimated via OLS have lagged endogenous variables; thus, the Durbin-Watson is a legitimate test for these equations. Among these equations,only hired labor demand has clear evidence of a problem. In this case, the author is much more comfortable with the unbiased but inefficient OLS estimate than with the estimates generated by the Cochrane-Orcutt procedure. Two of the simultaneous equations include lagged endogenous variables.~ The unavailability of an unbiased simultaneous equations estimator at the time these equations were estimated prevented proper treatment. No proper statistical test is used to test for first order autoregression for any of the simultaneous equations. One final comment relative to testing for autoregressive dis- turbances is that the above discussion, and most of the discussion in the literature, concentrates only on first order autoregression. The only legitimization for this fact that the author is aware of is that one often deals with annual observations. This fact would not seem to prevent the possibly frequent occurence of higher order autoregression. BIBLIOGRAPHY BIBLIOGRAPHY American Agricultural Economics Association Committee on Economic Statistics. American Journal of Agricultural Economics, 54:867-880, December 1960. Baker, Timothy C. "An Economic and Financial Projections Model of the U.S. Farming Sector: Supporting Data and Tables." Agricultural Economics Staff Paper 78-33, Department of Agricultural Economics, Michigan State University, 1978. Basmann, R. L. "On Finite Sample Distribution of Generalized Clas— sical Linear Identifiability Test Statistic." Journal of the American Statistical Association, 55:650-659, December 1960. Basmann, R. L. "Letter to the Editor." Econometrica 30:824-826, October 1962. Bhatia, Kul B. "The USDA Series on Net Investment in Farm Real Estate-A Critique." Journal of the American Statistical Association, 66:492-495, September 1971. Bhatia, Kul B. "On Estimating Capital Gains in U.S. Agriculture." American Journal of Agricultural Economics, 53:502-506, 1971. Binswanger, Hans P. "A Cost Function Approach to the Measurement of Elasticities of Factor Demand and Elasticities of Substitution." American Journal of Agricultural Economics, 56:377-386, May 1974. Bitros, George C. "A Statistical Theory of Expenditures in Capital Maintenance and Repair." Journal of Political Economy, 84:917—936, October 1976. - Bonnen.James T. "Improving Information on Agriculture and Rural Life." American Journal of Agricultural Economics,57:753-763, Decemberl975. Box, G. E. P. and G. M. Jenkins. Time Series Analysis Forecasting and Control, Revised Edition. San Francisco: Holden Day, 1976. Boyne, David H. "Cahnges in the Real Wealth Position of Farm Operators." Michigan State University EXperiment Station Bulletin No. 294, 1964. Bradford, Lawrence A. and Glenn L. Johnson. Farm Management Analysis. New York: John Wiley and Sons, Inc., 1953. 305 306 Brake, John R. "Capitalizing Agriculture in Coming Years)‘ In Emerging and Projected Trends Likely to Influence the Structure of Midwest Agriculture 1970—1985, ed. John R. Brake, Agricultural Law Center Monograph 11, University of Iowa, June 1970, pp. 28-54._ Brake, John R. "Impact of Structural Changes on Capital and Credit Needs" Journal of Farm Economics, 48: 1536-1545, December 1966. Brake, John R. "Future Capital and Creidt Needs of Canadian Agri- cultureJ'Dept. of Agricultural Economics, Publication AE 70/3, University of Guelph, 1970. Brake, John R. and Peter L. Barry "Flow-Of—Funds Social Accounts for the Farm Sector:CommentP American Journal of Agricultural Economics, 53:665-668, November 1971. Brake, John R. and Emanuel Melichan "Agricultural Finance and Capital MarketsJ'In A survey of Agricultural Economics Literature Volume 1: Traditional Fields of Agricultural Economics, 19403 to 19703, edited by Lee R. Martin, 1977. Carlin, Thomas A. and Charles R. Handy;"Concepts of the Agricul- tural Economy'and Economic AccountingJ'American Journal of Agri- cultural Economics, 56:964-975, December 1974. Carlin, Thomas A. and Allen G. Smith."A new Approach in Accounting for Our Nation's Farm IncomeJ'Agricultural Finance Review, 34:1-6, 1973 Cochrane, W. W. "Conceptualizing the Supply Relation in Agricul- tureJ'Journal of Farm Economics, 37:1161-1176, December 1955. Coen, R. M. "The Effect of Cash Flow on the Speed of Adjustment." In Tax Incentive and Capital Spending, ed. by G. Fromm, Washington: The Brookings Institution, 1971. Coen, R. M. "Tax Policy and Investment Behavior: Comment." American Economic Review, 59:370-379, 1969. Cromarty, W. A. "The Farm Demand for Tractors, Machinery, and TrucksJ'Journal of Farm Economics, 41:323-331, May 1959. Day, Richard H. "Discussion: From-Stock to Flow Capital Inputs for Agricultural Production Functions: A Micronalytic Approach." Journal of Farm Economics, 49:491-495, 1967, Economic Report of the President Transmitted to Congress, Washington, January 1975. Eisner, Robert, "Investment: Fact and Fancy? American Economic Review, 59:237-246, 1969. 307 Eisner, Robert."Tax Policy and Investment Behavior: Comment." American Economic Review, 59:379-388, 1969. Fair, Ray C. "The Estimation of Simultaneous Equation Models with Lagged Endogenous Variables and First Order Serially Correlated ErrorsJ'Econometrica, 38:504—516, May 1970. Feldstein, Martin S. and Michael Rothschild."Towards An Economic Theory of Replacement Investment" Econometrica, 42:393-423, May 1974. ‘Goldberger, Arthur S. Econometric Theory. New York: John Wiley and Sons, Inc., 1964. Griliches, Zvi."The Demand for a Durable Input: U.S. Farm Trac- tors, 1921-57J'In The Demand for Durable Goods, ed. by A. C. Harberger, Chicago: University of Chicago Press, 1960. Griliches, Zvi. "Capital Stock and Investment Functions: Some Prob— lems of Concept and Measurement." In Measurement in Economics, ed. by Christ, et al., Stanford: Stanford University Press, 1962. Griliches, Zvi. "The Demand for Inputs in Agriculture and a Derived Supply Elasticity." Journal of Farm Economics, 41:309-322, May 1959. Griliches, Zvi. "Estimates of the Aggregate U.S. Farm Supply Function.’ Journal of Farm Economics, 42:282-293, May 1960. Grove, Ernest W. "Econometricians and the Data Gap: Comment." American Journal of Agricultural Economics, 51:184-188, February 1969. Grove, Ernest W. "Econometricians and the Data Gap: Rejoinder." American Journal of Agricultural Economics, 53:366-367, May 1971. Grove, Ernest W. "Farm Capital Gains—-A Supplement to Farm Income?” Agricultural Economic Research, 12:37-42, 1969 Hall, Robert E. and Dale W. Jorgenson. "Tax Policy and Investment Behavior.“ American Economic Review, 57:391-414, 1967. Hall, Robert E. "Tax Policy and Investment Behavior: Reply and Further Results." American Econmic Review, 59:388-401, 1969. Heady, Earl O. and Luther G. Tweeten. Resource Demand and Structure of the Agricultural Industry. Iowa State University Press, Ames, Iowa, 1963. Hendershott, Patric H. "A Flow of Funds Model: Estimates for the Nonbank Finance Sector." Journal of Money, Credit, and Banking, November 1971. w 308 Hendershott, Patric H. and Richard C. Lemmon. "A Flow of Funds Model: First Estimates and Forecasts." 1973 Business and Economic Statistics 'Section Proceedings of the American Statistical Association. Herdt, Robert W. and Willard W. Cochrane. "Farm Land Prices and Farm Technological Advance." Journal of Farm Economics, 48:243—263, May 1966. Hoover, Dale E. "The Measurement and Importance of Real Capital Gains in U.S. Agriculture, 1940—1959.” Journal of Farm Economics, 44:929-940, 1962. Hopkins, John A., Peter J. Barry, and C. B. Baker. Financial Management in Agriculture. Danville, Illinois: Interstate Printers and Publishers, Inc., 1973. Johnson,IL Gale. "The Nature of the Supply Function for Agricultural. Products." American Economic Review, 40:539-564, September 1950. Johnson, D. Gale. "Agricultural Credit, Capital, and Credit Policy in the United States." Federal Credit Programs, Englewood Cliffs, New Jersey: Prentice-Hall, 1963, pp. 355-423. Johnson, Glenn L. Managerial Concepts for Agriculturalists. Kentucky Experiment Station Bulletin 619, 1954. Johnson, Glenn L., et al., editors. A Study of Managerial Processes of Midwestern Farmers. Iowa City: The Iowa State University Press, 1961. Jorgenson, D. W. "Capital Theory and Investment Behavior." American Economic Review, 59:247-259, 1969. Kmenta, Jan. Elements of Econometrics. New York: The MacMillan Company, 1971. Koopmans, T. C. and W. C. Hood. "The Estimation of Simultaneous Linear Economic Relationships." In Studies in Econometric Methods, New York: John Wiley, 1953. Kuh, Edwin. "Theory and Institutions in the Study of Investment Behavior." American Economic Review, 59:260~268, 1969. Lins,°David A. A Simulation Model of Farm Sector Social Accounts with Projections to 1980. USDA Technical Bulletin No. 1486, December 1973. Lins, David A. "An Analysis of Sources and Uses of Funds in the Farm Sector of the United States." Unpublished Ph.D. thesis, University of Illinois, 1972. Loeb, Peter D. "Specification Error Tests and Investment Functions." Econometrics. 44:185—194, 1976. 309 Manetsch, Thomas J. and Gerald L. Park, Systems Analysis and Simula- tion With Applications to Economic and Social Systems: Parts I and ll” Department of Electrical Engineering and Systems Science, Michigan State University, Second Preliminary Edition, August 1974. ' Melichar, Emanual, "Discussion: Capital Finance and Capital Flows Accounts." In A Proceedings of the Workshop on Farm Sector Financial Accounts, Economic Research Service, forthcoming. Melichar, Emanuel. "Financing Agriculture: Demand for and Supply of Farm Capital and Credit." American Journal of Agricultural Economics, 55:313-325, May 1973. rMelichar, Emanuel. "Aggregate Farm Capital and Credit Flows Since 1950 and Projections to 1980." Agricultural Finance Review, 33:1-7, July 1972. Melichar, Emanuel and Marian Sayre. "Capital Gains in the U.S. Farming Sector, Nominal and Real, 1940-1974." Contributed paper prepared for the 1975 AAEA Annual Meetings, The Ohio State University, Columbus, Ohio, August 10-13, 1975. Melichar, Emanual, Harriet Holderness,and Marian Sayre."Agricul- .tural Finance Databook, Annual Series." Division of Research and Statistics, Board of Governors of the Federal Reserve System, February 1976. Nelson, Frederick J. "An Economic Analysis of the Impact of Past Farm Programs on Livestock and Crop Prices, Production, and Re- source Adjustments." Unpublished Ph.D. thesis, University of Minnesota, 1975. Nerlove, M. "On Lags in Economic Behavior." Econometrica, 40:221-251, 1972. O'Dell, Charles A. "Econometricians and the Data Gap: Reply— Comment." American Journal of Agricultural Economics, 51:679-680, August 1969. ' Ortega, J. M. Numerical Analysis: A Second Course. New York: Academic Press, 1972. Penson, John B. Jr. "Toward an Aggregative Measure of Saving and Capital Finance for U.S. Farm Operator Families." American Journal of Agricultural Economics, 59:49-60, February 1977. Penson, John B. Jr. "An Aggregative Income and Wealth Model for the U.S. Farm Sector: Its Description and Application to Policy Analysis." Unpublished Ph.D. theSis, University of Illinois, 1973. 310 Penson, John B. Jr. "Demand for Financial Assets in the Farm Sec- tor: A Portfolio Balance Approach." American Journal of Agricul- tural Economics, 54:163-173, May 1972. Penson, John B. Jr.,David A. Lins,and George D. Irwin. "Flow of Funds Social Accounts for the Farm Sector." American Journal of Agricultural Economics, 53:1—7, February 1971. Penson, John B. Jr., Dean W. Hughes,and Glenn L. Nelson. "Measure- ment of Capacity Depreciation Based on Engineering Data." American Journal of Agricultural Economics, 59:321-329, May 1977. Pindyck, Robert S. and Daniel L. Rubinfeld. Econometric Models and Economic Forecasts New York: Mc—Graw Hill, 1976. Plato, Gerald. "Agricultural Commodity Projections: A Multi-Market Equilibrium Approach." Contributed paper prepared for the 1975 American Agricultural Economics Association, The Ohio State Univer— sity, Columbus, Ohio, August 10-13, 1975. Ramsey, James B. "Tests for Specification Error in Classical Linear Least-Squares Regression Analysis." Journal of the Royal Statistical Society, Series B, 31:350-371, 1969. Ramsey, James B. "Models, Specification Error, and Inference: A Discussion of Some Problems in Econometric Methodology." Bulletin of the Oxford Institute of Economics and Statistics, 32:301-318, 1970. Ramsey, James B. "Classical Model Selection Through Specification Error Tests." In Frontiers in Econometrics, ed. by Paul Zarembka, New York: Academic Press, 1974. ‘ Ray, Daryll E. and Earl 0. Heady. "Government Farm Programs and Commodity Interaction: A Simulation Analysis." American Journal of Agricultural Economics, 54:578-590, November 1972. Reinsel, Robert D. "The Aggregate Real Estate Market." Unpublished Ph.D. thesis, Michigan State University. Schuh, G. E. "An Econometric Investigation of the Market for Hired Labor in Agriculture." Journal of Farm Economics, 44:307-321, May 1962. Schultz, T. W. "Reflections on Agricultural Production, Output, and Supply." Journal of Farm Economics, 38:748-762, August 1956. Scott, John T. Jr. and Earl O. Heady. "Regional Demand for Farm Buildings in the United States." Journal of Farm Economics, 49:184- 198, February 1967. 311 Scott, John T. Jr. and Earl 0. Heady. "Econometricians and the Data Gap: Reply." American Journal of Agricultural Economics, 51:188, February 1969. Scott, John T. Jr. and Earl O. Heady. "Econometricians and the Data Gap: Reply to a Comment." American Journal of Agricultural Econom- ics, 53:101-102, February 1971. Simunek, Richard W. "National Farm Capital Accounts." American Journal of Agricultural Economics, 58:532-542, August 1976. Theil, Henri. Principles of Econometrics. New York: John Wiley & Sons, Inc., 1971. Tomek, William G. "R2 in TSLS and GLS Estimation." American Journal of Agricultural Economics, 55:670, November 1973. Tostlebe, Alvin 8. Capital in Agriculture: Its Formation and Financing Since 1870. Princeton University Press, 1957. Tweeten, Luther G. and James E. Martin. "A Methodology for Predict- ing U. S. Farm Real Estate Price Variation." Journal of Farm Economr ics, 48:378-393, May 1966. Tweeten, Luther G. and C. L. Quance. "Positivistic Measures of Aggregate Supply Elasticities: Some New Approaches." American Journal of Agricultural Economics, 51:342-353, May 1969. Tweeten, Luther G. and C. L. Quance. "On Statistical Intolerance In Supply Analysis." American Journal of Agricultural Economics, 53:675-677, November 1971. Tweeten, Luther G. and C. Leroy Quance. "Positivistic Measures of Aggregate Supply Elasticities: Some New Approaches." American Economic Review, 59:175-183, 1969.. Tyner, Fred H. and Luther G. Tweeten. "Simulation as a Method of Appraising Farm Programs." American Journal of Agricultural Econom- ics, 50:66-81, February 1968. Tyrchniewicz, Edward W. and G. Edward Schuh. "Econometric Analysis of the Agricultural Labor Market." American Journal of Agricultural Economics, 51:770-787, November 1969. U.S. Department of Commerce, Bureau of the Census. 1969 Census of Agriculture. Farm Finance, Vol. 5, Part 2, August 1974. U.S. Department of Commerce and U.S. Department of Agriculture. 1972 OBERS Projections: Regional Activigy in the U.S. Series E Population, Vol. 1, Concepts, Methodology and Summary Data, Economic Research Service and Bureau of Economic Analysis, April 1974. 312 U.S. Department of Agriculture. 1972 OBERS Projections: Regional Economic Activity in the U.S. Series E' Population Supplement. Agricultural Projections, Vols. 1, 2,and 4, ERS, May 1975. U.S. Department of Agriculture. Majgr Statistical Series of the U.S. Department of Agriculture. Agriculture Handbook No. 365, 11 volumes. U.S. Department of Agriculture. Supplement for 1959 to Measuring the Supply and Utilization of Farm Commodities. Agricultural Marketing Service, Supplement for 1969 to Agriculture Handbook No. 91, September 1960. U.S. Department of Agriculture. Proceedingg of the Workshop on Farm Sector Financial Accounts. Economic Research Service, forthcoming. 'U.S. Department of Agriculture. Food Consumption Prices Expenditures. Economic Research Service, Agricultural Economics Report No. 138, July 1968. U.S. Department of Agriculture. Food Consumption Prices Expenditures. Economic Research Service, Supplement for 1974 to Agricultural Economic Report No. 138, January 1976. U.S. Department of Agriculture. Balance Sheet of the FarminggSector. Economic Research Service,Agricultural Information Bulletin 389, September 1975. U.S. Department of Agriculture. Farm Income Statistics. Economic Re- search Service, Statistical Bulletin No. 547, July 1975. U.S. Department of Agriculture. 1975 Changes in Farm Production and Efficiency, A Summary Report. Economic Research Service, Statistical Bulletin No. 548, September 1975. U.S. Department of Agriculture. Changes in Farm Production and Effi- ciency: A Special Issue Featurinngistorical Series. Economic Re- search Service, Statistical Bulletin No. 561, September 1976. U.S. Department of Agriculture. Agricultural Finance Outlook. Economic Research Service, APO-16, November 1975. U.S. Department of Agriculture. Agricultural Finance Statistics. Economic Research Service, AFS—2, June 1974. U.S. Department of Agriculture. National Food Situation. Economic Research Service, NSF-156, May 1976. U.S. Department of Agriculture. Farm Real Estate Market Developmentg, Economic Research Service, CD-8l, July 1976. 313 U.S. Department of Agriculture. Farm Real Estate Market Developments- Economic Research Service, CD-80, July 1975. ' .U.S. Department of Agriculture. Farm Real Estate Market DevelOpments. Economic Research Service, CD-7l, December 1968. U.S. Department of Agriculture, Farm Real Estate Market DeVelopments. Economic Research Service, CD-67, August 1965. U.S. Department of Agriculture. Agricultural Prices Annual Summary 1974. Crop Reporting Board, Statistical Reporting Service, Prl-3(75), June 1975. U.S. Department of Agriculture. Farm Labor. Statistical Reporting Service, numerous reports. U.S. Department of Agriculture. Farm-Operator Family Living Expendi- tures for 1973. Statistical Reporting Service, SpSy6(9-75), September 1975. U.S. Department of Agriculture. Agricultural Statistics. U.S. Govern- ment Printing Office, Washington, 1976. ‘ U.S. Department of Agriculture. Agricultural Statistics. U.S. Govern- ment Printing Office, Washington, 1975. U.S. Department of Agriculture, Agricultural Statistics. U.S. Govern- ment Printing Office, Washington, 1972. U.S. Department of Agriculture, Agricultural Statistics, U.S. Govern- ment Printing Office, Washington, 1967. Van Horne, James C. Financial Management and Policy, Second Edition, Englewood Cliffs, New Jersey: Prentice-Hall, 1971. Weeks, Eldon E. and others. "Farm Income and Capital Accounting- Findings and Recommendations of a 1972 ERS Task Force." Unpublished report, Economic Research Service, U.S. Department of Agriculture, July 1972. Yeh, Chung J. "Simulating Farm Output, Prices, and Income Under Alternative Sets of Demand and Supply Parameters." Contributed paper prepared for the 1975 American Agricultural Economics Association Annual Meetings, The Ohio State University, Columbus, Ohio, August 10-13, 1975. Yeh, Chung J. "Prices, Farm Outputs, and Income Projections Under Alternative Assumed Demand and Supply Conditions." American Journal of Agricultural Economics, 58:703-711, November 1976. HICHIGRN STQTE UNIV. szIz LIBRQRIE 5IIIIII I III - - - - - - I