MEASURING UTILITY 0F WEALTH AMONG" FARM MANAGERS r Thesis for The Degree of. Ph. D. ‘ MICHIGAN STATE UNIVERSITY Albert Nelson Halter 1956 TH :2“! This is to certify that the thesis entitled p IIeasuring‘Utuity of Wealth Among Farm Managers presented by ". Albert Nelson Halter has been accepted towards fulfillment of the requirements for Agricultural L degree in _.1_— Economics w“ W Major yoiessor Date October 25, 1956 0-169 CAM I375 2:” .‘vl - ..:»' LEASIBIIIG UTEITY or mum more MIL-1 1-m».TA(}3;gs By Albert Nelson Halter ’Ik. A THESIS ' Submitted to the College of Advanced Graduate Studies of hichigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics LIB RA R Y Michigan State Uniwrmty 9" ‘91}?m‘ ‘i"" "1 . ’1 A .‘7- 7-- LhnbflLflUunhhTS The author desires to express his appreciation for the encouragement and guidance received from Dr. Glenn L. Johnson during the course of this study. Thanks are due to Dr. L. L. soger for his part in providing financial assistance and to the members of the author's graduate committee for tLeir stimulating suggestions during the program of graduate work. Special thanks to fellow graduate students who helped get the thesis in final form and to W. A. Cromarty for assistance in meetin" tne deadlines. L) The author also wishes to thank D.J.I. for her enduring Spirit during tne writing of the thesis. \I ‘l \l \I \I \I \I \I \I \I \' - c. _. .. .— .7 -_ _. ‘_ c» —_ c I\ l\ /\ l\ l\ I\ I‘ l\ I‘ I\ I\ P.“ MEASURING UTILITY OF WEALTH AEONG FAiM HAMAGflfi By Albert Nelson Halter AN ABSTRACT Submitted to the College of Advanced Graduate Studies of Michigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics Year 1956 Approved .fi’w V L82? :L‘qxlfij J. Both historical and recent develOpments of the concept of measur- J. able utility provided the basis for deducing thel rp0tneo 5 tested in rso 1ypot1e sis is that there exists a te hnique by d ,C 1.1 U) fl 6 U) P e U) o l- :5 \ J F.) }.J. whicn utilityo oft ealtb can be measured. The second hypothesis is 1 i that a correSpondence can be discovered relating past and present J. a. characteristics of indivicuals to iuture manare al behavior via esti— mates of numerical utili The formal model developed by von heunann and horgenstcrn, further explicated by Friedman and Savage, and applied to farming situations by Johnson, provided the operational p1ere ,uisites and theorems for a correSpondence between utilitv entities and numbers. The Interstate Kanagerial Survey provided data for testing the hypotheSes. A set of questions as}: ed Q29 farm managers in seven mid- western states whether or not tnev would accept certain odds in hvpotne eti- cal risk taking and insurance situations. he que stions we re constructed so that the elements of the questions could be identified wit 1 the relevant aSpects of the model. The answers to the questions were thus eitl1er consistent or inconsistent with the specifications of the mocel. The main analysis derived utility curves for the farmers who gave consistent answers. From the utility curves estimates Cl lT_LM ;—-,‘ marginal utility were made. These estimates, which are interne1sonall comparable, we re relates to other chcracteristics of the farm manag ers iv v .5. ,._._‘_ . ._ 4.. r. _,_ k. V”. 094‘ id.‘ —. ~ I ‘3‘. v, A‘, s ~. ‘.- \~ _ “a ‘ '~ '_ \M ~‘v‘ ' o “. \ -H “h - \v ’1 n“ -‘ \ —.«., v \ —\ ‘y ‘. ‘ 5.. — .~ ,5 . \- "'4- T» ~ ~ intervi iewed. Since these estiuates are 1 lative to the assignment of an ori3in and a unit of measure, they are useful for pre.ictin behavior but not for making welfaret CCOHOIlC reoom1ent ations. The variables which were found to be related to the estimates of relati e marginal utility include: (1) net worth, (2) gross income, (3) debt position, (L ) t; pe oi farring and (5) concern for the two types (statistical) error. In g eneral, as t1e marrinal ut ili per dollar oi additional wealth increases, (a) t11e ir1divi duaal's amount of debt increases, and (b) he tends to be enga3ed in more risly types of farming. As the marginal disutility per 6.0 W11 lost wealth increases, (a) the individual‘ net worth and income decreases, (b) he tends to be engaged in less risky types of farmin3 and (c) he is more concerned about not taking action when he should. It was concluded that t11e tec1nique used in the Interstate hana3er ial Survey provides some estimates of cardinal utility which distinguish individuals on the basis of meaningful managerial behavior. Further, it was concluded either (1) that in some cases the technique either was not used as extensively as it should have been or (2) that the 1 O interviewers had Qlf' f]. .L iculty in communicating the questions to the farmers. These shortcomin3s were made explicit and remedial steps proposed. \J TALE CF COLLSIJS CiAPTflR I Iluw)O—JLC-LJ-(J'JOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.... T11eoretical I atuie of han................. Review of Literature...................... Statement oi Iypothes-:s................... Organization of Thesis.................... I: SOLI:)C~J CF DAl'i-IAOOOOOOOOOOOOOOO000.00.000.00... Origin of Interstate he naje rial buu1»..... illuVlSUde 119.118.:91 Lal sulfiVV’jt—jooooooooooo Schedule Construction and re-t-asting Inter'i ewer School.................. Statistical Sample.................. Interviewing........................ Coding Procedures................... ’cports and Criticism.................. III LOb‘LIL. ALD‘ l‘L‘JCiLI.‘ QUE E‘OA. 141511-41”; UT ITY.... Tfle lLOdeloooooooooooooooooo0.000.000.0000. Technique of Quantii* nying gUtility....... surLl‘LarJrOOIOOCOOCOO0......0.00.00.00.00. ./ IV TEE LODEL ELABOJATiD TO E TAHLIS Ei CO'STS”JA‘L \CLASSIi‘tlc.‘¥fIOLJS.0000000....00.00.000.00... Elaborated Wlo el.......................... e Patterns of Consistency............ Within-Odds Consistency............. With lndiiierence Point.......... Between-Odds Consistency with Three Indifference Points............ Between-Odds Consistency with Two Indiffere ce Points............ Between-Odds Consistency with One Indifference Point............. Without Indifference Points.......... Between-Odds Consistency* with No Indiffelaonce POj.-ntOOOOOOOOOOOOO Within-Odds Inconsistency................. Betwee -Odds Inconsistency............. None of the Odds Answered................. vi Li 1.3 l7 19 19 KI.‘ \JJ L—4 K." 1- LL) —q'~.—‘ 1"' « . ¢ .11-“; v. ~ . ~-..——11 . w a .1. ...- s l .1 TABLE OF C 1”TEITS - Continued CHAPTER Page Statistical Comparisons of Observed and Expected i: ull‘UClkis 00113154581103”. 0 o o o o o O I o o o o o o o o o o o O O 0 Statistical Comparisons 01 Observed and EXLect3d “ tamee —Occh ConSIStenc....................... U2 Ha — ‘1"1~-1"‘1 ‘m 7'11\r“r1 fl 1' Tfr'i *‘1 T. "If" I ‘Tfi ‘j-*'71".-:-I\-l - 1-) '4 V 111.: b-LIJLJ-iJJAO JOB) 0.}: 4.1.4.1 DCLLJDUqJLD A5114) I;:.L:11;LJ‘.1I:_L.bS. O I O o 0 Q Q Q g Q 5) , ,1. COCALH}? Proc%u{1d-L~30 . O C I O O O O O O O O C O O O O C C O O O O O O O O O C O O O O O C O O O p) \f I \j ' COd'ad In101‘filation.OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO.I. Answer Groups Defined.................................. 56 Between State Di“m‘ rences by Answer Groups with Special Reference to Interviewer Bias....................... 59 For Gai ns........................................... 59 For Losses.......................................... 52 Characteristics by States........................... 66 Characteristics of Farmers in EaC11 Ans er Group on Gain Questions............................. Characteristics of Farmers in Each Answer Group on Loss Questions............................. 4.] DO: D C: H a .1 TILG P16001311 O.L Intelrv'ieI‘JCI‘ Bias..................... (JO SLJ-IEIETLQTL’v.‘.OOOOOCOQOOOOOOOOOCOCOOOOOCOOOOOCOOOOOOOIOO... 92 VI DERIVATION Afil EJAIUATIUN OF EEPIRICAL UTILI H'Um. $33.. 90 Developmento of Derived Types........................... 98 Derivation.c1 Utilit}r Function...................... 90 Derived Types Dei‘ined............................... 101 Analysis of Relationships Between Erpirical U tility Functions and Relevant Variaoles.................... lC3 Derived T3 es for Gains Questions................... 103 Derived Types on LC ss Questions..................... 113 Derived Types ior Both Gains and Losses............. 127 Involving Weak Types of Consistency Unly......... 132 Involving Strong Types of Consistency Only....... 131 Summary....................................... 1 Pre dicuions from the Utilit by Mqu tions.............. 1 Reliaeility of Preeictions.......................... la E\raluation00000OOOOOOOOOOOOOIOOOOOOOOOOOOIIOOOOOOIOOOOO lII-JpI—l‘ VII 81]-;sz MID CC1IU LUSIC::300000000000ooooooooooooo.oooooooooo 1.ng I._J -1-w— ~T-v1r1’1r1 "" fi 1‘1 Lab—LLALUJJD 01.144410000000000...00000000000.0.0.000000000000.0.0000. 1/) O p {DICESOIOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO...0.0.0.0000... l’\8 I-‘r‘ APPE vii g. V 1 gm». a 5-..- ,. r... r 4‘5- A. ‘ 4 i; q 1. J...“‘ . _“ "«C.7_ . U "‘ U “ ‘- dkv ‘ 1‘- . -_- u.,A . __'_ , _ \ d-~~~ . ,\ V ‘ . J! ' *- U, .1.“ 4 U'YT' a..‘ I ‘, 4"” ‘V \ .\- a-, ‘ ‘ _ ‘. ._ R‘ A} ‘ 4~-‘ . \. \ h‘ . \‘ ‘ ~V- I" V- .1 .. . 4‘“ . - '5 b. I '-"‘r V‘ . ‘ J- ..*’*v- 1 v 1. Va. ‘ " —P“I . » \ s. Hy,“ ‘ ‘ ‘.~ . y , 4 y. u-‘- ‘ ‘9 ~ - «.‘ . TABLE 10. ll. 12. LIST OF TAJLES Characteristics of the Sample of Bi: gut Strata ior the InteI‘SI/c'l” V 1:1118-3331 *a'l SUI )jerOooooo:oooooo00000000000000... Description of Gain and Loss Situat:Lons Used on the Inter— state hanagerial Study Schedules........................... Observed and Expected Number and Proportions of Within-Odds Consistencies for Gains and Losses......................... Number of Individuals in Each of the Four Tgrpe s of Between- Odes consisuencrerOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Distribution of tm 3529 Schedules over the Six Answer Groups, ior Ga ir 18. Distribution of tne Schedules for Each of the Seven States Over the Six Answer Groups, by Per Cent, for Gains......... Distribution of,Eacn Group Over the Seven States, by Per Gen—t, for GainSOOOOOOOOOOOO...OOOOOOIOOOOOOOOOOOOOOOOOOOOOO Distribution of the 529 Schedules Over the Six Answer Gl‘oups, iior LOSS-35...o.00.0.0000...ooooooooooooo.0.000....- w Distribution of the Scheiules for nacn of the Seven States Over the Six Answer Groups, by Per Cent, for Losses........ Distribution of Each Group Over the Seven States, by Per Cent, for LOSSGSOOOOO0000......OOOOOOOOOOOOOOOOO0.00.0.0... Average Net WOrtL, Gross Income, and Debt Position of EtesponCients, b3}. StatESOOOOOOOOOOOOOOOOOOOOOOQOOI00.0.0.0... Average Years of Respondents' Farming Experience, Age of Respondents, Iumoer of Respondents Dependents, by States.. Ratio of Debts to Assets, PrOportions, by States........... Proportion of Total Acres Rented, by States................ Proportion of Income from Farming, by States............... Type of Farm, Preportions, by States....................... d H. H' Ho 0‘ \ \J) 65 '7 - O . —.. . . < . ' - .. ., . - F « . .o \ — . F. 4. . ., a" . *’ ' 5*. ‘ ‘ . . ‘ . - n v . I - u ’ ‘ ’0 ‘-. , .. , . .- . , | ’ V‘ . . r v a ' ‘ I ,. 4 ‘5. . .. » . , , . . .. 1 r x . r * ‘ ' ‘ I ‘10 v ” > , 5‘. u v ‘| I L _ _ _,.. . -: x. p ‘ — . ‘1 I “ - ’ . 3 V - S - ' . S-. . - . . e . ~ - 9 ~ 0 t ' l ”on s I. . 0’ 'I a . . up i ’ " " ’ ~ ‘ ‘- - r I “ ‘ - '- ' . . r. k I. M . » , , c ~ . . .. - . c v . av. tn . a‘. ‘ ‘p‘ ' ’ ; -. , . a . ., ° I 4‘. o a ' .' ~ ) . '4. . \- - . ._ .. . .. a c - ' . . -h - .. ‘- A Ir, Uh ‘ ‘\—_ . '1... Q.... -“ ‘ -6“ va.‘ I" V» "V '1‘" N " C) ‘ d‘\_ &_ ~ .I 8“ . l -. ‘-I’ 7. H, C‘ ‘ -.: V .". \‘J‘ - LIST OF TABLES - Continued k,“ \JJ Concern for the Two Types of Error, by States............ Number of Years of School Attendance, by States............ Average Not Worth, Gross Income, and Debt Position, by Answa‘r GTOUPS, 0n Gainsooooooooooooooo0.0000000000000000... Average Years of Farming '"perience, Age, and Lumber oI Dependents, by Answer Groups, on Gains..o...o.............. Proportion of Total Acres Rented, by Allower Groups, on GairlsOOOOOOCOOOOO0.0.0.0...OOOOOOIOOOOOOOOOOOOOOOOOOOOOOOOO PrOportion of Income from Farming, by Answer Groups, on Gainsooooooooooooooooo00000000000000.00000000000.000.000.00 Concern About the Two Types 0; EIror Ior Six Answer Groups, on GainSOOO...000......-OOOOOOOOOOOOOOOOOOOOOO00.00.00.000. (1 Type 01 FaIm, PrOportion by Answer Groups, on G8ins........ Avera3e Let Worth, G: 088 Income, and Debt Position oI Six Answer Groups, lor LOSS'BS.OOOOOOOOOOOOOOCOOOOOOOOO00.0.0... AV era5r, e Years of Farming Exper'ence, A39 and Number of Dependents of Six Answer Groups on Losses.................. PrOpOI tion OI tLLe Total Acres Rented for Six Answer Groups, onLOSSBSOOOOOOOOCOOOOOOOOOOOOOOOOOOOOOOOCOO0..OOOOOOOOOOO. Concern for Two Types of Error by Answer Groups, on Losses. Type of Farm, Proportion by Answer Groups, on Losses....... Ratio of Debts to Assets, PIoportion by Ans Jer Groups, on LOSSeSOOOCOOOOOCOOOO0.0.0.0....0......OOOOOOOOOOOOOOIOOOOO. "omparison of Indiana and Eicnigan Interviewers on Per Cent of Scziedules Falling Into Each AnSIIer Group, on Gains...... roportion of Sacn AHSWJT Group on Gains Taken by Inter- . o .- , _o I. [:3 _: ‘1 4' r I_ ’\ VlGWOTS ln LhClr XJSPGCGIVS budLUSooooooooooooooooooooooooo f Sa c1 Answer Group on Losses TaLen by Inter- heir AGSPGCtiVG StateSooooooooooooo0.000.000... Proportion viewers in F30 N] (? bl o<“ rvn ‘. ' \ pct/n . - ‘IHH . ,. uL—J . . 1—. 4-——... A --.' - I 0, (1‘ I1" ’_I . d-. l . 5 ‘_ 1". oc' -_ ‘_~v '. v , '5 . --—». . . :" JIO v_‘ ~ \ . v-_ - , 0. ‘ ‘— .~0 . C \_ -~ (VI.-. '~4 J]. ' T .‘a-~~ - . » ..,; r sU. - 'v- .4 ,n. ..__ H “-0 5-“ "d. ‘_ V v ' _ ~-.3 . ‘| ‘ - --.. ~“ ~~ . .~ {V ‘55 O _ r- b.” - Q ~\ . ~v ~ -_V ~ .. fl. *5 «. I - r. Jv ‘ ‘v I V- \ n- a I: a V5... 5‘ h .. ‘ r I 3-.-; ‘1. u» - ‘. “a 2 N ~V~ "s ‘V V “I .. ‘ I .,_~ . \, n. \v‘ t 'V l LIST OF TASLE‘ - Continued TAELE .- 1 2h. KL ’ \J—‘L . \A ‘ O\ DB. It. ‘49 . 1:, (j\ DY. Relative marginal Utilities for Derived Types and Number of Respondents at 30,000 Dollar Gain.......................... 3e .et worth, Gr 53 Income and Debt Positions for the Derived Types on 30,000 Dollars Gain....................... Aver 5e Length of Farming Ex.erience, Age and Number of Respondents' Dependents for Five Derived Types on 30,000 DOll-ar GainOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOCOOCOOOO00...... Composition of Each Derived Type for 30,000 Dollar Gain by ~43 +53 DUaUr'JSoooooooooooooooooooooo000000.000oooooootooooooooooooo Types of Farming, Preportion in Derived Types for the 30,000 Della-LI. Ge'inOOOOOOO0.0.0.0000...OOOOOIOOOOOOOOOOOOOOO Relative Dar3inal Utilities for Derived Types and Dumber of RespondeIts at 7,300 and 33,000 Dollar Losses.............. Average Net Worth, Gross Income and Debt Position of ReSpondents in Derived Types on Losses..................... Farming Experience, Age, and Number of Dependents of Individuals in Der'ved Types on 7,300 and 30,000 Dollar 1.108888000000-000000000000000.0oooooooooooooooooooo0.00.0... Composition of Each Derived Type on 7,300 and 30,000 Dollar Losses byS'bateSOOOCOOOOIO....0...O..I.O...OOO..OOOOO00.... Type of Farming, Proportion in Derived Types on 7,300 and 30,m0 Dell-8.1.. Losses.0......0.0...OOOOOOOOOOOOO0.0.0.000... Concern for the Two Types of Errors by Derived Types on 7,)00 and 30,000 D0113? 1.1088680000000000.000000000000000... weak and Strong Consistency on Both Gains and.Losses, Losses Only, and Gains Only, by States..................... Average Net Worth, Incoae, and Debt Position of Weak and Strong Consistencies in Both Gains and Losses, Losses Only, and Gains OIfl—brOOOOOOOOOOOOOOOIOOOOIOOOOOOOOOOOOOOOO0.00.... Forms of Weak and Strong Consistency, by Years of School... 108 115 “,5..- -. _‘--1 “1...; ,4. x .4.. . .1 . - . . ,A . .. ,~o -._..; - 9.. I.~~V.. . H. x , ,‘Q L. .. t‘" ”‘ . .‘- V F 1\ ~ B.-- ,'001 ‘ "\— - h I C A.-- V‘. "A ~. J...; - n. _ ._ .°'OI .h, V‘ .. ~ #1. r . 4 . ‘_. . -,‘_ ~__ ‘ -, . u . _ - l‘ 1‘ A. f . ~- J u 1‘“ .. v "’ - -';‘. . . ‘V ;~,. J '— ' . ‘ D.” ' s. :r- a. 1 1-. "‘ .- ‘A.._ (- -.. a ‘ -__ \ -H .,__ ’«or. ', ....._ «~A ‘“’ .J. .. - I ~ . “Q "t ‘7‘ ' ‘UV ‘ . ‘ ‘V ~--, 0 - .U c IST OF TELES- Continued :2. U p pt: 0 _4 \n r O __, I", L. : / / ' \"1. C‘\ o 37. ’T} "J l 1"“ " if \L) va Concern Tor Two Tvp3s ; Error by Jesr(nccnts U“o W3 re w;;k and Strong Consistent on Dotn Gains ano Lossas, Loss 3 HGrl,, and Gains omdrOOOOOOOOOOOO0.0...0..OCOOOOOOOOOOOOOOOO:’OOOOO 13 A erao 3 N- t worth, Gross Inco orae, and Debt Tor wea :Con- sistencies on Botn Gains and Losses and Gains 0nly......... l33 Forms of‘Wea Consistency by the Lumber of 1 rs of Sclool AttenC1anCGOON:OOOOOOO0.00.00.00.00...OOOOOO...:.O..OOOOOOOOO 133' Concern ior the Two Types of Error bv Ind’viduals WLo N3 re Weak Consistent on Both tne Gains and the Losses, and Ge ins onlzerOOOOO0.0.0.0000...0..0.0.0.000000000000000030.00....0. 131:1- Averagc Net worth, Gross Income, and Debt Tor Strong Consistencies on Gains and Losses, Losses Only and Gains onl3r00000000.0.0.0000...OO0......OOOOOOOOOOOOOOOOOOOOOOOOOO 1‘03 Per Cent Strong Consistency on Both Gains and Losses, Losses Only and Cains Only witnin States. . . . . . . . . . . . . . . . . . . 130 Per Cent of Individuals Who Attended the Specified Nunber of Years of Scnool ior tna Strong Consistency.............. l3? Concern for the Two Types of Error by Individuals who were Strong Consistent on Both Gains and losses, Losses Only, me: Gaj—I‘ls onlbrOOOOOOOOOOOOOOOOOO0.0...OOOOOOOOOOOOOOOOOOOOO 1:7 Amount of Ca in or Loss Lecessa ry to Induce Acceptance of an Uniaj-r OCld.OOOOOOOOOOOOO0.0.0....000.000.000.000...0.0.0.... 1—er flankin“ of Sr; 3n Insurance Schemes by your Der’vative Groups on Losses 1’ O...OCOOCOOCOOOOOOOOOOCOOOOOOOCOOOOOOOOOOO. lLL-j Distribution of Cuestions on tne Field Schedule Used in the 10:10.8 surveVOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.0.0.... lS}O A~ 3.. 3.. q " p . ya“ ‘ U "a y. . .. .. '“n-. p -‘ v- a -....‘-¢- . ‘ A A--. P . L . 2 '“-~_ ntvg, - t S -.-u__ . _ . V‘" ‘-' '- ~ h‘ 0 .‘-~“‘ . O “AA N- _ v ‘ . -..-.~b. u.‘ t » , ~. ~ - ‘ .. ... ~T" ~-,._.' ». g, .- ‘4‘ "H-- ’ ‘- , _ x . He“ 1;) ‘»'y-.‘. z .. - ‘ .‘QIUH“‘ - 4.. ,N :_ “a “r as3~u_\,u ‘7: . :~~ “W : _- s , .‘3 .~ ‘»_ "h uh? ‘ «4“.‘ u ‘“3 .34 . ‘ 3 ¥ Hr~r . - A s. I” .. - . ‘ ..~‘c a . y ‘ -’\-_ . ‘1'. . . dy~‘ ~‘p‘ ' ' r 'v ..\.‘.. g ‘~.T V». . - u‘ .. ._ 1;» 1 M“ .d.‘ A~M‘.“ V This thesis, looked at from a vantage point which provides a broad prospective, concerns the nature of man. The theoretical or formal propositions presented about his nature apply in every situation, i.e., explain all oi his motivated behavior. The empirical content of the thesis concerns the nature of a limited field of men, specifically farm managers. As in any scientific investigation the objectives are to describe reality with theoretical and empirical laws and predict the course of reality on the basis of these laws. The description will be of the behavior of individuals in Speciried situations; these situations are regarded as risky, changing or uncertain, since there is incomplete knowledge of the future. Behavior in managerial studies is called decision making. Lescription of this type of behavior in these situ- ations is important, not only to the scientist, but to individual farmers, teachers, politicians and administrators who can use such facts in combination with ethical propositions in formulating statements concerning both public and private policies. These individuals construct recommendations about how to reconcile the ethical philosopher‘s concept about "what ought to be" with the scientist's concepts about "what is." The redictions concern what decisions or behavior can be eXpected under certain circumstances. These predictions, of interest to the "Y" - . ‘..v 1...... a . . . ..-.-- O» :- u J -. ... , V‘.‘ M ,r ‘ . ~ “." . .1 '7 ‘v‘tfiw - _ _ fl .-\..., a v..,-- -4 . ~ ‘l .2”““ n~ -‘2 v “ *‘...4. ‘ --~.< .- . ._ a .-\ V‘- U-ldv. ' ‘ _~ ' I -‘ nfl‘x-“na ... -,_\ — r- i‘lv l L - i" “w ‘6. ~.‘- A.- V- -— iz-l, . ~ _ ”vuyv. " s. - er a, , . ~'~."-’.‘ tnvl ._ ._.‘A “‘¢v-l‘ ‘v‘ . ‘ .4 s“, ‘ .f'. C ‘ e N“ ‘V 3_ _ ~ \ A "' ‘ "3‘. ‘V i i -u-‘ 2"» . I ~ ‘ ‘M -c — -. U wr- R‘ .. x. .- ~ ,_ -.,. Q » A~ ‘ " scientist from the standpoint of systematic knowledge, are of ultimate importance to the individual policy maker. It is these individuals who upon the basis of the scientists' descriptions and predictions attempt to preserve, alter or leave unattended the existing situation. The method of scientific inquiry involves deducing hypotheses from a theoretical structure and subjecting them to empirical tests. If empirical laws can be established from this procedure, then broader theories relating these laws to other empirical laws can be formulated. This process is intended to lead to more complete knowledge concerning the nature of the universe. In order to graSp the content of the thesis the theoretical nature of man must first be stated. Obviously the entire theory can not be tested by one study and only a few particular hypotheses can be tested; nevertheless, the results and conclusions must reflect upon and question the entire theoretical structure. Theoretical Nat re of Man * Man is thought to be an animal that is possessed of a free will; he is motivated by his desire for pleasure and aversion from pain,1 and is basically constituted to maximize this pleasure. A free will means man has freedom of alternative choice; this means he possesses lThis conception of psycholOgical hedonism is easily confused with ethical hedonism and with ethical utilitarianism. Ethical hedonism maintains that pleasure is the only positive ultimate value, i.e., intrinsic good; whereas ethical utilitarianism maintains that the right act is the act which, of all those open to the individual, will produce the greatest amount of pleasure in the world at large. —\ .. u’ ".9 6;--- ..-- _- 'u\_--.;~ 'Q‘ 'C‘A‘ — ‘- . . . r - .. , a..,_c ' ... -‘ 1* vv.-v-... , ~ VA ‘,:..‘; '. 5":U‘¢u ,, . u.- .fl. - ’_ V- ‘ " 4 “‘4 r...‘., u.. _ . ~ .2; :n 3 _. V“' ‘-‘v ...:.., A a-.- . ~ ~~q ’ vukyv’ v.., ‘ .. ‘. 3’. ~.. ‘ v”. .“O--‘- "" ~.._ 111 I ’A '- '~‘»v-, r ‘V.‘ n, ‘ . ‘~,‘ h ‘ l‘ _- . ‘t-: In . .\~ ’. \ V ., :31“: . «‘y. A \ \‘N‘ v ‘ .“g i - ~-\ flu) ( (I) the ability to choose among alternative possibilities of action. This procedure of choosing between alternative courses of action is called decision making. It is assumed to be a conscious rational procedure; that is, man intentionally uses his reasoning power to deduce his alternative courses of action. It is supposed that his desire for pleasure and aversion from pain corresponds to a scale; this means the act that produces the greatest pleasure is at the top of the scale and the act that produces the greatest pain at the bottom. he naturally chooses the acts that produce the maximum amount of pleasure. An appar- ent difficulty arises when comparing qualities of pleasure or pain. It seems that there isanbther dimension of pleasure besides just its quantity. However, this difficulty is overcome by supposing utility as the factor abstractable from every human experience and thus the true motivating agent. Definitions of utility as being isomorphic to hedonistic pleasure, satisfaction or gratification, have been equally meaningless from an operational point of View. None of the terms have been given meaning by a set of operations orpropositionscu'empirical significance. (For purposes of this thesis the term utility will be used to denote a certain set of operations that makes it possible to set up a correSpondence between utility and numbers.) This thesis is concerned with the measurability of utility and its usefulness in making predictions concerning human behavior. The history of the measurability of utility is not only interesting but provides the background for understanding the complications involved. The next section will review three aSpects of this history. These are: - 5'. .. . . .2 . K v” vi. . a Aflu .. l. My .c- nu— .» n. —- X: .4, a. ,. . ».. ..v a. .V y... I —k (l) the initial theorizing concerning utility, (2) the intermediate stage of forming concepts concerning its measurability, and (3) a recent stage involving experimental attempts at quantifying utility. As a review of the literature regarding utility is beyond the scope of this thesis and has already been done by others,1 only revelant examples of the literature pertaining to the above three aSpects will be considered. Review of Literature (I Utility theory was brought to the forefront about the beginning 01 the nineteenth century by Jeremy Bentham.2 he suggested the measurement of quantities of pleasure and pain; however, his purpose was different than the one proposed above. His primary objective was to construct a more rational system of civil and criminal law. Thus from its inception, utility was construed to be both a motivating factor and a moral factor. It was thought that, from the intrinsic nature of man, rules and principles could be discovered that would prescribe "right" action for men. This thesis attempts to abandon the notion that utility has a moral connotation and restricts itself only to the motivating irplications.3 1G. J. Stigler, "The Development of Utility Theory I and II," Journal of Political Economy, Vol. 57 (October, 1950). 2Jeremy Bentham, introduction to the_§rinciples_of Morals and Legislation. (Oxford: Clarendon Press, 170;). 3M. Friedman, and L. S. Savage, "The Expected-Utility Hypothesis and the heasurability of Utility," Journal of quitical Economy, Vol. 60 (December, 1952), p. hTh. The authors point out the wideSpread con- fusion in using "the same word-utility-to stand for two quite different things: on the one hand a quantity that it is useful to regard an individual as maximizing in interpreting his behavior and predicting r1.,|r.‘ ;’ . .1 4»- o A‘c h...» U~"‘ an. .4 ..A C. 3. Q. .1 p: C .H. I O ...c. .... .3. H. l‘ .- _ i . i 7. i. t. F. .g u ..v at“ pH! a" A 5“...- '- ‘4. d\§0~ . FL . . P.V . . . e U ad.“ : a ‘u ... _ «O \ To . . D. .1» .7». I an.‘ ..E ‘v A». .7" ‘ Ly J $ . ‘ \‘\ ~ \ I. .3 . .n. H . a .. C\ :5 .d\ .. u. ... .\ F4. .3“ ~_\ ‘v. * C a Q E. ... \v u. A 5 .. This procedure is based upon the author's belief that "no study of what men do can avail to teach us what they ought to do." Later in the nineteenth century three economists, Jevons, henger, and walras attempted to make explicit the consequences of the measur- ability of utility in the concept of marginal utility. Of the measur- ability of utility Jevons said: A unit of pleasure or of pain is difficult even to conceive; but it is the amount of these feelings wnich is continually prompting us to buying and selling, borrowing and lending, labour- ing and resting, producing and consuming, and it is from the quantitative effects of the feelings that we must estimate their comparative amounts.l In constructing a way to measure utility he employed the familiar measuring stick of money. The price of a commodity is the only test we have of the utility of the commodity to the purchaser; and if we could tell exactly how much people reduce their consumption of each important article when the price rises, we could determine, at least approxi- mately the variation of the final degree of utility. . . . For the first approximation we may assume that the general utility of a person's income is not affected by the changes of price of the commodity; so that, if in the equation 1¢ ){ -.m-¢/c we may have many different correSponding values for x and m, we may treat.1Vc, the utility of money, as a constant, and determine the general character of the functioncf x , the final degree of utility.2 his reactions to changed circumstances, and, on the other hand, a quantity that he 'should' maximize or that society 'should' maximize or help him to maximize." 1W. S. Jevons, The Theory of Political Economy (hth ed.; London: Macmillan and Co., LimiiEb:“lyll), p. ll. 292, git., pp. lhé-th. The final degree of utility is its marginal utility. ..-, v I! : tr C. -4----v vv \- 'r-‘O—.‘. r -uvu‘.‘. ‘-‘¢- » ‘ f n :v‘ 51* ,4 ‘7» Q -.'t.‘. - . 'v ..V In . ~u.- ‘ ’ .‘- .“‘ t O I. . ’4 f.. H-» P .' .. J - I . . . a g . 0 c.” ‘ . G a a V 'v.c- : '_ w‘~‘_ . v‘- f‘. ~~‘~ _ -‘~...V\ ‘ c. I" 'A W _ -. “-A v s. , v . s‘. V -. ~-. g I - . 'V‘I— Ir’ ‘ 'h.’ " ‘. J ‘ ' , “- _ . ‘ s, “.i Although Menger gives no empirical technique for measuring utility he attempts to distinguish value and utility. Utility is the capacity of a thing to serve for the satis- faction of human needs, and hence (provided the utility is recognized) it is a general prerequisite of goods-character . . . what distinguishes a non-economic good from a good-subject to the quantitative relationship responsible for economic character is the circumstance that the satisfaction of human needs does not depend upon the availability of concrete quantities of the former but does depend upon the availability of concrete quantities of the latter. For this reason the former possesses utility, but only the latter, in addition to utility, possesses also that significance for us that we call value.1 A translator's note2 points out that henger thought the concept "utility" is entirely objective and lacking in psychological content. He pictures it as an abstract relation between a species of goods and a human need. walras does a masterly job of avoiding the empirical job of measur- ing utility. Instead he says, I shall. . . assume the existence of a standard measure of intensity of wants or intensive utility, which is applicable not only to similar units of the same kind of wealth but also to different units of various kinds of wealth. . . .3 Analytically, if we are given effective utilities as func- tions of the quantities consumed according to the equations . . . and . . . then the "rarete's" are designated by the derivatives, . . . and . . . .4 The translator of walras' work points out that it would have been better to have chosen a word less vague and less ambiguous than 1Carl Nenger, Principles Q£,ECQEEWiCS: trans. and edit. J. Dingwall and B. F. Hoselitz (Glencoe, 111.: Free Press, 1930), p. 119. 292. cit., p. 118, No. 6. -1 , a, . ._ _ -—. . my; 3Leon walras, Llements of Pure hconomics, trans. William Jaiie (Homewood, I11.: Richard D. Irwin, Inc., 1920), p. 117. 492. 313., p. 120. 5 ‘ :- ~..._,_‘ ’- 1' ¥ K. ;. \— L __’ ‘ .- ... ‘ ‘Nfi. . _‘ ~ 1.4 . ‘ ‘ ‘ . P“ "rarete‘s" to express his mathematically precise concept; but it was clearly out of filial piety that he perpetuated in his own work his father's favourite term.1 he also points out that "rarete's" has the same significance as Jevons’ "final degree of utility" which Jevons defines as the differential coefficient of total utility considered as a function of quantity.2 Although the main application of utility theory has been to the concept of demand, more serious attention to the measurability of utility functions than was given by the preceding three economists was given by Fisher, Pareto, and harsnall. Fisher constructed a technique for measuring utility after formu- lating the following mathematical system.3 Postulate: Each individual acts as he desires (1) Definition of utility: utility of A i'utiiity of B if the given individual at the given time prefers A to B or neither (2) utility of d A n.-:=n utility 01 d B if the utility of d A = utility of ndh (h total) and utility of d B = utility of d M 7 (.2) g g nwrginal utility (l) d 11' <1 2 unit of utility (util) A being given I p m. 11bid., p. 506. 2Ibid., p. 506. 3LFisher, hathematical_Investigations in_the Theory of Value and Prices (New haven: Yale University Press, 19907; I § I,‘ ' “U - ‘O l‘ .— ; . -. ‘.._, .. .._."_._ "u _ g '~ ." . ‘ ‘Q‘ ‘. r ‘I v ”‘4 .._‘:‘ ' §' ‘ ““s “t ”on *'u '~J‘ .5 ‘ I .,_~ .L1JCV , s -4 '.« ~'§-_’_ r“ V‘ .. «'15 \“5a‘_ . v. ' .1. I. . . “ ‘ A V ., ‘1 ‘v‘ 3““ k. :‘n ‘ u“ n- “*v n. V“ ‘ G 'fw‘ '.“"_a-_- ‘ v4“ "7‘. ‘VL ,. «t. .‘ _. ML.v 1:“ CO A d v . (p) 3732 d A = total utility O (O) A ' g—% = utility value A (7) d V d v . __._ . d _ _.._ _ n ,, d A A A (.1 A _ balIl O Assum tion: d _V - a . q p : Function 01 A only 3.! Corallaries: From definition and postulate, hen B is exchanged for A 910 u <: U1 u C319» :1:- <1 u> From (2) and assumption, in the equation: utility of d A_ utility of d B =n, tae value of n is independent of the particular commodity and of its quantity u used in the definition. The method of measuring the marginal utility was to utilize data of family budg gets and prices so as to compare the wants of two typical families of different incomes, in the same community, by using as a yardstick or criterion, a third family having identical tastes, but differing in the amount of income and living under a different scale of prices for foods, rents, clothing and other items of consumption.1 Further details of the technique are of no consequence here; however, a quotation found later in the same paper sheds light on Fisher's Ultimate purpose. ". . . according to which way this product differs 1Irving Fisher, A Statistical he thod for Measurin‘ "ha”;lnal Wfimr—J Utiljiy“ and estinL the JUSbiCe of a Pro glessive Income Tax in Economic -fi--.‘- Mt... ”9‘ Esszyfs Contributed in i1.onor oi J011n Bates hClarit. {New York: Macmillan,” 192 I ), pp. 199. (“5- ‘,7,.. --- ..v.. a..- v‘ Q v»‘.'p' - P" US“.C‘.-~.A . . TVL- ,. "O c '- ‘>\w--;c ‘2“ v“ l «U Vfi. u .~ . ..-- - -__ - .1 ‘v---d C ‘, _ t 'r '1- “ .3 " ' r “~144J .' . , . .a. -« ‘a 1 - — . ’ VII-.. u. — .- n' -A. ~ “~\-.- - - n - C 4- c. . .,-j‘. I ~ ‘ ’0 q-. “-¢-uu a--- 1" , .. _ , c ‘-._ :«~. - .‘v‘ . g“. * 3' : . ‘ . ‘ ‘4 '*~,._ 0 h ' | cc, ~» ‘R‘... “"v..‘ “..| "' "U I e “vs. I" -.. *“ — ‘4‘ -' :~- ~ V. ’ V- n.‘ ' “c1 ' - \‘ «1 a _v-‘.~ . ‘ A 9.. fiat“. y~ -. ‘-_, ‘ . \I v . - . - ~ V.” ' .. . x .-r ‘— ~ v H 1‘V‘.‘ ' -"~-\ ~‘ , . 'vn -1 ~a .3 : "U ‘ R r: « ‘4‘ '_u- \ U‘ 5 . s ,_ V, ~‘ ~.- “ V“*\,, .p_ J from unity, we have a justification for pregressive or regressive taxation . . . ."1 Pareto in addition to arguing that the slopes of indifference curves can be deduced from budgetary data wondered if a unique total utility surface could be integrated. He answered by saying that a unique total utility function could be constructed if the consumer could tell the magnitude of the utility gained by moving from one indifference curve to a second relative to the utility gained by a move to a third indifference curve. Pareto doubted that the consumer could rank these utility differences.2 In harshall's analysis of demand he makes an assumption of constant marginal utility of money. However, in regard to risk taking and insur- ing, which is more appropriate to what is to follow, he says: 1 . . . from the general law that the utility to anyone of an additional £1 diminishes with the number of pounds he already has, involves an economic loss, even when conducted on perfectly fair and even terms. For instance, a man who having £600 makes a fair even bet of £l00, has now an expectation of happiness equal to half that derived from £700, and half that derived from £500; and this is less than the certain expectation of the happiness derived from £600, because by hypothesis the difference between the happiness got from £600 and £900 is greater than the difference between the happiness got from £700 and £600. . . , the direct converse . . ., is that a theoretically fair insurance against risks is always an econo.ic gain. . . .3 Perhaps the most outstanding feature of the development of utility theory through Marshall has been the emphasis upon the consequences of 1:05.40 , p. 1.85. 2From secondary sources. Stigler, pp, cit., p. 381. 3Alfred Narshall, Principles of Economies (hth ed.3 London: Macmillan, l9h7), p. lb). .. .-.. .-.— ,.- ‘u--_-' >-v‘v-— o .- -. . ~ , " . ’- “O-l' u‘—-‘_4 - / or- r“'<--- « ~ ‘0‘ ..\n vogu- \v — . ,- -. a _ . " ‘-'--‘a..- -. \u. .o.. . ,~ ‘----d| ‘-v"_v- 1 1---..'.‘ -L...,._~-_. . ‘C -‘ \~-~. .. -u .. ~~. '---..“ s I o 0 Cr. ' ...v“' a ». , ‘1 “s. 4. ~. v- ., .- - v ..,. —..... . n. . . .. _ «&-..‘. v3. ._“. l.“ . l‘ d-‘. a . *- Pc“ I .. Q v. a r ~- 2 ~«._' . '~.. 4 aa -, a r. -,v \ 3., -- .. n." .rn ‘5», Q _ ~v . “__1 . N. “\, ~ ‘ -‘I A \- 1- A ‘ \ ’- “Q h -- - a“~ '~ ., . ~ A‘ \~ ~ . “ “U .. . W. ‘ x L. ~'- _ .. ° l. _' , . _Q. ~, 1-.“ 4 ~u‘ "«I 5! ,7 ‘_ , . ‘ P ‘. r. \‘§ -.‘ n a . ~IO ‘0 - ‘1. _ “ 3“.. ‘ . is A- 4‘ \ . . I L I lO utility measurement derived by the use of the mathenatics of physics, .- e.g., calculus and differential equations. it is not until von Neunann and Norgenstern conceived economic behavior as a game of strategy that the mathematics of set th 3ony and p1 coaoility theory play 3d a role in utility measurement. They point out that under the conditions on which the conventional utility concept is based, very little extra intuition is necessary to reach a numerical utility. They sav: . . . We expect the individual . . . to possess a clear intuition Whether he preiers the event A to the 50-30 combination of B or C, or convers sely. It is clear that if he prefers A to B and also to C, then he will preier it to the above combination as well; similarly, if he prefers B as well as C to A, then he will prefer the combination too. But if he should pre_er A to, ' B, but at tne same time C to A, then any assertion about his preference of A against the combination contains fundamentally new information. Specifically: If he now prefers A to the 50-;0 combination of B and C, this provides a plausible base for the numerical estimate that his preier ence of A over B is in excess of his preference oi C over A.1 These notions which have been made explicit by application of mathematics by von Neumann and I-Eorgenstern will be used in the latter chapters of this tiw is However, it rem ins to point out some of t.e consequences of this new approach. There are essentially two groups of individuals who have tried experimentally to measure utility; these are economists2 and statisticians. 1J. von Ne eumann, and O. Morgenstern, Theory of Games_and Economic Behavior (Princeton: Princeton University Tress, 947). 2The amount of literature in economic journals is increasing so rapidly that it would go beyond the scope of this the si s to present a complete bibliograph. The references cited usually Contain references to other literature. A good source of reierence material up to 1953 is A..A. Alcnian, "The Leaning of Utility Leasurement, Anerican EconoLic Review, Vol. h3 (Harch, 1953), pp. 25-50. 3The interest in utility measurement by statisticians is evident from the surge in literature. It is not the purpose oi‘ this thesis to " '.--'D“ .iafljl 4-, .3 _ v. - u... .. .,.-- 3. ‘fl -‘ d“ “‘""-’--.t- ~~.- _ __v _ 4 ‘ . ~‘u...4- v... ‘ I ""‘oO-‘.., La..v:--Iv. o ."2 P»-.- o~ —' - "V ‘uw‘v “P;- A _' Iv ‘ ~ .3 ‘ ~56 »‘ i .— VA ‘ V.._ . ‘n‘_ . "v -.‘ _ ~“‘_ A 5A_ . Q "i. ‘I" P.— ‘a ‘. ‘~5-b” ‘1 .. - ~.‘ F~‘v ‘~Au H “‘... M“ V . V _‘-‘- O ‘(n .‘ .M A . n... x -- HI“ . ll Experiments performed by economists, such as hosteller and Nogee, were attempts to measure utility experimentally using real money. However, the size of gain was small in comparison to what would be con- sidered a return on investment in a farm enterprise to a farm manager. Losteller and Nosee's subjects were 1? Harvard students and national guardsmen, lh of whom finished the experiment. Their conclusions, though tentative, were that subjects did choose among uncertain prospects on the basis of the utilities of the amounts of money involved and on the basis of the probabilities associated with each. One respect in which the recent worm of the statisticians differs from that of the economists is that as they proceed thev are develOping a decision making model of which utility measurement is only a part. The economists believe they already have a decision making model in what they call marrinal analysis and that utility measurement is the only component lacking before the model will predict behavior. The statisticians' point of View is illustrated in the quotation from Suppes: . . . The increasing advocacy of subjective probability is surely due to the increasing awareness that the foundations of statis- tics are most properly constructed on the basis of a general theory of decision making. In a given decision situation, sub- jective elements seem to enter in three ways: (i) in the determination of a utility function (or its negative, a loss function) on the set of possible consequences, the actual conse- quence being determined by the true state of nature and the decision taken; (ii) in the determination of an a_priori ') compile a complete bibliography but the references cite sually contain f ' ' 3 u references to other literature. A good source 0 referenc material is P. Suppes and W. huriel, "An Axiomatization of Utility Based on the Notion of Utility Differences," Lanagement Science, Vol. 1 (April-July, 1935). ‘-.. , av _. -“O “_A I . Q- . .. ,"F.: \ ~---\: u| .. 7‘." a . - 'VV. 5.. . a? w ' . ro v-§ 'a . c.-. L - n.-. -..“ v.— “ n. - - - C”* .‘ s. V' n... a ! a ‘ v.‘-. ‘W "“h- ‘ g -‘- “-- ..\ . i "o ‘4- .,. m‘ - ‘ . "‘._. ‘x ‘4_" “l v.“ a ”a \. “’- probability distribution on the states of nature; (iii) in the determination of other probability distributions in the decision situation. One of the experiments in utility measurement carried out by Davidson and Suppes used a linear programming model to measure cardinal utility and to predict further choices. The general procedure was as follows:2 Music students were used as subjects, with long-playing records as outcomes. Each subject came to three sessions; all testing was done individually. In the first session a utility curve for six records was determined by the linear programming method. . . . In the second session a utility curve was found for another set of six records, two of which were drawn from the set used in the first session to permit the construction of a joint curve. The joint curve was used to predict choices be- tween untested combinations of the ten records used in the two sessions, and these predictions were tested in a third session.’ ‘Experiments carried out byEdwards:3 attempt to emphasize the importance of subjective probability. He says people have a consistent, stable pattern of preferences arbng probabilities in gambling situations, and that this pattern of preferences among probabilities is another factor, in addition to the sub— jective value of money, which may cause human behavior to differ strictly from the expected utility hypothesis. Before proceeding in the next section to the literature that was the inSpiration for this study, consider two important characteristics ¥P. Suppes, The hole of Subjective Probability and Utility in Qecision Making, Tech. deport he. 3, Project No. 1033, Office of Ordnance hesearch (June 1, 1955). 20. Davidson, and P. Suppes, Experimental Measurement of Utility by Use of a Linear Programming Model. Technical Report No. 3, Office of Naval Research (April 2, lypo). 3W. Edwards, "Experiments on dconomic Decision hahing in Gambling Situations," Seminar on the Application of Mathematics to the Social Sciences (University of hicnigan, November, 1952). A. "' u .. L. b..v u- .‘ - - -OI— " 'n. ‘0' U \.J . *-n~ 0-— - .. LI... Q”.: I >,» . V‘ ‘ " -2'.‘ ... ..--.-~_._ ._- .. . .'1 v ,. ~“'"‘"9 h -mx. 1 ""-\ *h e, v :..c.,,.,_ w ‘ ‘ M" 'r v R.v,~~ .. V' 3" “Cub-A’ F5». . - \ "flg V‘OQM“-.“‘~ 4" - '- --t ya . -I‘ - 5-. 9%.}, VV‘ U . I "tvnu, ‘. I‘u-‘U:--.. -. ‘h- . v “k -. ‘rAs ‘ . . L ' ' ~ - v.4, ‘ O" H .‘_.- “ -‘.K \ ’v-s f 1“ Q 2 v ’ v.~.o~_.‘ b .‘H ‘—. ”Mt: _, 1-,. (‘ Uuu.‘ H‘— .. " t'- . ‘ . n‘ Q~H a -_"“ ‘v‘n I- .‘ “ s n . . h . _ ’ § .. .‘ . "H‘ . ‘hv ‘thx . x“ .. 1‘ up .. ._‘~.- ~ . ~ I l‘. i 'n. I A " ..’ 44‘ C_‘ v,“ .__-‘ ‘ "-2 .. ‘w'v . *- I -‘ fi‘ 1 \ | d. —. ‘ “‘1‘ H Kg) of the development of utility theory in summarizing the above discussion. (1) The two connotations of utility in economic literature make the word ambiguous for scientific research, and (2) The use of probability in measuring utility of specified objects or events, for example money income, raises the question of subjective probability and the utility attached to probability distributions. Any discussion or empirical work concerning the measurability of utility can not be carried out without consideration of these two notions. Some further discussion of these two ideas will be presented at appropriate places in this thesis. There remain two contributions to utility theory which are so im- portant to this thesis that they must be singled out. The problem under discussion in this thesis is ~so intimately related to this previous work that a statement of the hypotheses to be tested can not be linder- taken until the works of Friedman and Savage at the University of Chicago and Johnson at the University of Kentucky and hichigan State University have been studied. Following the approach introduced by von Neumann and horgenstern, Friedman and Savage produced an argument concerning the shape of the utility function for monetary gains and losses which was intended to rationalize the reactions of individuals to risk.1 The two classes of risk situations to which individuals react are those regarded as 1M. Friedman and L. J. Savage, "The Utility Analysis of Choices Involving Risk," Journalgof Political Economy, 56 (August l9h8), pp. 279—30h or M. Friedman, and L. J. Savage, "The Utility Analysis of Choices Involving Risk," Readings in Price Theory, ed. G. J. Stigler and K. E. Boulding (Homewood, Ill.: Richard Irwin, Inc., 1932). . 1 Ln..~—-.-¢ -b-’ ..;v.,Aw-.r: ~ v- ...”a .‘v .4..u — -' . -n‘ . I“ . LIA "' u y “ w;~.»‘.,_ “’»l ‘, - 0. ‘ ." r {a — _ —‘— a. _. ‘ a “‘v .“-' a... ., I..- v . . —' 1" ... -.-"‘~t -. \4 ~' .' i s. ' - h. ‘ v..- 4 n -F §~ I ~_ 4‘0".“ n-._-. .. .dg- 4" - . ~ «xv. . - 'V- AF. . .n \ -.,. . o . . '.. ,. “" 5. .. ‘ l- W, ~lu“ - o u 0 F- ‘~'I.‘ vgg"- a 'v.. a, -~‘." “~:- 1* ‘.~‘ ‘u,j 1h gambling and insurance. In addition they react to other economic phenomena involving risk. The hypothesis proposed as stated by Friedman and Savage (Chapter III uses the von Neumann--Morgenstern notation) is: In choosing among alternatives open to it, whether or not these alternatives involve risk, a consumer unit (generally a family, sometimes an individual) behaves as if (a) it had a con- sistent set of preferences; (b) these preferences could be completely described by a function attaching a numerical value-- to be designated "utility"--to alternatives each of which is regarded as certain; (c) its objectives were to make its expected value as large as possible.1 ‘he conceptual experiment for determining the utility function offered by Friedman and Savage is not the one used in this thesis. However, the procedure they offered, although somewhat impractical, is an alternative to the one used. They suggest:2 Select any two incomes, say $500 and £1,000. Assign any arbitrary utilities to these incomes, say 0 utiles and l utile, respectively. This correSponds to an arbitrary choice of origin and unit of measure. Select any intermediate income say';600. Offer the consumer unit the choice between (A) a chance a of ibCO and (152) of 21,000 or (B) a certainty'of‘téoo, varying a until the consumer unit is indifferent between the two. . . . . . . In this way the utility attached to every income between $500 and {1,000 can be determined. Friedman and Savage say that a utility function obtained in this way can be used to compute the utility attached to any sets of possible monetary outcomes and associated probabilities and to predict which of a number of such sets will be chosen. The function they hypothesize to describe the utility of money income has the following properties:3 1Ibid., p. 267. 239m” p. 292. 3Ibid., p. 303. aavqficvs ,- f. vi...‘l, ~_ ‘ I ' “‘ ~v-p » ~. a“ u- ..4. -.._ _ 'v-,~._-C r‘ -cntv.,,,. V '7“! ;vq~-.- -\.4 a . -o..y,,'u v . '"""v~A ... . ‘ ‘ ‘w_.e..‘. A»- W- - ”-o.. D “ . . _ n- u_“ . . We... :“’;v~~'. .‘u_ .,|' “ b ~-.-~ ‘ .. ~..,.- ~’ . I -_ ‘v‘obv__. 3“ ._’ (‘_ ' “fi‘u. , a " — . ~ U~.§4_ ‘..‘. [A4 “\- : ‘3 w.“ ’u u" ‘ I ‘u‘ .1'.‘ -\ _’ , .‘ a .‘3 H ...._,. U‘ a? ' v VA. ‘ .. VA ‘ ‘I‘: ‘g :‘ Av. _ ‘.v, H “.- 2": .' a ' 31a . “ “Q ~".»:. Y‘ A‘C‘L -. 1‘ s h .1 (a) utility rises with income, i.e., marginal utility of money income everywhere positive; (b) it is convex from above below some income, concave between that income and some larger income, and convex for all higher incomes, i.e., diminishing marginal utility of money income for incomes below some income, increasing marginal utility of money income for incomes between that income and some larger income, and diminishing marginal utility of money income for all higher incomes.1 One of Johnson's contributions to the field of farm management has been his recognition that the Friedman and Savage hypothesis concerning consumer behavior in risky situations has applications to many of the 'risky events occurring in farming. He pointed out in a book2 written jointly with L. A. Bradford and la er at the Bozeman Risk and Uncertaint" (I a Conference3 that farm managers need not have either a positive preier- ence for stability in order to insure or a preference for gambling in order to engage in risky enterprises. Johnson noted that all that is necessary, according to the Friedman--Savage utility function, is that (l) the disutility of losses in assets or income increases at an increas- ing rate, and (2) the utility of gains in assets or income increase at an increasing rate. 1In Chapter V only the first two stages are derived. 3L. A. Bradford, and G. L. Johnson, Farm Management Analysis (New York: John Wiley & Sons, 1953). 38. L. Johnson and C. B. Haver, Decision Fakinngrinciples in Farm Management, Kentucky Bulletin 593 (Lexington: University of Kentucky, 1953), and Ge L. Johnson, "Learning Processes: The Individual Approach," Proceedings of Research Conference on disk and Uncerpainity in Agri— culture, Bozeman (Fargo, h. D.: North Dakota Agricultural Experiment Station, 1933). -‘v ....‘ 16 In reference to farm managers' behavior in gain situations he says: The action of a considerable number of farmers, everywhere, imply that the; value gains in income-~producing ability at an increasing marginal rate. In technical economic terms this is the same as saying that they have an increasing marginal utility for income and assets. It is this type of belie“ that led Johnson to undertake a study of farmers' managerial processes including a test of the Friedman-- Savage hypothesis. The Interstate Managerial Study, to be discussed in the next chapter, in cooperation with other researchers is the first large scale attempt at studying these processes. It is also the first time that quantifying of utility has been attempted in the field of farm management and used to describe farm managerial behavior. Thi latter phase of the Interstate Managerial Study will be the direct concern of the author in this thesis. Statement of hypotheses The theoretical construct that man maximizes a measurable quantity called utility makes it possible to deduce the hypothesis that there is a means by which utility can be quantified. A second hypothesis can be deduced from the notion that managerial behavior can be predicted from the numerical utilities. This hypothesis states: There exists a correspondence between the numerical utilities derived from this technique and such characteristics of individuals as age, number of dependents, years of farming experience and place of residence. A further cause and effect association exists between the numerical .-6 '.....- ":v a»- -.. . n-~--- «- - w n '4‘... _.. --.~ wov' ,_ .._,_‘7, M‘ V.a_p a - v‘. r ‘-.—‘.‘. (1') Va '1“. ~_\ V " ‘54! ..~.~~_«- | l . . ‘ . i . .4 A, ~ _ “u, : fi. V"‘~ . \ d r‘ ;_“‘ ‘. ‘*-k . '_ J I ‘1‘ u‘v-As ‘V‘:,_.[ ‘ _ —‘ h \v._ . ' \— ‘ ~fi..,m. :‘ -1 \ \v\~ \ ~ “ , ‘K‘K \‘ d ‘ ‘.. A ~ 1 \7' ‘V‘. : S‘s. ‘. v" - “ ‘\ -.. \s ”‘4 ‘ l 7. utilities and certain managerial behavior such as income received, net worth position, debt position, and behavior in other situations requiring managerial action. By relating numerical utilities, common characteristics and managerial behavior of the past and present, future managerial behavior can be predicted. ‘grianization of Thesis In the chapters whicn follow the original objectives and procedures J1 of the Interstate Managerial Study will be discussed (Cha ter II) and to the technique of quantifiying utility used in the Interstate Managerial“xax\\ Survey (Chapter III). Then the ef ectiveness of the Interstate hanagerial Survey technique in eliciting answers will be evaluated (Chapter IV). Next the relevant data for testing the hypothesis concern- ing the significance of the numerical utilities will be presented (Chapter V). Finally, a summary and an evaluation of the technique will reveal the significance of this method of measuring utility and suggest precautions to be taken in future research concerning the use of measurable utility in explaining managerial behavior. The implications for farm management teaching and extension are not given in this thesis. To adequately accomplish this task would involve displaying various ethical propositions,and deducing from these and from the statements of fact presented in this thesis, recommendations useful to farm managers in solving their problems. These recommendations can -.- A H J..- -\ lo and should be presented in other literary form than in a thesis. Furthermore, publishing of these recommendations can not be undertaken by the author in a scientific role. The capacity he will be serving when writing the implications of this study is, a non-scientist, a policy-reviewer. ..... - , _. . " a, v‘ .‘. 7 7 ‘ ‘V-d .- L4,. ‘ _‘ .- ‘ Q— r . ~ "‘ _;° - :- 5-..-av-.‘ " 0‘ . '_', ,,, ~. "‘-’ -’.‘._.. a.-‘ .‘ . ,‘ C .4 m... .... _ V...» 4 \‘ ‘ .\, ‘Jfiu_‘_ ~. _ ..--, ‘ \ . ‘_~ -. a " ‘.~ ‘..‘ »_.\~\_v~ - , Vg\u-‘~ h ‘ v fi." . - the. V ~ ‘A a..(.., V“ 5" ‘ ‘ Q. r 4‘ ‘¢ . --_ A \'I V._‘: ~N .,_ v‘ V -~‘_ Cw ~,_ ~ . dk“- CHAPTER II SOURCE OF DATA Origin of Interstate Managerial Study Jr... The Interstate Lanagerial Study, henceforth referred to as I.M.S., is based upon the ideas and concepts stated in Johnson‘s and Haver‘s bulletin called Decision-hakinggErinciples in Farp#ganagement.1 The main contribution of this bulletin is the concept that management may be viewed in a functional-situational framework. The five functions that management is thought to perform are:' (l) obser- vation, (2) analysis, (3) decision concerning the problem under consideration, (u) action taking, and (5) acceptance of economic reSpons- ibility. The situations in which these functions are carried out are V characterized by changing conditions. The varying degrees 0 H) knowledge concerning (1) price structures and changes, (2) production methods and reSponses, (3) prospective technOIOgical developments, (h) the behavior and capacities of people aSSOCiated with farm businesses, and (j) the economic, political, and social situations in which a farm business operates result in changing the conditions.3 The five degrees 1G. L. Johnson, and C. B. Saver, Decision Lahin;_Principlcs in o ,—. ‘4 A. Farm hanagement, Kentucky Bulletin 595 (Lexington: UniVeislty of Kentucky, 1953). . L. Johnson, "Needed Developments in Economic conomics (Vol. 32, Nov. 1950) p. 1151-;2, 2Ibid., p. 8 or of. G Theory," Journal of Farm 3 3Johnson and Haver, pp, cit., pp. 6 and 9. 19 . 'V“ i . ' 40,- u- ..4- ‘ b -vfifi" ‘lv ,4 . _ up] -ounv- . . ..---.-r.. ,- u-uddv-~... .. , p.‘.«,. ,__ _. L:-~..;..'4..- - "“P\.\- R, ,. " ‘ a 'l.’“--’M - A Q - _‘ N 0...; 4.. - . . ’ " av~~- ...i . “‘ ~-~-:.-.. "v-‘v, “»a~__u - 'I"‘ , ‘n—., ~ ; -x.‘ -:.-_ a.” . A4"‘ a n. f n._ n “ .4 "w~.. u.‘.. DA . J. ‘ , 5 '1",- . ‘ w-“ ~UtAl_: 0 cl‘ r U_ u c‘.“‘ ., Au. ' 0‘- ""‘U “g - ““ V K ‘w ‘ ~ «‘_: -\ ‘.. .w 2‘4 ‘5“ M“ -3 (\D of knowledge that are delineated are:1 (l) subjective uncertainty, (2) inactive situation, (3) the learnirg situation, (h) forced action situation, and (5) subjective certainty. These concepts which form the background to the study have been subject to discussion2 and some empirical work. The most important discussion of these concepts from the standpoint of initiating empirical study occurred at the tie: and UHCBI tainty Conference at Boz nan, Montana, in 1953. Altiouga Johnson presented in formal meetings some hypotheses that could be empirically test ed, 4 various interested individuals at informal meetings did most of the conceptualizing for the I.h.S. Aiter consider aele discussion, an inter- state survey was decided upon as a means of obtaining data to test the concepts set forth in Johnson's and Eiaver 's bulletin. The sections that follow will be concerned with the operating details of the su vey. Interstate Kanagerial Survey The Interstate hanagerial Survey was conducted in seven states and obtained a total of lOTS interviews. The seven state institutions 1Ibid., pp. ll-lu. 2Procec din; of les>arc1 Sonference on hisv and _Uncertainty in Agriculture, Great Plains Council Publication ho. ll (rar‘o, N. D.: North Danota Agricultural College, 1955). 36. L. Johnson, Rana 'erial Concents for A‘“iculturalists Bulletin 619 (Lexington: University of hentuc y, l)? u)—' L. Johnson, "Relevant Theories, Concepts and desearch Tech— niques; and I earning Processes, The Individual Approach." Proceedings of ReSearch C nference, Publication No. ll (Eargo, N. D.: North DaKOta Agricultural Co llene, 1955 . ...',.' M”... L...” yL-:. -. c-‘evv r~lr u'-V' V. .. J..— r . . . ‘ - »~ .- ' . w) “H... m. P..- I‘sA I ' ".mo Vu--.,_.’ ‘ V I (S‘Wf‘ngq r - ”dfi -ku b- A0 9 -2 '” \_ i‘ H... .v.- v. :r -- ' " ~ PW < - -“ “~Vds _ .. ‘- , ~fl~.-‘~‘ ' I.v .‘\ ' 7‘ I‘V“ . j v- . ‘— - "p- .. N 7‘“, '( v ‘4‘, y - v.‘ n ' - “"";“:'r' n. . Q U . \ 4‘ :“e. . fi.‘ V‘V-‘Afi‘ . N IQI v. vac I- .g .‘ \ J:~‘ . ~ .. . "Vb-“n, ~... C .. . . 21 which cooperated on setting up and running the survey are: (l) Univer— sity of Kentucky, (2) University of Ohio, (3) Purdue University, (h) Michigan State University, (5) North Dakota State gricultural College, (6) Iowa State College, and (7) Kansas State College. The services of the Farm foundation and the Risk and Uncertainty Subcommittee of the North Central Farm Management Research Committee were utilized in establishing the c00perative relationships. Michigan State University, as originator and a primary sponsor of the survey, arranged for and contributed the services of a survey expert for use (a) in constructing and pre-testing survey schedules and (b) in training interviewers. Schedule Construction and Pre-testing The development of the schedule used in the Interstate hanagerial Survey proceeded through four stages: (1) a proposed list of questions including the objectives and hypotheses to be tested, (2) a tentative schedule showing further design of the questions, (3) a schedule for pre-testing, and (h) the final field schedules. The preposed list of questions, objectives and hypotheses was prepared by Glenn Johnson and the author and was presented to the Risk and Uncertainty Subcommittee of the North Central Farm Management Research Committee in November, 1953} After considerable debate, the committee agreed that the subject area was well enough defined to proceed to the question design. For this purpose the services of Joel Smith of the Michigan State University Sociology department were 1Unpublished report, November lh-lT, l953, at the Farm Foundation Office, Chicago, Ill. q-;‘-\~~. 5 J “-0-! u.»'.;..._ "- ‘I n -v- . .- 3‘ V-nd vv.-..~v- - :,“.:""‘3 -‘n‘ Uv-Au‘dgq '4‘. ' -A ~v ‘1’. ~1 as ‘.-;‘ ._ . 2‘3th ."‘2 c v.4. - ‘VU“V U “3’”wa - - e'vv‘ .u v, “in V'-‘ ‘ “w’. ‘x ‘ ”é:g:s #0 - . a _ “-11,.1 ‘, r; “\A. ‘*§q ‘ ‘. “b - i ‘. ‘:~ '. ‘M " 1..” ~' ‘evct‘t ° ‘1‘.’ h.,. v" a ‘ — r~~ : as-” shit. .- ‘- ‘3‘ - “*5 a ~,. ‘DV .. - . .. v- 1" > . ‘HA ‘.h~ A~:“P‘r~ ‘ ‘H‘:¥. ‘0‘” . ~.~_;; ‘ "I 7. ‘ 3' n. ‘v‘ Uh _‘ - _.C .. “W ”A”. fi‘v ‘7‘? ‘ V I ‘5‘. 8-3:; V- contracted. With the aid of Joel Smith a tentative schedule was con- structed. More emphasis was placed on design and wording of questions than in the previous proposed list. This schedule was presented to the subcommittee in March of l93h.1 After careful scrutiny, and after the committee presented its revisions of the questionnaire, a third schedule to be used in a pretest was constructed. Copies of the pretest schedule were sent to each of the cooperating states where the questions were subjected to field conditions in the respective states. The schedules complete with farmer responses and the interviewers' comments on the effectiveness of each question were returned to Michigan State for further analysis. From this valuable pretest material, a final schedule was designed. The results of the pretest based upon a criterion of workability showed that the total list of questions which required an average of three hours to answer was too long. The belief that this length of interview would cause some of the respondents to become fatigued and disinterested resulted in a major change in schedule design. The pre- test also revealed that certain questions were ineffective in elicit- ing responses consistent with the objectives and thus were either modified or eliminated. The total list of questions which appear in Appendix A was pregrammed into six shorter schedules requiring an hour to two hours to complete. Each smaller schedule was intended to com- prise a unit in itself, that is, those questions which bore a close 1Unpublished report, March 23-2h, l95h, at the Farm Foundation Office, Chicago, Ill. M.. v A "f' ' fin , ~ F ‘ -- OL.‘ J‘ A . -_. ““C' u .. ‘ IVV..-o-\t’- -‘ . ’u‘t '~ U «.. - Qv-fi;r «C ‘f‘. ign‘yu ‘4'. V‘ I ~--. ~v ‘ , ‘ . :V‘HL, \1/ S‘ "Wards“ ~. Vs‘v.o v.¢u.u ‘— . , ‘~» _ _ ‘ ‘ C rh.‘ w...) b-\¢..~ _ '1“ _ O , r‘ a “u “I”: . c. “,3.’ ' “""r .- ,._ ‘1. '. I‘d- ”“w : I. ‘ '1 t. ‘ ~ ~. h ',. “.3 ‘ . s44 v.4._a ‘ ‘ . V“\~- u . ‘ ‘ ~ h~‘ __ “4‘ 1 ”- V .‘L: ’ . . ..: :5. TF‘.‘ ‘Ved . ‘. I’~- . -' .nn ‘ “U h - I «A “‘5 v:“.l"L_- a”: .: . ., . A k‘lb 1dr: ~ v Av “‘ . I . ‘r v. Q \‘Vv‘ la- ‘_ ~ .9 U4." -‘|~ ‘ s N LC 3"“ . "~4Wr~ .. "‘ I ~. \ \k‘ A; v ‘ q ', . u£ relationship to each other in terms of the hypotheses to be tested appear on the same schedule. The distribution of the questions over the six schedules appearsin Appendix B. The order of the questions on the final schedule was also carefully considered. The sequence of questions attempted to follow this pattern: (I) a few easy, single reSponse, attribute questions, (2) those open- ended questions which required a free response without prior inform- ation,1 (3) some ranking questions with information aid cards, (A) some open-ended questions concerning Specific decisions, (5) open-ended questions requiring short answers, (6) questions requiring a "yes" or "no" answer, and (7) further attribute data including gross income, net worth and debts. The order of questions on the final schedule is also shown in Appendix B. Interviewer School In June lQSh, an interviewer school was held for one week at Purdue University. Joel snetn with the assistance of Glenn Johnson and the author instructed the interviewers of the seven participating states. Iowa and North Dakota had representatives at the school who later instructed their interviewers. The purpose of the school was to acquaint the interviewers with the study, the survey and the schedule, to instruct the interviewers in the proper techniques of interviewing, and to supervise some practice interviews under actual conditions. A mimeOgraphed review of the objectives of the study, the intent Of the questions and sampling procedures helped acquaint the 1This order was such as to avoid "build in" answers. -.— ‘5 May wflfl" ' .~.. filthy-‘5" A ‘ . . ,V..-. -— -..¢‘0.... _.. .’-\ . [.1 _ s ....... \’/ --.~v--A- vvfi’fi ---.-e.\, -. . ~(‘ A'-. ‘ -5--~|.‘__ .,‘ - .‘u‘ Q-'~ - 9.- I .h‘- l\_ 5,, q-A,A- . « ' v ' ‘ e. ".-< u-.. v . " ---u.-. ,-_. fina— ...Q n .. i _ m§.~ : :1. ‘ ~~ 's-s.. N I x -' U. x.._. vxv L'. . ‘ I .V ‘v- ._ ~" ‘. >.*’-‘ ~.‘I """~;«y - Q ~.. «- ~. vu" ’~ ’~§ x. a ..v ” .J_“~ AI. R ‘..- . U. W. ». "N", ‘ Ci ..‘~ ”v . 1. ~- a, ~ V. C.» " A ‘ .9. - A‘iw. '1. ' LC“"r . . « .y‘ s 4‘ ‘1. \“‘q ,. 4...‘ P‘s '.. S "v. .a" .'.‘ "- “5". i - p . \F A «.“j l w" .‘ v \ v..d 4' .I' ‘R.._ .,r' d ‘ ". C‘s- ‘ .; . ‘7‘ -3‘ . 'w ‘3‘ '. \‘.v interviewers with the study and the surveV. The schedule was fully discussed A mineograpl ed set of instructions which included (1) general interviewing instructions, (2) general instructions for the schedule, and (3) instructions for specific questions aided in explaining inter- viewing procedures. Following the formal instruction, each 01 the interviewers com- pleted a schedule with a farrer in the vicinity of Lafayette, Indiana. Joel Smith discussed the reSpcnses and reactions obtained from the respondent with each interviewer. Further instructions were riven to C.)— J.‘l those intern iewers who had difficult; on the first interxi ew. In some cases a second interv ew was taken which was again e i>weu b" Joel Swith. One primary objective of these practice sessions was uniformity in interVicwi. Statistical Sample Representatives of institutions cooperating with the north Central Risk and Uncertainty Subcommittee Specified the area and units .‘ to be samp ed. The area consisted of eight geographical regions con— taining contiguous croups of waole or part counties located within the seven states. The units to be interviewed consisted of r ral commercial farms (census definition) with gross income of tZSOO or more and which have single hous- ehold mana: erial units. Farn.s characterized by live— stock share leases, father-son arrangements where both have a separate family and household, and re~ular ousiness partners ips between two unrelated individuals were ineligible for interview. I a. ver- ‘ gouflv- .. H. ‘ . ...3v---—v~ ..- 31 D“ n--- ‘,‘ .. ‘ _ *bvd. dv-«- A.. “9:; fl . -- '5-4 '4. ......7, . \ 'n---vr~ r a ,,_ v " - A -. " «“rp ‘ -, \2 .q,._._l~ ~_ I d .... ~_ A..- -V..._ .. . \ .- l “J y] - -.“’ an. ...c‘ _ . .nfl "-\..-. A- . ~"' .. \ ‘ .p ‘a \- ~._~ '\ .4. “ .‘_ ‘a‘- ’ 7"”: 'w. I“ - u~-~ .. V “O“ h“ -\ ‘1 E n ‘A.. I e..- ‘7‘“- . ..._‘-_ I “ ~u.\.. ”New. . V... ... 1.,“ -. . \.l, “' ‘u n ‘ ~ ‘ "v. ‘u.. “.74 "—«_‘. . V :- ./ .‘ “ l ‘| '-.,‘ 2 ."I . "k; ‘ . *‘l \, ‘«.“ Q -. . ‘ L." V fin“, ‘ 5 ‘5- ., . A .‘, ' s . n .n. . ~..: -. -, ‘ I l ."\ ‘- x s, \4 “A The statistical laboratory at Iowa State College desirned a .3 stratified random sample of area sampling units. Each 0 f the ei ght areas was a stratum and each stratum was subdivided into area sampling 0‘ units. Each unit was ei acted to contain two eligible farms (in the case of Kentucky sampling units which contained an average of three eligible farms were used). The sample drawn was completed using the 1950 census of Agriculture and the 19h] devised Raster Sample Katerials. The following procedure was used in selecting the farms to be interviewed: l) The number of eligible farms present in each whole or part county was determined. (Number of 1950 commercial farms with wross incomes of 2500 dollars or more, less the number of livestock share leases and 20 percent in order to adjust for partnerships, father—son arrangements and changes in the number of farms since 1950). 2) The total number of area sample units with two eligible farms within each county was determined. 3) Master Sample Materials were used in subdividing the county into area sampling units of the desired size. h) A random sample of the desired number of area segments was drawn from each stratum and these segments were numbered and indicated on one-half inch scale count; highway maps. The sampling characteristics of the eight strata and the number of interviews taken are shown in Table l. __ ——.-—-—~. -... V“ - s “ ‘nnp ‘ A -‘.a- .,I . ‘A"‘ ‘0‘ .4 "w‘ . ‘ ., . "\ Q"‘- .‘U‘ -fl ‘ \v - F“-'M ‘ "l‘ .1. .u _. :_ ,, . - 'w “h; I“ . we“. ’-Q .4 w.‘-~._ ‘_"_ V ~-. \ '~x V~;*h C‘ r ‘ I - v“ U -4 _ o‘vv "Lina . u U.' |I ‘1 A S: -. , V F”- K.‘ 26 TABLE 1 CHARACTERISTICS OF THE SAMPLE OF EIGHT STR TA FOR Th3 IN‘iRSTATE hAflAGEiIAL SUBVEY ——— m “.~v~“»- - q..-” flu...“— v- ~..—.-— *. Estimated Estimated Expected Actual Number of Eligible Numoer of Number of State~ Eligible Farms per Interviews Interviews Farms Sampling Taken Unit Kentucky 1,790 3 150 12h Ohio 23,599 2 200 137 Indiana 15,769 2 200 189 hichigan 37,5h5 2 22h 199 Michigan 39h 2 30 30 Iorth Dakota 9,30l 2 150 l2? Iowa 23,6h9 2 140 120 Kansas 6,985 2 206 1&7 Interviewing During the summer and fall of lySh, twenty-three interviewers in the seven states contacted the eligible farm managers. The interviewers were instructed to adopt the following procedure: 1) A segment map should be copied onto the applicability sheet and the locations of farms should be entered. 2) A drive through the segment before starting should allow making any necessary changes on the map. (Each prOSpective interviewing location should be given an identifying letter in sequence.) 3) All apparent farm residences in a segment should be visited to determine whether the occupant qualifies according to the criteria stated above. h) All prOSpective reSpondents should be accounted for on the applicability sheet. (A total of three calls should be made, if necessary, to account for a potentially eligible farm.) 5) All interviews should be numbered in the sequence in which they are taken in addition to the segment number. 6) The six schedules should be rotated in sequence. (When a sample member is not at home on the first visit, reserve the questionnaire until the interview is finally made.) 7) When 10 to 20 interviews are completed, the schedules should be sent to Joel Smith for review. Joel Smith examined the schedules as he received them and, if necessar , made suggestions to the interviewers on how they might improve the quality of their completed schedule. This type of control was intended to produce m re uniform interviewing and to insure against unfinished schedules.' diin39Proqedures After most of the interviews were completed, the personnel at hichigan State University had the task of constructing a code which would make it possible to transfer the data from the schedules to IBM punch cards. This task advanced in four stages: (1) preliminary code construction, (2) revision and testing of the codes, (3) actual coding, and (h) cross tabulating. Ira DFA‘ - .. mn‘ ,..' w I I‘ a "0 n7 -7”. . ~ .‘n‘-~' * .-.l l u - . ‘ '\r O ‘ 4 . w- o “A .I‘ ‘ oinau UK; as" a. . “‘1. V‘ - .- . l ‘l~.- “ “7'\ --§ . ‘ ’ s -— ~‘q ‘_ fiwv‘ » .f ~_ ’ L. .. \. :~ . «.‘ a ‘ . ‘ v' -____ -. -~ - "~-.:. 1‘ “wk . V . \-L V'”, . Q— A ~ , - «“5“ Q“ "3 n." .. " "" ~._ UV " .5 ‘ ‘ "'v b-t“ -- -. ”H. V.~._ ~ ~ ‘ ‘W.V: 'D . V n. ..b - u.‘ 3-1 "rQ- 'V-F; Una”: . *- u. ‘a rhf -I’ .' ‘Vvt’ ‘ .._M ‘\ ‘4: no, _. V » \‘. - . The first step in the preliminary code construction was to type a large number of the responses to the open-ended questions. With (1) many answers to one question before them, and (2) the general theoretical bachground of the study in mind, Joel Smith, Glenn Johnson, and the author proceeded with the second step of defining categories to which numbers were assigned. The answers could then be represented by a number. In the case of the attribute data code numbers were assigned to intervals of numbers or to the actual number given as an answer. The preliainary code, constructed in this way, was presented to the Risk and Uncertainty Subcommittee. The code was revised according to their recommendations. Later the sections concerning open-ended questions were subjected to reliability tests. To test the code a definite procedure was followed: (1) Two persons would code 15 or 20 actual questionnaires randomly selected from the seven states. (2) The code numbers assigned by the testers for each item were compared for agreement. (3) When the numbers did not agree, discussion. of the reasoning followed in coding the item led to one of the members changing his mind to agree with the other or to both agreeing to Change he code. This testing of the code provided the necessary background to instruct clerks in the coding procedure they were to follow. The coding of all the schedules was carried out in three steps. First, Joel Smith and the author taught the clerks by acquainting them with the code and then instructing them on the proper procedure to . I .. - I-~-‘—p~~y~.. ' ' ..- &:a--.oo ,- 7. g y"..,.., 3v. A a .a- ‘ .4 ‘ ‘JA "vL‘..L y . ' — . a. C"‘ 1' ‘ . U- a..;_ ...._. ‘ . O I'""‘ A -- “‘_. L‘ . y..\/ ., ‘ o I‘.A‘\' “d-.. '1” . " -‘." ‘ '»‘-‘A fi:v:~s “:= A“, «*4 “*9 V‘ V YA- :._-‘ _ ’ d I A..V 15" \h-.., ‘ -.- ‘ . b - :‘f—C“V~ . ..4 ““‘-L.~ 4‘... 9 ~-. ~V: T A- " iv.~. U ~: ” A.-J_ a- .‘ U~“ ~l. V. H'r-r ‘ .. a“ v“: 3' «uh. C‘ .\ _. .kv‘t‘ “I . “\~ “1‘ t '71 _ \v. I. H ‘- -“-' a“. “27";4.‘ VI. HO W. 5‘. . Li“ N‘ a. - "‘v° [~“‘ ‘ vv,.."_ ‘ [‘0 ‘0 follow in assigning numbers to the answers on the schedule. The next step was the coding; this involved reading the response on a schedule, deciding which item in the code best fitted the response, and writing a number corresponding to the item on a code sheet. Checking was the final step; this involved a repetition of the coding procedure by a second person. If the two clerks did not agree on the code number to be assigned to a particular response, they discussed their reasons for their choice until either they agreed or a third person was asked to make the decision. The final stage in getting the responses from the schedules to IBM cards was cross tabulating. After all the lists of responses which did not fit the code were either (a) fitted into a broadened category of the existing code or (b) included in new code items, the number of items that would appear in a particular column of the IBM card was checked against the number of items that would appear in a related column. . When the coding procedures were complete , the tabulating depart- ment of Michigan State University punched the code numbers into IBM cards. The data on each schedule required a total of hSO columns on six IBM cards. The punched cards were again checked for interrelated punches between the columns for each question. After initial marginal tabulations were run these Checks were repeated on punch and column totals. a l ’ “ g; 11.. 7‘ my: v -.. o,‘ .4 ‘ W Vuv . .s. ,zsn _,' . ‘v... ,-‘ -.\..‘--\‘ . a..~..1.lc. u ax '- V _ A” p n‘v—Ck-vg. v . :fi- -,.. Q‘ ‘” ~A V.nv Q‘_ n '- .A‘A‘n‘- -.,. _‘I_,‘_' F ,1 .d - ‘VVIJ U..a u.--.' . Q .“3 nw' . , _ r . I -. _ h"" “0 an-...- ‘g . _ . ’ an . -[ C we - .. .- m- u¢g-_.,"l .v 's- ‘ - V ~v.— ‘1‘“... pl w ... ~.« g . u —C C 4 . - ,. “‘~—b. ... ""‘ k\ u .' ' val“ ‘fi 5"». . d ‘ “‘ 'V-ov w- .4. ‘_ .. ~0...S. s Wru‘. . "I O “3.-.“, .“flh‘, n. 3., , “" v‘u ‘ Au‘i ‘ _ . s-..._ "C: _“‘h ”an. «cc — “""’~' 30 Reports and Criticism In August 1955, a report of the progress of the I.H.S. was given to the American Farm Economics Association at their annual meeting. Harold Jensen discussed the nature of t:e study by pointing out the relation between the managerial concepts develOped 0v Johnson and Haver and the survey questions.1 Haver of the University of Chicago Spoke about the universe of farms studied2 and Joel Smith discussed some of the problems of methods in the I.M.S. survey.3 These papers parallel and supplement the content of this chapter. Glenn Johnson presented a paper of more direct concern to this thesis.4 He pointed out how the Friedman-Savage utility hypothesis was used in constructing the technique dealing with rains and losses in the I.M.S. survey. Since there were no data available at the time of his report, no evaluation or conclusions could be drawn concerning the effectiveness of the technique. lHarold Jensen, "PrOgress and Problems in Decision Making Studies; The Nature of the Study," Journal of Fa“m_.hconon'cs Proceedings No. 5, (December, 1955). 2C. B. Haver, "Progress and Problems in Decision Making Studies; The Universe of Farms Studied," Journal o£_Parm Economics Proceedings No. 5, (December, 1955). 3Joel Smith, "PrOgress and Problems in Decision Making Studies; Some Problems of Method in the Interstate Managerial Study," Journal of Farm Economics Proceedings No. 5, (December, 1955). 4G. L. Johnson, "Pregress and Problems in Decision Making Studies; The Friedman-Savage Utility Hypothesis in the Interstate Managerial, Study," Journal of Farm Economics Proceedings No. 5, (December, 1955). a n f - - vhf! ,- "V C r ‘\ \v.. x. '-~J—vo.~ - — . P""‘ A» -‘~ —‘ ‘- oo.4uv (L. _, ‘~ ' “ ‘.- “\r r “. '. I 1. 'a‘d‘C‘ .._ _ _ . U l — ‘- F' .2 r3- \ ,. v. bvo-u-\". ” ~ _ . P‘n- __ . ._\ ~a.v """-"4- ‘ ~ --‘ "~" .- 4 ‘ d n.- J #v..v“.“ --., I ‘Ar.'-" ,. ‘A- a- C N..- v \m ' --4 “~d a- ‘ -...f -..-_' V I v ..~\-:b-\. "“5: 0v ’l“-.:'\-?"~ _, '_V V»_,‘, .-‘ V > :Y‘ $~~._~ ‘.. ‘3‘, - ,_ nu'n V.“‘ v'n ‘ ~ P I. ‘. ‘v *-. n" ‘Q I "‘v.-. ‘ a’A ». “\U‘ V'f' r I. ‘ us.“ a. v ‘1‘ r... ‘ \r‘ .u,.‘.v. I P‘-.. — . “1:1 “ n {'1 _ h .. U. D ‘r - Q ~ 4.3 A -. ‘~~V. Ln... 5" "' ~ , v '5 z. .. ._o ‘ \ ‘1’- \ ?‘ ‘ ‘ I. . ‘ 4. L. ’II- V‘_ V v ‘V‘. “w: mu . ‘~‘:; ' | ’ ~§ ‘ ‘ ‘— .“ ‘..n ' I . u) ,_‘ , ‘Jh . s. - ‘1‘ 31 C. F. Sarle has criticized the total study in two published papers. In the first, he points out two conceptual difficulties that he en— visions on the basis of the list of questions used in the survey. These are (l) the study assumes that the decision process is an indi- vidual process rather than a social one, and (2) the study imputes decision processes to the individual that are of a nature foreign to the 0‘ (1) st findings of psychology.1 In reference to the questions on the schedule he believes them to be phrased in terms of an abstract managerial decision pr cess lacking in empirical refe'ence. In his second paper, Sarle further criticizss the queStionnaire for its use of abstract terms and concepts. he emphasizes how important communication between the researcher and the respondents is for sound socio—economic survey research.2 In reply to Sarle's criticisms, Johnson and Smith claim Sarle's evaluation was premature in that he was not well acquainted with many important aspects of the study, including the data.3 In a second paper, Johnson states4 lC. . Sarle, ”Research on the Dynamics of the Farm Managerial Decision Process," Journal of Farm Economics, Vol. 38 (February, lyjo), r‘" r / f pp. 133-100. 2C. F. Sarle, "Comment on the Rejoinder," Journal of Farm Economics, Vol. 38 (February, lQQ6), pp. loY-lTO. 3G. L. Johnson and Joel Smith, "A Rejoinder," Journal of Farm Economics, Vol. 38 (February, 1950), p. 103. "IJ 4G. L. Johnson, "Kore Ado About Sarles Suppositions Regarding the Interstate Managerial Study," Journal of¥§arm sconomics, Vol. 58 (may, 32 . . . (l) we have been aware of many of the dangers stressed by Sarle since early in the design of the stu‘y, (2) we have take many positive steps to avoid these dangers, (3) we have succeeded in avoiding these dangers in some instances but not in others, the pattern of successes and failures bearing at best only a vague relationship to the degree to which we followed, ex ante, Sarle's suggested methodologies, and finally (h) we are trying to analyze the data so that no unjustifiable conclusions are reached as a result of failures in the questionnaire. It is important to note that in none of the above criticism of the questionnaire were the Specific questions to be analyzed in this thesis singled out. Thus it remains the task of this thesis to evaluate the specific questions, dealing with gains and losses directly or in- directly. A technique for quantifying utility which was made operational with the set of gain and loss questions will be evaluated in the chapters that follow. CHAPTEi III 7 RUDEL AHD m33hhl"tfi EUR nEASUJING UTILITY This chapter deals with the mathematical model that is the basis for utility measurement and with the technique used in the I.fi.S. to quantify utility. ' he liodel It has already been assumed that the nature of the manager is such that his aim is to maximize utility. Utility was further supposed to ‘ 6 be an undefined entity or an "intervening variable."l It is desire O that this variable be mapped or correlated to numbers in order to make it possible to use the numbers as a basis for prediction. There may be many such mappings and the passage from one mapping to another is called a transformation, the totality of these mappings forming a system of transformations. The description of the variable by numbers is said to be unique up to that system of transformations. The empirical relationships from which utility is supposed to be .1. abstractable is (l) preference among events, and (2) indifference ”etween combinationscfiTevents with stated probabilities.2 1Intervening variables are those which intercede between empirical relationships, i.e., the concept contains only words which are reducible to empirical laws. 2This model was proposed by von Neumann and fiorgenstern, Theorv of Games and Economic Behavior. (Princeton: Princeton University Press, 19a?) II' ’" ”II K») Le \JJ N4: 1. Thus consider a system U of entities u, v, m... In U, a relation is given u > v, and for any number 6! , (C- v is a complete ordering 01" U, i.e., u < v when v > u. A. For any two u, v, one and only one of the rollowing relations holds: 1) u=v 2) u>v 3) u-v, v >-w, then u >'w. II. Ordering and combining A.uv, thenu>0Lu+ (l-a.)v C. u w >v, then there exists an a. such that oLu + (l -o: ) v >w III. Algebra of combining A. au+ (l-ot.)v= (l-d)v+aLu B. a (Bu+ (l-B)v)+(l-0L)v=7u+(1-?)vw‘nere7=a-B Two important theorems that von Neumann and horgenstcrn deduced and proved2 from this set of axioms are: - w ———v—- 1An axiomatic system is a linguistic structure in which no identi- fication with empirical constructs is mace. ”The proofs of the interceding lemmata and theorems as well as the proofs of the two theorems listed is given in the appendix of von Neumann and horgenstern, 0p. cit., pages 017 - 052. *- -‘ld' -- .- “ =_ , V “at: U—- "" .. fi’vfl' v- —‘ .. I‘- . F 'd‘ ._‘\. . . Theorem 1) There exists a mapping of w-——> q (w) of all w on a set of numbers possessing the tv.o properties a) Monotony (not decreasing) b) For O.< ex < l and any u, v q[(l-o~4)U+06V]=(1-0c)q(u)+ocq(V) Theorem 2) For any two mappings of q' (w) and q (.1) pos ms in the properties a) and b) of Theorem 1) q' (w) = wO q (w) + w1 with two suitable but fixed W0 and w1 where wO >-O. The first theorem provides the correSpondence between utilities and numbers whereas the second theorem says the numerical mapping is determined up to a linear transformation.1 Technique of Quantiiying Utility The set of axioms I, II, II] and theorel ms 1 and 2 provides the ormal structure for obtaining a numerical utility. If a technique that incorporates the concepts of the arzioms can be deveIOped ior Obtaining data, then the theorems provide the basis for deriving a numerical utility function. The technique used in I.K.S. has, as its objective, the collection of such data. The remainder of this chapter deals with the construction and use of this technique. v 1M. Friedman and L. d. Savave, "The Expected Utility Hypothesis and the heasurability of Utility, " Journal of Political Economy, Vol. 60 (Decen1ber,l,$2), p. hob, pro sent the alternative theorem: There are numbers 01,...., On such that u <'v, ii and only if E uici.5 Evici. 1;oreover, any two such sequences of numbers Ci and ci' are connected by an equation Ci' = s + tci ior so.rne s, t, with t >»O. ‘1—.-P-’~"'9‘ ' Jyg‘J- — 4- n ..»- -. ,_-- C'. - .. .‘m .. 0‘...) I'- “f:~‘.._,. ‘ I Mh‘ “' ‘ "_L“ -Q\ . A. 'r-.-‘_ u... r- ‘ “-'—.. .‘\ a, 5‘.A-_. CH“ ‘ . - . A‘ I/V ~.. 5. ’\» S I -.\..'j v ‘. -- . _ l I‘ u + f‘ _ \. \ “i 1~. h ‘ I LA _ . g 'r I i t. v \_‘ U" \/ I l . J '5] ‘ ‘ ~w + '. \. H»- . ~ \. “v "n , U‘P‘P‘S- ‘ . ‘U ..-L‘ ' ~...L. W" ‘1 In > . . , b \ ‘Qh ~d > ‘3 a > ~, ‘ i J-x \ ‘~‘ ‘ - J A 30 Consider this situation. "If you knew that one person out of a group of hO would get a piece of property worth $1,000, at no further cost to him, would you be willing to pay 10 doll.rs out of your present income to become a member of that group?" This situation could be considered as consisting of three entities, (l) the utility of the present income position, (2) the utility of the possible 1,000 dollar gain and (3) the utility of the possible position of having 10 dollars less than at present. It is assumed that a person I could order these entities by the three relations (I) indiiference, 9; IT“ (2) more preferred, and (3) less preferred. The operation ygu + (l.-o4) V' W'iS made meaningful by letting: a) u = utility of 1,000 dollars gain b) v = utility of the position of having 10 dollars less (payment to play) c) w = utility of present, income d) c% = probability of 1,000 dollar gain = 1/ho e) l - on = probability of losing payment to play = 39/h0 In this situation, the operation would read l/uOu + 39/h0v >'w or l/bOu + 39/u0v < w depending upon the answer to the query. he transitivity axiom is given intuitive appeal by considering that if u >rw and w >‘v in the above situation, then it seems plausible that u >-v. Further, the ordering axiom seems clear when one considers u >-v in the above situations; then u >fil/h0 u + 39/h0 v since the two alternatives are mutually exclusive there is no reason to eXpect «‘w-r L .w‘..n.-« A nu.~n~. ,‘ ...fl-s,, 5 \"._.“ ‘Vflv. 1-... . ,k- L-uv ' n- - .~‘r"‘r ,\- ’ . .:..\-C- - U1 . -— . " " C .Q Li...» - _ V") r ,, ..- (1-, «.t--.- _. c ‘ 1 V. a..- - “' »- _. .. "d.' a: ”C L'_ - ’4” a. .. .. ‘ \-n,..~ . U.\_hv\; f .- ., v... dub.“ S ‘_~ ~ . A U." n_H‘—.,._ n .v “v.5”.b 3‘; A ‘Lcfifxx ~‘~"VV‘U~..":-". ‘ _ ,_'. .(1 ‘ ‘ ‘1 ;/‘!~.\ .4 ’1‘. ‘ , C «, , ‘15L " IL ““1 ". a) 4' g N M ,, U) " = 14 c“ S: n‘ ‘ ML“: VLJ _~_..: “‘5‘, Sr“). ‘ .u": 1.. ,4 . k v: Tfl‘. ‘--4_ - , "“-— complementarity between the utility of a 1,000 dollar gain and the utility of the position of having 10 dollars less. The ordering u> w > v surely implies the existence of an ocwith 04 u + (1 -oc) v > w; however, this technique does not attempt to find the exact. 04. The algebra of combining axiom makes it possible to interchange he two entities u and v in the above situation and obtain the same answers. The technique does not take advantage of this axiom. how'sup- pose there were two situations, one exactly lihe the above and another analogous except that a) size of gain = 5,000 dollars and b) 04 = 1/200 Suppose that someone answered "No" to the first situation and "Yes" to the second situation. Then the following operations would hold respectively: 1) l/hU u + 39/n0 v < w 2) 1/200 u'+ 19:;1/2oo w ‘.> w‘ Where a) u' a utility of 5,ooo dollars gain b) v‘ = utility of position of having 10 dollars less c) w = w' = utility of present income Since the operation czll+-(l.-CK) v = w is postulated, it must exist somewhere between these two situations. It seemed quite impracti— Cal to find the exact cm for each interviewee and the utility entities ‘Which would satisfy this condition. This was particularly true as an aiternative approximation to this condition produces the desired results. Thus, it was assumed that an indifference point exists within the interval between the two quantities of gain. If the true points of indifference are uniformly distributed over the interval then they can be represented by a point located half—way between the two quantities of gain, then the operation is 1/200 u" + 119/120 v" = w", where a) u" = utility of 3,000 dollars gain b) v" = utility of position of having 10 dollars less 0) W'= I' = w" = utility of present income. A belief that the large number of cases used would reduce the apparent inaccuracies of the mean values or of group data also justifies this procedure. The complete technique used in the I.h.S. consisted of two sets of similar situations, one set dealing with gain situations and the other with loss situations. A description of each of the situ- ations for the gain set and the loss set is shown in Table 2. The range of alternative gains fr m 500 dollars to 50,000 dollars, .e of possible losses from 100 dollars to 50,000 dollars were intended to cover the range of gains and losses which would be meaningful to the respondents in the survey. A previous pilot study conducted by the author and Chris Beringer showed that married college students would accept smaller loss situations than gain situations. Thus the loss situations start at a Smaller amount than the gain situ- ations. The range of probabilities from 1/2000 - 1/20 lor the gains situ- ations and from 1/2000 - 1/h for the loss situations was kept to TABLE 2 Desceirrion or GAIN .AND Loss SITUATIONS USED on THE IQTERST TE LAHAGEAIAL STUDY SChJDULBS Amount of Probability Expected Amount Types of Gain (P1) of Gain Gain.(czPl) of Payment Oddsa (dollars) ( 0L) (dollars) (dollars) ‘ Gains Situations 500 1/20 25 10 2; to hr 3 UR 1,000 1/40 25 10 25 to {F F 0? 5,000 1/200 25 10 25 00 33 r p 10,000 1/000 2) 10 25 40 n3 r 0: 25,000 1/1000 2; 10 25 40 LP 3 0r 50,000 1/2000 2; 10 2; 00 1? F Up Loss Situations 100 1/0 23 10 2; he MI F 0r 500 1/20 25 10 23 00 n; F Uy 1,000 1/200 2; 10 2: 00 Hr F 0r 10,000 1/400 25 10 25 no my F up 25,000 1/1000 2; 10 2; no me E 0r 50:000 1/2000 25 10 23 10 1:5 5 05 #7 —~‘ *‘ ‘v.— .- aThe three sub—columns under these two headings correspond reSpectively to more than fair, just fair, and uniair odds. All l0 dollar payments are ME, 2; dollar payments F and ho dollar payments UF. hO a minimum, to try to avoid the possible coniounding effect 0: the utility attached to probability distributions.1 The ran e of payments or stakes from 10 dollars to 10 dollars 10 or both sets of situations was intended to avoid the possib e disutility effect of this loss as well as to keep the eXpected gain or loss equal and small.2 The situations were arran,_;ed randomlJ on two sheets of legal sized paper with the loss situations on the first sheet and the gains situ— ations on the second. The iorms used are shown in Appendix .A. The words used in communicating the probabilistic situationwwmxaintended to avoid the connotation of "roulette wheel gambling."3 The interviewers (1) read an example4 of the loss situation to the respondent, (2) answered respondent's questions concerning this example, (3) aslce he reSpondent to check Yes or No under each of the situations presented to him on the loss sheet and (L) if necessary, rerramed the situations in more meaningful terms, i.e., using fire, Windstorm or other possible farm losses. A similar procedure was t11en iollowed with 1C. H. Combs, and 0. Beardslee, On Decision-Making Under Uncertainty, 'formulate an extensive model that incluCi es not only tile utility 01 the gain (or loss) and the utility 01 the "payment to play” but also the utility of probability distributions. Experiments patterned aiter t11is model would include the enti: e ra nge of probability i‘rom O to 1. This iormulation can be iound. in R. W. Thrall, U. H. Coonbs, and R. L. Davis, Qecision Proce esses (New York: John Wiley &:Sons, 193a) pp. 255-05. 2In a later chapter it is assumed that the marginal disutility of a dollar over the range of the three stakes is constant. 3Pretest results indicated that the gains questions were interpreted as gambling to which many respondents objected. The attempt at avoiding this connotation was not completely successrul. 4The example was unlike any of the situations in Table 2 in respect to amount or loss or gain, probability or payment to play. the sheet of gain situations by substituting gain for loss in the first three steps and meaningful terms in the fourth, e.gy,a.small investment (‘D in some larm enterprise. Summary This chapter has presented the mathematical structure for utility measurement. It has shown how this structure has been identified with certain aSpects of the technique used in the I.N.S. to quantify utility. The next two chapters will be concerned with the results of using this technique with 529 farm managers. ChAPTEd IV DEL jLA1b1AT‘“ TO US1A3LT‘jh COLSTSTfim LLAbSZFl”ATlU1S This C1apter elaborates the utility model and as a Tirst step in the analysis oi the data elas si1ies responses as to desfi ea oi consist- ency with the elaborated model. 1bis is only a preliminary to the main analysis of this thesis. This classification is used (1) to test the eigectivene ss 01 the schedule and t;e reliability of’inte rvi etin5: and (2) to not only test t1e technique of quantifying utilitv, but also in determining relationships ‘etween utility estimates and related vari- ables. Testin5 of the schedule and the reliability of the interviewing will take place in the 1ext chapter, while the second step is carried out in the sixth chapter. Elaborated hodel In what Tollows the model is elaborated to include a utility func— ‘tion consistent with the formal structure set Torth in the previous (ihapter. There, it was assumed that an individual would say 'les' to a situation 11' at u + (l — at) v > w where u = the utility of the position of acquirin5 the 5ain v = the utility of the position 01 11avinr lost the payment to play w = the utility oi the present income, and 0C — the probability of the gain. u2 L13 This condition is illustrated by the function shown in Figure 1.1 Wealth is plotted along the horizontal axis and utility along the vertical axis. In Figure I, the following symbols represent the quantities of wealth represented by the property to be gained in the hypothetical situations-- P0 - the present income PO - P1 = the value of the property to be gained, and PO - P2 - the payment to play. Utility A p-----—-- "U [.4 Wealth Figure I. Hypothesized utility function for an individual W110 accepts a fair bet. Choosing at such that the expected gain of wealth is equal to zero, i.e., a 1?1 + (1 -oc ) P2 :- PO,’2 makes the situation a fair bet. The expected utility E, is shown at A where 041.1 + (1 - a) v = E. The utility of the present income w is shown at B. The operation ¢n11-+ (1 - at) v > w is satisfied where {I > w. Only a function concave from 1The shape of the hypothesized function is due to Friedman and Savage, pp. cit., page 71;. 2The equal Sign means the monetary position of the individual is mmhanged from his position at P0. In Table 2 a fair bet was indicated whenarpl . p0 .. P2. above can describe the utility attached to the gain of wealth for a 53. person who accepts this kind of a fair bet in maxinizine LJ his expecte utility. In the loss situations where u = the disutility of the position of incurring the loss W = the disutility of the position of having lost the payment Of a premium (insured income) v = the utility or the present income position, and d = the probability 01‘ a loss, a convex function from above shows the operation c1 u + (l - a:) v < w to be satisfied. 1y du+ (l - at) v =uis shown In Figure II, the BL ected utili at C, and the utility of the insured income, w, at D. PO — Pl - the value of the property ta be lost, and "U o I *U m l the amount of the premium. Assuming a, is such that a; Pl 4- (l - a) P2 = P0,:L then the situation is a fair insurance scheme. For an individual to accept this situation Ineans that E < w. Similarly for all the other odd situations, that is, more than fairy and unfair, the utility function for gains will be considered concave from above and for losses convex from above. ~.— — lThe equal Sign reans the monetary position of the individual is unchanged from his position at PO, i.e., the cxpeeted loss is zero. In Table 2 a fair insurance scheme was indicatmiznnHI agPl = P0 - P2. P1 P2 P0 Wealth Figure II. Hypothesized utility function for an individual who accepts a fair insurance Scheme. The Patterns of Consistency In this section all possible responses to the hypothetical questions are w and. cn11' + (l - at) v' < w' given by a particular set of answers. From the indifference operation the utility of either gains or losses can be established. By using the concept of indifference points and assuming that these points fall on a numerical function, it is possible to determine types of answers which are consistent or inconsistent with the hypothesized utility function. First those answers which either do indicate or could indicate an indifference point will be defined as being consistent with the axiomatic indifference operation. These consistencies will be called "within— odds consistent". Second, the location of the indifference points will be specified in order to define consistency with the hypothesized function. This kind of consistency will be called "between—odds consistent." Other types of answers will also be considered. In additior some statistical comparisons will be made of the observed : v ; - 0 ° 4” ° and 8-4390 UGO. numoers 0... 0011818 be“ (3188. Within-Odds Consisten ’) g These cases are consistent with the axiom of indirierence. There are two kinds of within-odds consistency, (l) with the indifference point and (2) without the indiirerence point. With Indifference Point In this case the expression "within-Odds consiste t" refers to the series of answers in which an indirierence point can be established. Such points can be established if an individual first said 'No' to one or more of the questions, starting at the smallest gain or loss and then said 'Yes' to the remainder of the questions within one set of odds. The intervals in which indifference points can be established occur along the horizontal axis for the gains and for the losses between two adjacent possible gains or losses, to one of which the respondent said 'No' and to the other of which he said 'Ies'. There are five intervals :in which the indifference points can occur. The location of the indifference points for each set of odds defines "between-odds consistency." This kind of consistency not only 1 agrees with the indifference axiom but is also consistent with the hypothesized utility function. The derree of consistener ’5 determined by the number of indifference points indicated by the answers to the U? gain or loss set. The three degrees of consistency to be defined contain three, two and one point(s) respectively. Betweentgdds Consistency with Three Indifference Points. - A pattern of answers for a respondent which displayed between—odds consistency with -three indifference points (1) has one indifference point for each of the three odds which (2) can be joined by a line over the interval in which the indifference points occur that in the case of gains is concave from above and in the case of losses is convex from above. To be consistent with the hypothesized function, the indifference point for the fair odds has to occur at an amount equal to or greater than the amount at which the indifference point for the more than fair odds occurs. A simi- lar relation has to exist between the unfair and fair odds, i.e., the indifference point for the unfair odds has to occur at an amount equal to or greater than the amount at which the indifference point occurs for the fair odds. By considering the five possible indifference points for each odd, the number of possible consistent cases of this type is 35 for either tie gains or the losses. Actually, 2h and 29 cases for gains and losses respectively were manifest in the results. Eetween-Odds Consistency with Two Indifference Points. - This type of between-odds consistency requires the following conditions: (1) one indifference point per odd for each of two adjacent odds, (2) either (a) the point for the least fair odd occurs between two amounts (of gain or loss) larger than the interval in which the other point occurs or (b) both points occur in the same interval, and (3) the odds for which there is no indifference point are (a) answered all Yes in the case of ht the more-than-fair odds and (b) answered all No in the case of the unfair odds. The number of possible consistent cases of this type is 30 for either the gains or losses. When all five possible indifference points are considered, the results showed 19 and 22 cases for the gains and losses respectively. Between-Odds Consistency with One Indifference Point. - This type of consistency occurs when (1) there is only one indifference point and (2) the other odds are answered all Yes or all he according to the following patterns: a) If the indifference point occurs in the more-than-fair odis then all the fair and unfair situations have to be answered No. b) If the indifference point occurs in the fair odds, then all the more-than-fair situations have to be answered Yes and the unfair situations No. c) If the indifference point occurs in the unfair odds, then all the fair and the more-than-fair have to be answered Yes. When the five possible placements of the indifference points are counted for either the gains or losses, the number of consistent cases of this type is 15. For the gains lh cases and for the losses 10 such cases were actually found in the results. In total for either gains or losses there are 80 cases of the three types of between-odds consistency defined above. A Chi-square test run on the sum of the squared differences between the observed and the expected number of types of consistency showed a significant dif- ference (3 percent) for gains and a difference significant at 30 percent for losses. . -. . .— -,.- .._,. '- L—“a n, H---J J-.\I v. . n,‘« ~.‘. ... Hu..a‘-A‘43 -- . n- . ._ , ‘__ ""“w 3,. . ‘ fl .. a. n-.S d :., .K .. ,_ - ‘y‘ ' ‘ "“3. L :I . a, E A I.. , C‘. n-_. _._ ' o _ ‘v-~-. ‘ ‘f‘.v‘~ ‘OH“"- J'.:“ c .. ""., ll . -,‘ " . .- .. _ V‘.v a... r .‘- . . K “¢ ‘ ‘4 \‘ A . “v ‘._‘A ' “V4 - - ~u_‘ V .‘ “ . . ‘4_ “ Z A“ '1 . vuga‘xfla " ‘ h V L ‘4- .u "‘"-\ ”-‘ ...:-- ~ “'V s~- " v._.- z. w. ( I.“ d‘v .3 U N. “V‘: ‘h’fl “v. ‘\ IV V. -L p, ‘ 1W , v” a. , V. K) ‘ "‘Q - d £- \0 A smaller number of consistent answers were obtained for the gains than were obtained for the losses. This, no doubt, increased the prob- ability of finding a greater number of consistent answers of the different types for the losses and explains the greater degree of sig— nificance for the gains. There were almost twice as many usable schedules from the standpoint of showing indifference points for the losses than for the gains. The apparent difficulties involved will be 0 55. analyzed in the next chapter (see pages E£9to ,, , Without Indifference Points A wea?er definition of "within-odds consistent" is a series of answers in which there is no indifference point, as the reSpondent said either 'Yes' or 'No' to all the situations within one set of odds. This is a weaker definition of being consistent with the axiomatic indifference operation because there was no indifference point indicated by the answers. These answe's can not be considered to be inconsistent with the axiom since a smaller or larger gain (loss) than was included in the questions would allow the respondent to indicate the location of his indifference point. They are consistent only in the sense that five chances to reveal inconsistency failed to do so. Between-Odds Consistency with No Indifference Point.- This is the ‘-..-- weakest type of between-odds consistency. It occurs when the reSpondent answered either all 'Yes' or all 'No‘ according to the following patterns: 1) No to all three odds, 2) No to unfair and fair, Yes to more than fair odds, 3) No to unfair, Yes to fair and more than fair, h) Yes to all three odds. Any one of these is called a weak form of consistency because (1) there is no indifference point, (2) the pattern 2) above is not inconsistent with a hypothesis of diminishing marginal utility for gains or diminishing marginal disutility for losses, and (3) pattern 3) above is not inconsistent with a hypothesis of constant marginal utility. None of the cases are inconsistent with a hypothesis of measurable utility. There are only these four possible cases of this type. Within-Odds Inconsistengy The answers in this group are those which showed more than one indifference point per odd after adjusting the answers ( see pp. 55- SS ). Also included in this group are cases which were inadequate for assessment, i.e.,one or more, but not all, situations were not answered. Between-Odds Inconsistency These cases of inconsistency, although containing one indifference point per odd which made them "within—odds consistent," had the in- difference points located so that they were not consistent with the hypothesized utility function. None of the Odds Ansggred In some cases, the respondents refused to answer the questions; these cases are included in this group. Statistical Comparisons of Observed and Exp ected titlin-Odds Consistency The number and percentage of observed and expected within—odds consistencies are shown in Table 3. The expected number assumes that the questions were answered at random. That is, by chance seven cases are expected to be consistent wit1 the indiii 3rence axiom out of a total of on ways in which six Yes or No answers can be permuted for one odd. TASLE 3 ‘CT‘I I7J L3 ELL AID P.t(.PU’I‘lOi.S OL MITII' 'ILI- ODDS 'ISTfinCIESL W GAILS AND LOSSE Type of Answer Odds Observed Observed Expected Expected Number Proportion Numjer Proportion Gains Consistent Kore than Fair 363 68.6 57.8 10.9 Inconsistent Kore than Fair ldo 31.h h71.2 69.1 Consistent Fair 370 (9.9 57.8 10.9 Inconsistent Fair 159 30.1 h7l.2 89.1 Consistent Unfair 369 69.6 57.0 10.9 Inconsistent Unfair 160 30.2 h7l.2 84.1 Losses Consistent More than Fair :90 73.7 57.8 10.9 Inconsistent More than Fair 139 26.3 J"(1.2 59.1 Consistent Fair 356 67.5 57.6 10.9 Inconsistent Fair 173 32.7 n71.2 69.1 Consistent Unfair 37h 70.7 ‘57.8 10.9 Inconsistent Unfair 195 29.3 h71.2 89.1 t is apparent that there are six times as many observed within- -oo dds consistencies as there are expected by the random model. Ob. iously random answering as a null hypothesis must be rejected. 52 Statistical Comparisons of Observed and Erpected BetIeen-Cdds Consistencv In total for either gains or losses there are 8h cases of the four +x as f b t 'm - d‘~ a: in . :.r- . y w 1 on H- a type 0 etwe-n 0 Us conQIStdncy as UQIIHEQ above. ine numoer 01 individuals who revealed each of the four types is shown in Table h —,.. The total number of interviews was 529. TASLE h Lunezi or INDIVIDUALS IN EACH or TE? roux Tress or aerwazw-opgs CQLSISfew Number of Indifference Observed lumber Points Gains LII II Losses 3 >6 127 2 240 7 3 l h? a 36 0 .3: 112:5... a” ._ . . . .. . . . ‘ ror reasons g ven in a later Chapter tiis does nOt include all No answers. It is apparent that a hypothesis that would state an equal prob— ability of occurrence for each of the 8d cases would be rejected. A hypothesis of the probability of occurrence could be formulated from this study that could be tested by later studies. Subsequent chapters will consider some of the reasons for these differences in answers between the degrees of consistency and between the gains and losses questions. 1Considering the gains and losses together, there are 7,036 possible combinations of cases consistent with the axiom and the measurable util- ity hypothesis. No attempt has been made to formulate a probability model for the 8h or for the 7,056 cases. CE" "21.311 v 131-«130'111133'1ss OF T1113 SCEEDULL 5 Am II-JiEiVIMT-ns This chapter considers the reSponses to the schedule of questions concerning gains and losses in reSpect to its eI;ective ”13 s in elicit- ing answers. In ord to carry out tras evaluation it is iirst necessary to review the procedure followed in getting the answers i‘rom the field schedules to IBM cards. This review indicates the adjustments that were made in the answers and the kind of inrormation Iinally coded. Then the responses are cross classified between the types of answe and the state in which the schedule was taken. Finally, between states differences in types of answers are related to possible determining variables, including interviewing procedures. This chapter is divided into subsections as Iollows: (l) coding procedure and adjustment, (2) the information coded (3) answer groups (D de Iined, (h) between stat differences by answer groups with special reference to inteiw iewer bias. Coding Procedure The first step in the process of getting the responses onto IBM cards was to copy the answers Irom the Iield schedules onto the reSDective columns of work sheets. The work sheet is shown in Appendix C. The ext step was to make certain replacements and tranSpositions which prorided some additional information for testing a methodological hypothesis. A replacement is the changing of a Yes to a No or vice versa in order that there will be one and only one indifference point per odd. If therevwnwano indizierence point, then no replacement was necessary' however, if two indiirerence points were indicated by the pattern 0: answers then the point at the greatest gain or loss was eliminated by a replacement. A common replacement, for example, would be to change the last No to a Yes in the iollowing sequence. No Yes Yes Yes Yes No A transposition is the interchanging of a Yes and a No in order to make a pattern or answers show one and only one indifierence point per odd. A common transposition would be to interchange the under ined Yes and No in the sequence of unfair odds. Fair odds: No No No Yes Yes Yes Unfair odds: No No No Yes £9 Yes This will make the indierrence point for the unfair odds occur at a larger amount than for the fair odds. Either a replacement or a tranSposition was allowed in each set of odds but not both. Only one of either kind of adjustment was allowed because only one indifference point and hence only one area of inde- cision is implied by the model used in this thesis. Whenever a replace- ment or a transposition was made, a special code number was indicated. Whether or not this adjustment for consistency provides additional observations that would otherwise be lost is the methodological hypothesis to be tested by this procedure. The Special coie number furnishes the test data for this hypothesis. The hypothesis is based upon the fact that the probability of a consistent answer occurring by chance witnout adjustment would be extremely small and the probability of making an unintentional error of the type adjuStsd is lar e. Thus, a superior criterion for testing the hypothesis is wnether or not the adjusted cases are significantly different from the unadyusted cases in relevant respects. The hypothesis n is tested in ChapterEKEwnere lurther preliminary constructions are given. The third step in coding was to summarize the placement of tne indifference points on the left hand side of the work sheet. All the information necessary for further coding then appeared on the work sheet. Coded Information The following were coded and placed in 29 columns on IBM cards: 1) A summary of the placement of the indifference points for each of the 6 odds. 2) A special punch noted whether or not the odds were converted to consistency by repla ement or tranSposition. 3) The exact answers for each gain and loss situation. h) The numerical utility corresponding to each indifference point after adjustments were made. (These computations will be illustrated in the next cnapter.) \n c x S) The patterns or types of between-odds consistency. / ..- r “(w ._ 'I a - - r. J_ 1_ o - . . , . o) A summary oi conoinations ior both gains and losses of within 7 and between-Odds consistency. Answer Groups Defined The 529 schedules were divided into six groups based, in part, upon the consistency classes developed in the previous chapter. Further, some of these groups are distinctly different irom each other in the mode of answers. however, considerable variability still exists within several of the groups. The schedules of the first group (Group I and A) are similar in that all more—than-iair situations were answered Yes, but differ in the manner in which the fair and uniair odds were answered. The last group (Group VI and F) contains schedules which showed within and between-odds consistency; however, no dis- tinction is made concerning the location of indirference points until the next chapter. The six groups for gains and the six groups tor losses are defined below:1 Group 1.2 Gains questions answered showing one of the following patterns: a) Yes to all more-than—fair, No to all fair and unfair __ AA‘ —_— 1For the remainder of this chapter groups formed from answers to the gains situations will be indicated by doman numerals and the groups formed from answers to the loss situations will be indicated by capital letters. 2These cases are consistent with he indifference axiom and with the measurability of utility hypothesis; however,are the weakest type. b) Yes to all more-tnan-iair and fair, So to all unfair c) Yes to all more-than-Yair, Eair and unjair. Group II. Gains questions answered to to all odds. Group III. Gains questions answered out within-odds incon- sistent or inadequate for assessment. Group IV. Gains Questions answered but between-odds inconsistent. Group V. Gains questions not answered. Group V1.1 Gains questions answered and both within and between- odds consistent. Similar groups ior the loss situations were Lormed as follows: Group A.2 Loss questions answered but shows one of the follow- 111 [J .1 patterns: a) Yes to all more-than-fair, No to all fair and unfair b) Yes to all more-than-iair and fair, ho to all uniair c) Yes to all more-than-fair, fair and unfair. Group B. Loss questions answered but No to all odds. Group C. Loss questions answered but within-odds inconsistent or inadequate for assessment. - ~— 1These schedules are consistent with the axiom and with the hypothe- sized utility function. This group is considered again in Chapter VI. 2These cases are consistent with the indifference axiom and with the measurability of utility hypothesis; however,are the weakest type. m . _ .ria55rig JK) Group D. Loss questions answered but between-odds inconsistent. Group E. Loss questions not answered. Group F.1 Loss questions answered and both within and between- odds con istent. These groups will be studied for diffesences with reSpect to certain attributes and behaviors. The revea ed differences will, in turn, be used to eXplain differences between the groups. The attribute data are: 1) State in which the schedule was taken 2) Respondent's years of Iarming experience 5) ReSpondent‘s age A) Number of respondent's dependents 5) Type of farming engaged in by the reapondent 6) Number of years respondent attended school. The behavior items are: 1) Net worth of respondent 2) ReSpondent‘s average gross income for a three-rear period 3) Debt position or reSpondent (amount of debt in dollars) h) Proportion of total land managed that is rented by the respondent 5) Proportion of total gross income that respondent earned from the iarr 6) The reSpondent‘s ratio of total debts to total assets lThese schedules are consistent with the axiom and with the hypothesized utility function. This group is considered again in Chapter VI. . i I JI .If»»..l.’rp..‘llol El) 1% 59 7) The responient's concern for taking action wne he should not, or for not taking action when he should. Comparisons were made on the basis of the Student's "t" test on the mean values of— he groups and by the Chi—square test of independence for the attribute data. Only those results are reported which showed differences significant at the 30 per cent level by the "t" test between groups and indicated independence between variables at chi-square values significant at the hO per cent level. For each characteristic the level of significance will be reported. Between State Differences by Answer Groups with Special deference to Interviewer Bias For Gains The distribution of the 529 schedule over the six answer groups for gains is shown in Table 5 in numbers and by percentames. r: C) TABLE 5 DISTRIBUTION 0E Tdfi 529 SCHEDULES OVER THE SIX ANSWER GhOUPS, FOd GAINS —_* ' I ‘_ ——_" ____ _ Answer (Mode of Answer) Number of Per cent Group Schedules of Total I (Weak Consistency) 95 16.0 II (All No) 123 23.2 III (Within-odds Inconsistent) 83 15.7 IV (Between-odds Inconsistent) 21 u.O V (Not Answered) 65 12.3 VI (Consistent) 112 25.8 Total 529 100.0 00 . / -- 7. . .. .. a - . . ~. . . . In Table 0 the distribution 01 the scneoules ior the seven States over the six answer groups is Shown by per‘entages. TAJLE 6 DISTthUTIOh 0F T53 SGJhDULfiS roe SAGE 0E THE 53 3h TAEES OVER TILE SIX AJ‘QSWEI—i GROUPS, BY Piii 0313‘, FOR» GAILQ'SC‘ -‘ *7 1 11 III’”“ 1v v v1 Number (Weak) (All No) (w - 0 (B - 0 (hot (0on- Total or Incon.) Incon.) Ans— sist.) Sche- wered) dules Kentucky 26.2 13.1 lé.h 3.3 8.2 32.8 100.0 61 Ohio 1a.7 u.n 7.n n.n 27.9 A1.2 100.0 08 Indiana o.n5 3%.1 1n.o 2.1 o.n; 12.9 100.0 93 Michigan 23.2 1h.3 11.6 3.0 12.5 3h.8 100.0 112 N0.Dakota 15.u 20.0 20.0 o.o 12.3 27.7 100.0 6; Iowa 22.0 27.1 22.0 5.1 3.u 20.u 100.0 59 Kansas 19.7 1°.3 22.6 5.5 15.5 18.; 100.0 71 aChi:square is signiiicant at less than 1 per cent. The important point to observe in Table 5 is that Kentucky, Ohio, hicnigan, and North Dakota have higher percentages in group VI than any other group. Group I is second in per cent of schedules taken in Kentucky, Richigan, Iowa, and Kansas. Indiana and Iowa had the highest percentage of schedules in Group II; however, Indiana had over 50 per cent of its schedules in this group while Iowa had slightly over 25 per cent. Kansas had the highest percentage of schedules in Group III. All the states were low in respect to Group IV. The low percentages in Group IV probably result from coding. Schedules were probably disqualified from Groups I, II and VI on the basis of within-odds inconsistency before the between-odds factor was considered, i.e., if an individual was inconsistent within one odd, his between-odds con- sistency was not considered. In Table 7 the distribution of eacn group over tne seven states is shown by percentages. DISTRIBUTION 03 aacn GiOUP Gigi THE SEVEN STATES, er Pii'uivy 101 GAINSa 1 11 111 IV V VI (Weak) (111 No) (W'- o (B - 0 (not (ConSISt- Incon.) Incon.) Answered) int) Kentucky 16.9 6.3 12.1 9.5 7.7 lu.1 Ohio 10.5 2.6 ' 6.0 16.3 29.2 19.7 Indiana 6.3 63.9 15.6; 9.5 9.2 8.65 Michigan 27.6 13.0 15.6; 19.05 21.6 27.65 NO. Dakota 10.5 10.6 15.65 14.3 12.3 12.7 Iowa . 13.7 13.0 15.65 lh.3 3.1 8.ufi Kansas 1a.? 10.6 19.3 19.03 16.9 9.13 Total 100.0 100.0 100.0 100.0 100.0 100.0 a... . . .71 , . . uni-Square is Signiiicant at less than 1 per cent. The first table in this sequence (Table 6) showed how the schedules were distributed within each state. Table 7 shows the contribution of the particular state to each of the six groups. Thus, Michigan contributed the largest percentage of schedules to Group I, Indiana to O\ to Group II, and Kansas to Group III. Indiana, Michigan, North Dakota and Iowa were tied for a close second in Group III. hichigan and Kansas contributed the largest percentage to Group IV but Kichigan made the greatest contribution to Group VI. Ohio contributed the largest percentage to Group V. These two sets of comparisons for gains show the wide diversity in answering the gains questions both between and within states. If the questions had been answered similarly in each state, the distribu— tions by states would be approximately the same and the distribution between states would correSpond to the proportion of the total schedules taken in each state. This latter proposition is not true for the gains questions. The same two comparisons for the loss questions follow in the next sub-section. For Losses The distribution of the S29 schedules for the loss questions over the six groups is shown in Table 8 in numbers and by percentages. TABLE 8 DISTRIBUTION or THE 529 SCHEDULES ovaa THE SIX answei cacurs, FOd LOSSES A‘ _-——‘ —- ‘— V—v— ##W —— Answer (Mode of Answer) Number of Per cent Group Schedules of Total A (Weak consistency) 89 16.8 B (111 No) 16 3.h C (Within-odds Inconsistent) 105 19.8 D (Between-odds Inconsistent) 27 5.1 E (Not Answered) 51 ‘9.7 F (Consistent) 239 u5.2 Total 529 100.0 Table 9 gives the distribution of the schedules for losses within the seven states. TABLE 9 DISTRIBUTION or T33 SCHEDULES FOR EACH or ’Ha SE'EN STaTnS ovai THE SIX ANSwEn GROUPS, BI PER CEIT, FOR LOSSESd 1| A B C D E F Number (Weak) (All No) (w’- o (B — 0 (Not (con- Total of Incon.) Incon.) Ans- sist- Sche- wered) ent) dules Kentucky 31.1 h.9 16.3 6.9 8.2 3u.h 100.0 61 Ohio 8.8 O 10.3. 1.5 16.2 63.2 100.0 68 Indiana 20.6 5.3 25.8 6.5 6.5 5., 100.0 ,3 hichigan 16.3 1.8 15.2 5.3 10.7 52.6 100.0 112 HO. Dakota 9.5 6.3 9.5 6.3. 3 7.9 60.3 >100.0 63 Iowa 16.9 3.6 28.8 8.5 3.6 39.0 100.0 59 Kansas 18.3 . 2.8 33.8 2.8 16.1 28.2 100.0 71 a . . . . . Chi-square Significant at less than 1 per cent. The outstanding feature of Table 9 is that the highest percentage of schedules for all the states except Kansas is in Group F. Kansas has the second highest percentage in Group F which includes those schedules consistent within and between-odds and corresponds to Group VI for gains. This distribution is in contrast with Table 6 on gains where the proportion in Group VI for all states is less than in Group F. It is also noteworthy (seen in Table 9) that Indiana, hichigan, North.Dakota and Iowa have the second highest per cent in Group C 95 whereas Kansas has its highest per cent in Group C. This is the group in which either the within-odds were inconsistent or inadequate for assessment. In all stat:s except Kentucky at least 60 per cent of the schedules fell either in Group C or Group F. Group F dominates in all state S except Kansas a1id Kentucky. Thus, it appears tha' a high percentage of all the reSpondents was either totally consistent or in- consistent in answering the loss questions. This is in contrast to the results on the gains questions, where several other modes of answers, e.g., I, II and IV were also prominent. This ray mean that the loss questions were easier lor the interviewers to communicate to the re— Spondent than were the gains questions. At least this could have T sulted in the respondent attempting to answer the questions, regardless of whether or not he answered them consistent with the hypothesis; where as, in t1~1e case of the questions, othw ypes of answers may have appeared as the respondents attempted to avoid the "gambling type" Situations. In Table 10 the between states distribution by groups is Shown. In contrast to Table 7 on gains, where several states contributed a large percentage to a particular group, Table 10 Q 01 s Indiana ccn- sistently con’ributing the hirhest or at least a high percentage to roups A, B, an nd C; whereas, hichigan contributes the highest or a high percenta are to Groups D, 3, and F. This between-states distribution for losses corresponc ds more closely to t e plO“C crtion of t1e total schedulw ta} en in each state than the same distribution for the gains questions. . ‘\ .- ‘\’/ TABLE 10 DISTRIdUTICN or EACH snow? OVER ThEaSEVEN STATE , BY PEA CENT, roe LOSSES A B C D E F State (Weak) (All No) (W'- O (B - 0 (Not (Consist- Incon.) Incon.) Answered) ent) Kentucky 21.3 16.7 9.5 11.1 9.8 8.8 hio 6.7 O 6.7 3.7 21.5 18.1 Indiana 21.3 27.8 22.8 22.2 11.8 13.9 Michigan 18.0 11.1 16.2 22.2 23.5 2h.8 No. Dakota 6.9 22.2 5.7 1h.8 9.8 16.0 Iowa 11.2 11.1 16.2 18.5 3.9 9.? Kansas 1h.6 11.1 22.8 7.h 19.6 8.7 Total 100.0 100.0 100.0 100.0 100.0 100.0 a . . .1. . Chi-square Signiiicant at less than 1 per cent. ~The above two sections on gains and losses are presented tOgether to emphasize two important facts: (1) there are important differences in the way in which the questions were answered within each set (gains or losses) of questions and (2) there are important differences between states in the way in which the two sets of questions were answered. The first of these, that is, the differences in answer groups for the gains and losses questions, will be the main concern of the remainder of this chapter. In the case of gains certain answer groups were generally associated with a particular state and in the case of-losses two states contributed large percentages to the six groups. It is 66 highly probable that the attributes of the individuals in a particular answer group will be Similar to the average respondent in the state making up the majority of the group. However, if meaningful character- istics can not be found to explain the differences between the groups then the one respect in which states could differ, interviewer bias, will be used to explain the differences. In order to determine whether or not the characteristics of an answer group differ from the average of a state it is necessary to compare the characteristics of individuals in the seven states. The next section carries out this comparison for the characteristics listed on pages 58 and S9. Characteristics by States This sub-section presents the relevant characteristics of the re- Spondents in each state. No attempt is made to explain the revealed differences. The objective, instead, is to provide background and handy reference tables for the analysis that follows. The average net worth, gross income, and debt position of respondents in the seven states is given in Table 11. No ssignificant differences were found among Kentucky, Indiana, and Iowa in net worth; Kansas has the highest net worth of any of the states and Michigan, the lowest. Likewise, there are no significant differences among Kentucky, Ohio, Michigan, North Dakota, and Kansas in gross income. Only Iowa and Indiana are significantly different from the other five states and from each other. Comparisons on debt position 67 TABLE 11 AVEIAGT NET wears, GROSS INCOKE, AND DEBT POSITION or iESPONDBNTS,BY STATES State Dollars Net worth Gross Income Debt Kentucky hh,101.6h 7,1h2 2,710 Ohio 38,591.67 7,563 2,695 Indiana h8,670.59 9,h30 3,159 Michigan 3o,725.h9 7,179 2,770 NO. Dakota 36,020.00 7,619 3,117 Iowa h7,539.66 11,200 3,572 Kansas 64,632.8h 7,20h 2,687 0 fl show Iowa and Ohio to be different at a 30 per cent level of significance. All other comparisons on debts are not significant. The age, farming experience, and number of dependents for reSpondents in each state is shown in Table 12. Kansas reSpondents who had the largest number of years of farming experience showed a significant difference from other respondents except those from Kentucky who ranked second in number of years of farming experience. There is no apparent difference between the other six states in this reSpect. There is about a four-year difference in average age between Kentucky, Indiana and Kansas respondents compared to Ohio, Michigan, North Dakota and Iowa as groups. The average number of dependents per respon dent did not vary significantly between the states. T “ LE 12 AVENAGE YEARS OF RESPONDENTS' FAAIING EXPEAILICE, AGE OF HLQPCID‘T S NUMBEh OF IESPONDEKES' DEPSIDJII b, BY SIATJ‘S 3 —__.— -—.—‘ ~ c—* “W State Years of Farming Age Number of Experience Dependents Kentucky 21.9 50.3 2.8 Ohio 19.1 h6.7 2.6 Indiana 21.0 50.8 2.5 Michigan 19.9 h8.l 2.9 North Dakota 19.8 h6.l 3.1 Iowa 17.9 DA.O 3.0 Kansas 23.1 h9.2 2.7 The following six reference tables show highly significant differ— ences between states for (l) the ratio of debts to assets (Table 13), (2) the proportion of total acres rented (Table 1h), (3) the proportion of income from farming (Table 15), (h) the type of farm (Table 16), (3) the concern for the two types of errors (Table 17), and (6) the number of years of school attendence (Table 18). Each table shows the per cent of the total number of reSpondents in that state with the particular attribute. .9, . .3... a. in; TABLE 13 :33) RAT: OF DEBTS TO ASSETS,PHOPORTIONS, BI STATES State Ratio 0 < .1 .11 - .2 > .2 Total Kentucky 62.3 9.8 11.; 16.; 100.0 Ohio 55.0 8.3 21.7 15.0 100.0 Indiana h6.> 19.7 25.h 8.5 100.0 Kichigan 57.3 17.1 1h.o 11.0 100.0 North Dakota 30.3 2h.2 25.8 19.7 100.0 Iowa 32.8 31.0 19.0 17.2 100.0 Kansas 60.3 19.1 11.8 8.8 100.0 a,.. . .n. Lni-square Signilicant at 2 per cent. TABLE 1h A PROPORTION OF TOTAL ones RENTED, BI STATES ~—-.-. — . .‘fi-. - ‘m State Proportion Rented 0 < .5 .5 - .7 > .7 Total Kentucky 50.8 2h.8 8.2 16.; ' 100.0 Ohio ;O.C 22.0 11.8 16.2 100.0 Indiana 62.4 11.5 :“.? 1h 0 100.0 Michigan 62.5 25.9 8.0 3.6 100.0 North Dako+a 29.2 21.5 15.8 33.8 100.0 Iowa 35.6 10.2 6.8 37.8 100.0 Kansas 28.2 18.3 16.9 36.6 100.0 “flu-.- - a . . .m. . . . Chi square Signiiicant at less than 1 re“ cent. TABLE 15 ~- v -. -—1 t7 .v..-) --r- r! T“ t""‘7 '1‘: r1 ‘li‘ 8' PROPORTION 0r InCUmi BdOm rAimth, DI SLATES ——~ ----..—-—-s 4- .. d“. Proportion from - “-fl—‘W- c.- I; ‘1‘ “71"] Li? igzizflu ‘. Q -0”.-- 9 ‘ State "0 - < 19‘ .5 - .75‘ :37T75 < 1 ‘”“7EII‘ Total Kentucky 3.3 8.3 13.3 75.0 100.0 Ohio 8.8 2.9 26.5 61.8 100.0 Indiana 7.5 11.8 10.8 69.9 100.0 Ifichigan u.s 6.3 11.6 77.7 100.0 North Dakota 8.6 1.5 10.8 83.1 100.0 Iowa 0 1.7 6.8 91.5 100.0 Kansas 5.6 7.0 15.5 71.8 100.0 _ Aw a .. . . . Chi square Significant at 2 per cent. TABLE 16 "1 “1 ‘1 7; -\ ~ (mm 7. x 1'1 r *1 3' TYPE Or TAiL, PiUPCnLIONS, bY SLATES w”..- State Dairyb Fat dasn ucn- Iat Stock: Toooc- Ctnerll Total tockv Cropu er‘ele Cash Cropl cog Kentucky 3.3 39.3 14.7 3.3 9.8 29.5 0 160 Ohio 14.7 h1.2 22 1 8.8 7.3 h. h 1.3 100.0 Indiana 3.3 53.3 21.1 7.8 0.9 0 3.6 100.0 hichigan h2.3 9.6 21.1 11.5 2.9 0 12.5 160.0 No. Dakota 3.1 lh.1 73.4 1.6 7.8 0 C 100.0 Iowa 0 63.8 23.9 1.7 6.9 O 1.7 100.0 Kansas 1.5 13.h 71.6 1.5 11.9 0 0 100.0 Total Number 7" 7‘ 9‘ of Schedules 62 163 175 30 29 21 20 100.0 :Chi square significant at less than 1 per cent. CIore than ton of income fronl dairying. d:ore than toe of incorqe from fat stock, i.e., h0g5, eef, and Sheep. more than ten of income “rem cash crOps. HG) between 15 and hW of income from each of fat stock, dairy and other (or between 15 and ho; of income from each of dairy and two other. Either 50 aSI crop and 503 Iat stock; or be ween 30 and htp cash crop and betw ee 30 and to“ fat stock. 5; ore than 35; tooacco, usually in combination with crops or fat stock. Includes those with more than ho; of income vegetables, poultry or truck fc‘ming. .1.‘. 01181“ cash from fruits and {‘0 TABLE 17 fl If!‘1‘\\ '\ - 1 . a 1 "1 fi *‘fi ' h r i fir‘a COLUnnd POI THn TWO PIPES Ob him i, D1 STATDD Proportion Concerned State b C a lst 2nd Both Don't Know Total Kentucky 27.1 32.2 33.9 6.8 . 100.0 Ohio 20.6 16.2 52.9 10.3 100.0 Indiana 22.6 30.1 hh.l 3.2 100.0 Kichigan 31.5 32.8 35.2 .9 100.0 No. Dakota 29.2 hh.6 20.0 6.2 100.0 Iowa 18.6 no.7 35.6 5.1 100.0 Kansas 18.6 2h.3 "7.1 10.0 100.0 aChi square significant at 2 per cent. b” . . . _ ,. . . . Cmore concerned about their taxing action when Should nOt. \ '. .Aore concerned about not taking action when should. Are equally concerned about both types of error. 0 N] b.) TABLE 18 NUHBER OF YEARS or SCHOOL ATTENDANCE, BY STATfiSa State .1 < 8 8 > 8 and 12 > 12 < 12 Kentucky 2h.6 37.7 1h.7 1h.5 8.2 Ohio 8.5 25.0 11.7 39.7 11.8 Indiana b.3 26.9 21.5 33.3 1h.0 hichigan 10.7 h3.8 20.5 17.0 8.0 North Dakota 16.9 32.3 21.5 23.1 6.2 Iowa 6.8 27.1 23.? 35.6 6.8 Kansas , 12.9 uu.3 10.0 21.1 11.h Chara.te:;stics of Farmers in Each Answer broup on Gain Questions }. This section considers the characteristics of the respondents in the six answer groups for gains. It will be recalled that the purpose of this section is to explain the differences between the modes of answers on the basis of attribute and behavior data. The following table (Table 19) indicates the average net worth, gross income, and debt position of these respondents. Group II (all No answers) has the highest average net worth and is significantly different from all the other groups. It also has the highest income and the second highest debt position. Indiana con- tributed the largest percentage to this group and has the second highest net worth, gross income, and debt positions. TABLE 19 AVERAGE NET WORTH, GROSS INCOHE, AND DEBT POSITION, BY AhSWER GROUPS, ON GAINS _. .___ __ —‘—.— Answer Group Dollars Net WOrth Income Deb—— I (Weak) 15,118 7,618 1,593 11 (All No) 53,031 8,863 3,2h6 III (W—O Inconsistent) 38,597 8,h§6 2,12h IV (B-O Inconsistent) 28,760 6,611 1,853 V (Not Answered) h6,062 7,h72 1,1hh VI (Consistent) 38,922 7,7n6 2,972 Group IV (between-odds inconsistent) has the lowest net worth and income and the second lowest debt position. However, in contrast to Group II, all states contributed about equally to this group. Although Group I (weak consistency) and Group V (not answered) do not differ significantly on net worth, there is a significant difference in debt position. This may indicate that individuals in Group V are financially more stable than those in Group I. Notice that there is no significant difference between Group III (within-odds inconsistent) and Group VI (completely consistent) on any of the three variables in Table 19. All states contributed about equally to Group III, but Michigan contributed more than one-fourth of the schedules in Group VI. However, average net worth for Michigan is substantially less than the average net worth of Group VI. Table 20 gives the average number of years of farming eXperience, age, and number of dependents for respondents by answer groups. .. 1 \f‘ TABLE 20 A‘FliAGE 111.1113 or 1‘ 11.1110 1311311111011, 113, 11111 11111-111111 03‘ D:L'P:..’..).. .13, 1‘11: .11:s:;1.a: ours, 011 0111-13 ‘— c. Answer Group Years of Number of Farming Dependents Experience :» U"; (D I (weak) 20.0 h7.4 2.9 II (All NO) 23.0 50.7 2.8 III (W40 Inconsistent) 19.2 111.1 2.9 IV B-O Inconsistent) 20.h h9.9 3.2 V (Not Answered) 25.7 55.7 2.0 VI (Consistent) 16.8 hh.d 2.9 Group II respondents again are very much like the average Indiana respondent as shown in Table 12. Group IV respondents have the lowest net worth and income, a very low debt position, relatively short farming "periences and t1w largest number of dependents. Previous comparisons between Groups I and V indicated that indi- viduals in V may be financially more stable; this fact is compatible with the data indicating that Group V individuals we relatively older, have more farmin3 experience and fewer dependents than those in Group I. This contrast is also true when Group V is c01npared to all the otlie on the three variables of age, experience and number of dependents. Again notice that there is no significant difference between Group III and VI. Michigan and Ohio schedules form approximately 50 per cent of Group VI; however, the average length of farming experience of indi- viduals of these two states is considerably greater tlan the average for the whole group. The average age and number of dependents for this group are about the same as for Kichigan and Ohio. The following series of tables (Table 21 to 2h) present attribute data about the six answer groups which were found to be significant by the chi-square test of independence. Two attributes-—years of school attendance and the ratio of debts to assets--were found to be independent of the answer groups. TABLE 21 ' ‘ “r" I r ' 31 ‘r '71-. ('1'1' '3' \' ' T * ‘v *7' 8- PROPOfiTION 0: TOTAL ACnL dnjfhfl, bf AdSflhd GiOUPS, On GAth Answer Group ___ __frooortion;fiented O < .3 .3 - .7’ > .7 ’Iotal I (Weak) 111.2 23.3 8.1. 22.1 100.0 II (All No) 52.6 17.1 13.0 17.1 100.0 111 (w-o Inconsistent)h3.h 111.1. 16.9 23 .3 100.0 IV (B-O Inconsistent)h2.9 19.0 9.5 26.5 100.0 V (Not Answered) 66.2 10.73 12.3 10.75 100.0 VI (Consistent) h0.8 ”1.7 7.7 26.6 100.0 afi,. . ._, 1 uni-square Significant at 5 per cent. 77 TABLE 22 PROPORTION OF INCOHE FROM FARMING, BY ANSWER GdOUPS, ON GAIIESa ‘— —j— ____‘ __A __‘ Answer Group Proportion from Farming, <‘.3 .5;.75 .3.73<1 All Total I (weak) 3.2 o.h 11.6 76.8 100.0 II (All No) h.l 8.1 8.9 78.9 100.0 III (W40 Inconsistent) 7.2 1.2 8.5 83.1 100.0 IV (B-O Inconsistent) O 9.5 19.1 71.h 100.0 V (Not Answered) 9.2 1.5 18.5 70.8 100.0 VI (Consistent) 5.0 7.1 18.6 69.3 100.0 a .. . . “..” Chi square is Signiiicant at 10 per cent. TKEE23 CONCERN ABOUT TIE TWO TIPSS OF 111.1011 101 SD: AIISL‘Ei GROUPS, ON GAIL-Isa Answer Group ‘ Proportion Concerned lstU 2ndC 150tha Don't Know Total 1 (Weak) 23.9 3h.8 33.9 5.h 100.0 II (111 he) 21.0 31.1 h1.9 5.7 100.0 III (W-O Inconsistent) 32.9 30.5 31.7 h.9 100.0 IV (B-O Inconsistent) 9.5 66.7 23.8 0 100.0 V (Not Answered) 20.6 1h.3 52.u 12.7 100.0 VI (Consistent) 27.7 31.2 37.6 3.5 100.0 Chi square is significant at less than 1 per cent. More concerned about their taking action when should not. .dore concerned about not taking action when should. Are equally concerned about both types of error. e p o*m TASLE 2h 78 w --1 \" -—‘ r“: a ,3 -) -..' ”N .- '1‘!” rs ~ ‘ A? ‘r ‘ TIPo 01 1 1h, Piercanoh or ALSWbfi GdoUPb, Ow 0111.3D Answer Group Dairy Fat Cash Gen- Fat Stock—Tobac- Other Total Stock C 0p eral Cash Crop co I (Weak) 11.6 29.0 26.0. 7.5 10.3 7.3 5.h 100.0 II (111 No) 6./ A3.7 31.9 5.9 6.7 .9 1.2 100.0 111 (W-O Inconsistent) 9.0 33.3 hl.O 1.3 6.4 6.1 2.6 100.0 IV (B-O Inconsistent) 15.0 h5.0 25.0 0 5.0 5.0 5.0 100.0 V (Not Answered) 3.1 30.8 hl.5 h.6 6.2 0 1.5 160.0 VI (Consistent) 16.9 22.8 3n.6 8.8 8.1 5.1 3.7 100.0 __— a w , r\ n no .1. .See Table 16 Ior type of iarm delinitions. 01.. . .n. .. . Chi square Sicniiicant at 5 per cent. Ll The characteristics of Group II correspond of Indiana in the following Characteristic (l) Preportion rented (2) Income from farming (3) Concern about errors (h) Type of farm way 8 : Indiana Second highest in owners and second lowest in renters High proportion from farting High proportion con- cerned about both High proportion in fat stock, and cash creps rather closely to those Group II Second highest in owners and second lowest in renters High proportion from farming high proportion con- cerned about both high proportion in fat stock, and cash rops Tables 21 and 23 show two definite characteristics of Group IV: (1) the second highest proportion of renters, (i.e. equal to or greater than .5) and (2) two-thirds (the highest proportion) of individuals who are concerned about not taking action when they should do so. In contrasting Group I and V, the high proportion of owners in Group V as well as their concern for both types of errors (over 50 per cent of Group V) is consistent with the previous comparisons. It is difficult to find marked differences between Groups III and VI in the series of tables presented. It is,however, significant that though almost 50 per cent of the schedules of Group VI came from Michigan and Ohio, Group VI averages differ noticeably from these wo states in (a) proportion of total acres rented (Table lb and 21), (b) concern for two types of error, particularly in Ohio, and (c) type of farm (Table 16 and 2h). In respect to type of farm, the data Show a high preportion of dairy farms in Michigan in contrast to a low proportion in Group VI. Koreover, hichigan and Ohio have a relatively low proportion of cash crop farms whereas Group VI has a relatively high proportion. This section can be summarized as follows: Group I (weak consistency) -- This group has a high debt position suggestive of cautious individuals in risky situations. This group will be compared further with Group VI (totally consistent) in the next chapter. Group II (all No answers) —- This group is comprised of a high proportion of Indiana schedules and appears to be very similar in character to the average of all the Indiana respondents. Group III (within-odds inconsistent or inadequate for assess- ment) -- This group was found not to be significantly different from Group VI. 80 Group IV (between-odds inconsistent) -- This group has the lowest net worth, a low debt position, the largest number of dependents, and a high proportion of renters. Respondents are more concerned about not taking action when they should than any other group. However, the small number of schedules in this group makes it difficult to draw conclusions. Group V (not answered) -— This group has a relatively high net worth, 'he lowest debt position, and has older individuals with a longer period of farming xperience and with fewer de- pendents than any other group. It has the hi hest proportion of owners and its individuals aare equally concerned about the two errors. These characteristics are indicative of individuals wn would have no need to take risks for relatively larre gains. 0 This may explain their refusal to answer these questions. Group VI (totally consistent) -- This group is relatively young and thus has the shortest period of farming experience of any of the groups. This group, which will be analyzed further in the next chapter, does not differ significantly from Group III in any of the characteristics discussed above. In spite of the fact that Michigan and Ohio contributed almost 50 per cent of the schedules, this group seems to be comprised of individuals sub- stantially different from the average of the individuals inter- viewed on these questions in those states. Characteristics of 1a“~‘ s in Bach Answer Group on Loss Questions —— M— - This section will consider tlze characteristics of the reSpondents in the six answer groups on the loss questions. Table 25 shows the average net worth, gross income, and debt position of the respondents in the six answer groups for losses defined on pages 57 and 58. TABLE 25 AVELAGE NET WCATH, GROSS IKCOXE, AHD DEBT POSITION OF SIX AHSWER GROUPS, FOK LOSSES Answer Group Dollars Net Worth " Income Debt A (weak) 35,303 7,931 2,129 3 (All 150) 35, 306 , 0,069 3,567 C (W-O Inconsistent) h5,3h3 7,960 3,381 D (B-O Inconsistent) 37, 85h 9,h79 3,329 B (hot Answered) h7, 270 7,526 1,251 F (Consistent) 39, 2;d 7,92h 3,183 Group A (weak consistency) has a significantly h gher net worth than all the other groups with a large percentage of schcd.ules from Indiana and Kentucky. However, Group A respondents have an average net worth considerably higher than for those two states. Group B (all No answers) has the lowest net worth but is not sig- nificantly dif‘ierent from Group D and F. Only 18 schedules of the total of 52? fall into this group. Group C (within-odds inconsistency) and E (not answered) are simi- lar on both net worth and income and differ significantly only on debt a position. On debt position, Group ranks the highest but is not significantly different from Groups B, D, and F. Q? N) Group D (between-odds inconsistency) has the highest income whereas Group E has the lowest. Group E, however, does not differ sig— nificantly from Groups A, C, and F. Group E has the lowest debt position. Table 26 giv v88 Urea average number of years of farming eizpelience age, and number of dependents for respondents in the six answer groups. Group E contains the oldest individuals with the most farming experience and the fewest dependents. This also is true for Group V on gains (see Table 20). TABLE 26 1‘“ fioTr‘ AVLL‘ *AGEI YEA’S bOF E‘Ail-iIE-TG ELLT3 SHIJINL, ZA'ID TREES: OF T tuanliis OE' SIX "0 .SWZLL GROUPS U01; LCSSLQS ‘ ‘ __‘_‘ ——~ ’~---‘ 9“ m”..-. g ._ ____ . ___ ~’~1 Answer Group Years of Number of Farming Age Dependents Experience A (WBak) 2L.O 51.6 2.7 B (All NO) 20.5 h6.7 3.0 C (W-O Inconsistent) 18.0 h6.6 2.7 D (B—O Inconsistent) lh.7 h2.7 3.h 3 (Not Answered) 26.6 56.3 2.1 F (Consistent) 19.1 h5.5 2.9 The between-odds inconsistent individuals in Group D are the young- est of all the groups and have the most dependents. L1 lea st farming exper ience and the Tables 27 through Table 30 present data on attributes which proved to be significantly related to the answer groups by the chi-square test Proportion of income from farmin (3 s and the number of years of school attendancelflflfiifound to be independent of the six answer groups. TABLE 27 PAOPOATION or 133 TOTAL 30133 333323 FOR flxAfiwaiG&P%ONL&V“d ‘ -_ - -_-—,.._ “~- _Pr0p0Iti on he nt ed ”- Answer Group 0 < .3 .7 > .7 Total A (Weak) 39.5 19.1 6.7 13.6 100.0 B (A11 KO) 38.9 27.8 22.2 11.1 100.0 C (WFO Inconsistent) 36.2 21.9 10.5 31.h 100.0 D (B-O Inconsistent) 29.6 33.3 13.8 22.2 100.0 E (hot Answered) 60.5 15.7 11.8 11.8 100.0 F (Consistent) h8.3 17.1 11.7 22.6 100.0 I l l aq.. . ... . . Chi-square Significant at almost 2 per cent. TABLE 23 0030333 run 130 TIPES 03 33303 BI A.Sd"l 3.0333, 03 Loss 33a Answer Group __‘ Proportion Concerned lst5 2ndC hothd: Don' t linow Total A (Weak) 32.1 21.3 31.7 3.8 100.0 B (A11 NO) 26.3 15.8 36.8 21.1 100.0 C (W¥O Inconsistent) 22.3 36.9 36.9 3.9 lO0.0 D (B-O Inconsistent) 30.8 36.5 26.9 3.8 100.0 E (Not Answered) 16.7 18.7 50.0 1h.6 100.0 F (Consistent) 23.5 36.1 37.8 2.6 100.0 Chi-square Significant at 1 per cent. iore concerned aoout ti eir talIing action wh en should not. Ilore concerned about not t aking action ween should. Are equally concer ed aoout both types of error. (DOW 9: 8h TYPE OF FAifi,a PROPORTION BY AISHER GROUPS, ON LOSSESb m ‘ --.,. Dairy'Fat Cash Cen- FatStock-Tbbac- Other Total Answer Group Stock Crop eral Cash Crop co A (weak) 10.6 05., 16.5 3.3 10.6 7.1 5.9 100.0 B (All 36) 11.1 27.8 33.3 16.7 0 0 0 100.0 C (W-O Inconsistent) 8.0 38.0 37.0 2.0 10.0 3.0 2.0 100.0 D (B-O Inconsistent) 13.8 23.9 30.7 7.8 3.7 3.7 3.7 100.0 E (Not Answered) 10.0 26.0 06.0 6.0 6.0 C 6.0 100.0 F (Consistent) 14.7 27.2 35.3 7.3 6.9 3.8 3.5 100.0 a . , ,. . A. . . See Table 10 ior type 01 farm deiinitions. Chi—square significant at 5 per cent. TABLE 30 RATIO OF DEBTS TO ASSETS, PEOPOELTIOI‘I BY AIIS'LIEB (EOUPS, ON LOSSESa *- A ----m —_.—— -“ -—.-.-. .- ‘—_ v..— —— -—~ — —-. *.’-7 Qua—un— Answer Group _#_ Ratio 0 .01-.1 .11-.2 > .2 Total A (Weak) 51.3 21.8 16.7 10.3 100.0 B (111 36) 30.0 26.6 6.7 26.7 100.0 0 (wee inconsistent) 53.1 19.8 13.5 13.5 100.0 D (B-O Inconsistent) 30.0 3.0 33.0 12.0 100.0 3 (Not Answered) 75.0 5.6 11.1 8.3 100.0 F (Consistent) 35.1 20.0 20.5 10.3 100.0 a .. . . . Gui—square Significant at 2 per cent. F\ 0 Group E has the highest proportion of owners, the argest per cent of individuals who have no debts and who are equally concerned about both types of errors. Groups C, D, and F have high proportions of renters, but Group C has the highest proportion of individuals who have a debt—asset ratio of .l or less. Group D has the largest proportion who have a debt—asset ratio of .2 or more. Each group has a pattern of answers with respect to types of errors which seems to be meaningfully related to the answer groups for losses. Group A, where the reSpondent said either Yes or No to an entire set of odds, has the highest proportion of individuals equally concerned about the two types of error. The fact that a hiwher proportion were more concerned about the first than about the second error is also consistent with the fact that the group largely consists of individuals who accepted all unfair insurance schemes. The consistency between their attitude and their answers is evident in the following interpretation. In a situ— ation involving a high probability of loss, taking action when they should not is a successful evasion of the loss. however, in the hypo- thetical questions action had to be taken in a loss situation and the reSpondents accepted all the unfair insurance schemes to afford himself protection. Group B has the highest proportion of "Don't know" answers to the types of errors question, of any of the groups. Twice as many of the respondents who answered No to the loss questions were more concerned about taking action when they should not than about the second type of error. Group C has as many individuals who are equally concerned about both errors as it has individuals who are more concerned about not taking action wher they shovld. Group F has alhost identical percentages in 1 eaCh of ‘he categories, yet Group C was incorsistert and Group F was (J- consistent in answering the loss q estions. however, individuals in ‘ V Group F are crasi'errblv different from individuals in Groups A and B ' 1 in reSpect to their concern for the second tyne of error. Group E is not much I“'r-Ci‘e ent from Group C in the proportion of individuals concerned about the second type of error; however, it is different in the proportion concerned about taking action when they should not and in the proportion who are equally concerned about both errors. The type of farming in each respective group correlates markedly with the type of farming in the states making up the majority of the group. However, in the case of Group F, the distribution of type of farms rather closely conforms with the distribution for the total number of farms for all the states as shown in Table 16. In summarizing the results of this section, the following character- istics of each group will aid in understanding the differences between the six answer groups on losses. Group A (weak consistency) -- This growp, which is made up of a high proportion of individuals who took all unfair insurance schemes, has a high net worth and a relatively low debt position. CD *‘J Individuals in this group are slightly older and have more farm- ing experience than individuals in Groups C, D and F. There are as many individuals who have no debts as there are who have some debts. In addition, a high proportion of individuals in this group are equally concerned with both types of errors. These characteristics, plus the fact that a high preportion of indi-* viduals in this group took all the unfair insurance schemes, could mean hat they were ext_emely desirous of maintaining their relatively strong financial position. Group B (all No answers) -- No significant conclusions can be drawn about this group because of the small number of re- Spondents falling into this category. (It is noteworthy that tLe corresponding group for gains had l2} schedules with all He answers). Group C (within-odds inconsistency) -- This group is very similar to Group ‘ except for (a) highe“ net worth, (b) higher proportion of renters, and (0) higher proportion with no debts. However, these characteristics do not appear to explain the difference in the modes of answers. Group D (between-odds inconsistency) -- This group consists of individuals who are relatively younger and have fewer years of farming experienCe than any other group. Theie is also a high proportion of individuals in the owner-renter stage of farming. In general, they are concerned about not taking action when they should. These characteristics might fit an individual who, due co to his youth and intvjerience, would be prone to make mistakes in his attempt to maintain <3onsistencv. Group? a (not answerer) -— This group is characterized by individuals who have relatively high net worth and 101 debt positions. T11.3y ar ethe oldest inoiv‘drc‘" rith the longeSt period of farming experience and have a M ;h proportion oi owners among them and a ' i1 prooo"*ior with no de Wit They are equally IOtice the strilzin: fi-j) concerned about the two types of errors. similarity of this group with Group V on gains. A similar con— clus io n can be drrwn in tlis case; perhcns thCS" inF‘VIcuals lave U] U) F. C?- k: F1; 0 '1 Ti: 5. fl. 0 Po *1 D.) d. k.)- :5 H. :3 Ho :3 m c: *‘3‘ E? O (I) U) E.) gs 5 U2 6‘ L: :5 p 0 less no 0 tners. GroupF (totally con mi wnt)- - The outstanding feature of this group is the close r1sen1lanle 01 its characteris 'ics to (D those of th 52? respondents as a groul. The only noticeable exception is that individuals in Group F 1ere rela oively more concer ed about not taking action when they should, than indi- p Y viduals in ‘the total group. This group will be examined lurther in the next chapter. W O The Problem oi Inte"vie w er bias There e three facts from the analysi -s in the above two sections which remain without adequate explanation. They are: l) The high proportion of Indiana schedules in Group II on gains. 89 2) The similarity of char acte istics between Groups III and VI on gains. 3) The similarity of characteristics oet; :een Groups C and F on losses. There are, no doubt, several reasons that would exklain these facts; however, one of the most obvious reasons is the differences between intervievers. Whet m1 it be because of approach, personality traits or rapport with the farmer, differences between the interviews ‘ may account in some degree for the data that cannot be explained in any other wai. In an attempt to explain the high proportion of Indiana schedules in Group II on gains, comparisons are made etween intw ever from Indiana and Michigan in Table 31. TABLE 31 COE'IPASISOI‘J OF IIDIATIA ’11) I-IICIIIGAI‘I TII'L‘EL'ZV 131.4318 031 P31 ('1 T CF SCHEDULLS rAlJLII'G III‘TO E CH Al'Si'TJJ‘ { CL-CGUTJ, ON G..‘J1ib‘a -‘ -o--“ ‘“ “fl—O “-. -n— g... Interviewer ___ Answer Groupé"- Number I "‘" II :1 I V v I of (1: ea};) (All No (W—O Incon- (B-O Incon- (Con- Schedules Answers) sistent) sistent) sistent) Taken Indiana 1 6.25 37.5 12.5 31.25 16.7 16 2 h.8 66.7 16.6 0 11.9 A2 3 8.6 57.l 17.l 2.9 lh.3 35 Michig— 1 25.8 16.1 9.7 O h8.h 31 2 20.u 20.3 22.2 11.1 25.9 Sb 3 25.9 0 3.7 33.4 37.0 2? aChi- -square tests were si ni1icant at les othan 30 per cent. Group IV is not included because of the snalL numoer of schedul'es in this group. V \r It is apparent that Indiana interviewers, in contr st to Iichigan interviewers, had a high proportion of schedules fall into Group II. From comments written on the schedules by the Indiana interviewers, the following conclusions might be drawn.1 The interviewers often accepted a respondent's first impression that the questions dealt with gambling ‘1 (which they felt wa~ immoral) and did not attempt to rephrase the C questions to include the risks of farming. The second fact or the striking similarity between Groups III and VI on gains may be, at least partly, explained by the significantly dif- ferent results obtained by interviewers in the following states: Kentucky, Michigan, North Dakota, and Iowa. This comparison is given in Table 32. Examination of tne distribution of schedules within and between states snows extreme differences among interviewers between Groups III and VI. Some interviewers were prone to get Group III answers whereas others were prone to get Group VI answers. One would xpect in a random sample that approximately similar preportions of scledules of the two groups would be taken by the different interviewers. If these inter- views were random from the total sample, comparisons between the groups would probably snow similarities in the characteristics previously dis- cussed. 1The following are direct quotations from Indiana schedules. Interviewer 1, "Doesn't take chances like these." Interviewer 2, "wouldn't gamble except for worthy purpose." "Lot much of a gambler, anything on this line not favorable considered." "This is lottery, too much against one." Interviewer 3, "Do not believe in gambling and this is interpreted in this way." "Doesn't believe in this kind of gambling, is rot constructive." TABLE 32 PROPORTION or EACH Awswri GROUP on GAIvs Tigew BY INTEiVIEwris IN THEIR RESPECTIVE sums“l Interviewer answer Group? I II III V VI Number of (Weak) (All No (W-O Incon- (Not (Con— Schedules Answers) sistent) Answered) sistent) Taken Kentucky 1 11.1 0 22.2 11.1 55.6 18 2 31.25 18.75 12.5 0 37.5 16 3 25.0 12.5 25.0 12.5 25.0 16 A us.u 27.3 15.2 o o 11 hichivan l 25.5 16.1 9.7 O h8.a 31 2 20.h 20.h 22.2 11.1 25.9 5h 3 25.9 O 3.! 33.3 37.0 27 Northqgakota 1 27.8 55.5 11.1 0 5.6 18 2 7.5 5.0 35.0 15.9 37.5 no 3 28.6 1h.2 0 28.6 26.0 7 1 20.0 30.0 20.0 5.0 25.0 20 2 27.3 27.3 9.0 9.1 27.3 11 3 5.6 27.8 50.0 0 16.6 18 h 50.0 20.0 10.0 0 20.0 10 8All chi-square tests were significant at no less than 30 per cent. Ohio, Kansas and Indiana interviewers did not show the above patterns for Groups III and VI. Group IV is not included because of the small number of schedules in this group. b \ (f. F\) Likewise, the similarity between Groups C and F for losses may be explained by different results obtained by the respective inter— viewers. The comparison between interviewers is presented in Table 33. The fluctuation in percentages of Groups C and F between inter- viewers is particularly noticeable in Kentucky, Indiana, Iowa, and Kansas. The range from no group F schedules to 77.8 per cent of this 'group is the striking feature of this table. The range of proportions of schedules in Group C is from none to h5.5 per cent when all the interviewers are considered. It is quite likely that if the interviews that include the loss questions appear at random from the total sample, comparisons between the two groups will show similarities in their other characteristics. Summary This chapter has shown how the answer groups are constructed and 1as attempted to explain the differences between the answer groups on the basis of attribute and other data from the total schedule. Since there is no model for predicting the ex ected numbers of schedules in each of the answer groups, only a qualitative evaluation can be made of this schedule and interviewers witn respect to their effectiveness in eliciting responses. The following conclusions are apparent: (1) in using this technique to measure utility, it is easier to get answers to the loss questions than to the gain questions, (2) the technique is difficult if not impossible to use with individuals who display little evidence of need to participate in insurance or in gambles for large gains, i.e., individuals who are relatively older, LIJ TABL' 33 PROPoariow or EACH Awswsa GROUP ON LOSSES TAEEN BY INTERVIEwais IN THEIR RESPECTIVE STATgs' w Number of Interviewer Answer Group ;_ Schedules A B C D E F Taken Kentucky 1 11.1 0 11.1 11.1 11.1 55.6 18 2 A3.8 6.2 25.0 o o 25.0 16 3 25.0 12.5 o 6.2 12.5 h3.8 16 u 5h.> o 36.u o 9.1 o 11 Ohio 1 2 2 C h.h 0 15.6 77.8 hj 2 21.7 0 21.7 u.3 17.u 3h.8 23 Indiana 1 12.5 6.3 b3.8 0 18.7 18.7 16 2 11.3 b.o 23.8 7.1 u.8 h5.2 L2 3 28.6 5.7 25.7 5.7 2.9 31.1 35 Michigan , V 1 ' 22.6 3.2 3.2 3.2 3.2 6h.5 31 13.0 1.8 20.h 7.4 11.1 h6.3 5h 3 7.h 0 18.5 J 7 18.5 51.8 27 North Dakota 1 11.1 11.1 0 0 0 77.8 18 2 5.0 2.5 17.5 12.5 10.0 52.5 hO 3 1h.3 1a.3 o o 1h.3 57.1 7 Iowa 1 15.0 0 15.0 0 5.0 65.0 20 2 18.2 0 h5.5 0 9.1 27.2 11 3 27.8 5.6 27.8 16.7 0 22.2 18 h 0 10.0 50.0 10.0 0 30.0 10 Kansas 1 0 0 28.5 0 lh.3 57.2 7 2 28.6 1h.3 21.h 7.1 21.h 7.1 1h 3 O 0 0 100.0 0 0 l h 16.3 2 0 38.8 0 12.3 30.6 h9 aAll chi-square tests were significant at less than 30 per cent. _.. v .Kfclws. but. "‘.«H..~...i|. a‘.r..£oHDl. I .Iu* have more farming experience, fewer dependents, high net worths, low debt positions and own their own farms, and (3) the effectiveness of the technique in eliciting answers is closely related to the inter- viewers' ability to make clear to respondents the meaning of the various hypothetical situations. This latter conclusion is substantiated by two illustrations. That is, first, it is easier for an interviewer to accept an answer of "I don't gamble" and deduce from this that the respondent's answer to the gain questions would be all No, than it is for an interviewer to interpret the questions in a meaningful context and then to press the respondent for answers other than all No. The second illustration of possible interviewer bias is the apparent similarity between individuals who were consistent with the hypothesis and those who were inconsistent. Whether or not some interviewers helped the respondents acquire con- sistency is not known. In future studies the hypothesis of whether or not helping the reSpondent maintain consistency affects the usefulness of reSponses beneficially or adversely could be tested. Such hel could bias reSponses or produce more usuable unbiased responses. The next chapter will consider further the two types of consistencies for gains and losses. As a consequence of interviewer bias there is a high probability that many Indiana farmers are misrepresented by inclu- sion in Group II; thus, these schedules will be eliminated from the analysis presented. Group B, on losses, is similar to Group II in that both are consistent with a diminishing marginal utility hypothesis; however, Group B will not be used in the next chapter because of the oi ’4) “I small number of schedules. Groups I, VI, A and b remain to be analyzed more extensively in the next chapter than was possible here. The conclusion concerning the interviewer bias between Groups III and VI ad C and F is believed not to affect the usability of the data in the next chapter. The similarity of characteristics between the sroups indicated that either some data were lost in Groups III and C or some were gained in Groups VI and F. If it is true that some data were lost, then this means that valuable observations in testing the hypothe- ses are not available. However, if it is true that some data were gained, then two situations may exist. The observations gained may or lnay not be random. If they are not random, they may be unbiased depend- ing upon whether or not the lack of randomness reflects true relation- ships among the variables involved or interviewer bias. If they are :random then the additional observations create variance which may obscure Imit not bias estimates of the true relationships. As Groups VI and F arui Groups III and C are made up of demonstrably similar individuals trugre are grounds for supposing that true relationships will be Iwaflected when Groups VI and F are analyzed. CHAPTER VI hii~1IV'ATION AISD EVALUATION or amnion UTILITY FLHECTIL’JEIS This chapter contains the main analysis of this thesis. Numerical utility functions are estimated for those respondents showing consistency in their responses. Marginal utility estimates are derived from these total utility functions and used to classify the individuals into derived types. These types are then related to attribute and behavioral vari- ables. This analysis contributes substantially to the general evaluation of the utility measuring technique employed in this thesis. Further, the utility estimates are used to predict other kinds of behavior. More specifically, only individual responses displaying the types of consistencies defined in the previous chapter under I and VI for gains and under A and F for losses will be analyzed. The weak types of consistencies defined under I and A are not inconsistent with the indifference axiom or with the hypothesized utility function, but were considered a weak form of consistency. The much stronger types of con- sistency defined under VI and F are those which are consistent with the indifference axiom and with the hypothesized utility function in that they permit computation of one or more indifference points. The location of the indifference points for the latter consistent types will be the basis of the analysis, i.e., an estimated utility function will be fitted to these points and related to attribute, behavioral items, and pre- dicted behavior. First the utility functions will be related to the following attributes: 9b \C \‘l 1) State in which the schedule was taken 2) Respondent's years of farming experience 3) ReSpondent's age a Number of respondent's dependent 5 Type of farming engaged in by the reSpondent 6) Number of years respondent attended school. On‘a prior basis, these characteristics should be associated with the marginal utility an individual attaches to different quantities of wealth. If a complete and meaningful representation of an individual can be given by combinations of these characteristics, then predictions of the marginal utility attached to wealth by other individuals with the same characteristics as those in the sample could be made. Second the utility functions will be related to the following characteristics, here denoted behavioral items, which are interpreted as the indirect or direct consequences of the reSpondent's managerial action: 1) Net worth of respondent 2) Average gross income for a three-year period 3) Debt position (amount of debt in dollars) h) Proportion of total acres rented 5) Proportion of income from farming 6) Ratio of debts to assets 7) Concern for two types of error 8) Attitude toward informal insurance schemes, In agreement with the theoretical nature of man offered in Chapter I, a cause and effect relationship is hypothesized between these behavioral _‘—‘" ‘T ‘-—“—1 .___ _A ~ :w—r-—-.-— items and utility. This means that the same stable and consistent utility function manifested in an individual's answers to the hypotheti- cal situations, motivated the behavior implied in the above list. Third and lastly the utility functions will be used to predict the amount of gain necessary to induce the respondent to accept an unfair risk (gamble) and the amount of loss necessary to motivate him to accept an unfair insurance scheme. In the first section of this chapter a method will be developed for regrouping the individuals in consistency classes of Chapter V into "derived types." In the second section the relevant data for comparing these types will be presented while in the third predictions from the utility function are made. The last section will present a concise evaluation in regard to the technique's effectiveness in providing numerical estimates of cardinal utility. Develgpment of Derived Types The method used in reclassifying the individuals of answer groups I, A, VI and F was, first, to derive a utility function for each respondent; second, to make an estimate of he marginal utility per dollar by taking the derivative of the estimated utility function; and third, to classify on the basis of the estimate of marginal utility. Derivation of Utility Function From the indifference relation a u + (l - a) v = w and from Theorem 1, it is possible to compute q(u) given a unit of measure and an arbitrary origin. twuri .‘nii m. 0 “-2' .f 3 an. (a q: .'|.. .. \_'l .4.....a. .5 {.0 ....,. 99 Let q(u) the numerical utility at a specified gain or loss q(v) = the numerical utility at ‘the other alternative q(w) the numerical utility at a certainty position 62 the probability of obtaining the gain or of incurring the loss Let the unit of measure, say one util, be equal to1 q(w) - q(v) = 33—3—5 Then C1(v) = QW) - hill-2- Substituting into a q(u) + (l - at )q(V) = q(W) gives or(q)u+ [(1 -0¢) [q(W) ”Po ‘Pz ] J = q(W) 5 Multiplying out the brackets and dividing thru by a; gives q(u) _ [POS - P2] [P0 - P2 8 q(w) Factoring and transposing gives q<fi>=q(w)+ 1——=———°‘][P—_O-———T'P 1 Let the ori gin be where q(w) = O and P0 = O {es-1m since 32 is a negative quantity. then __ 1Thus the total utility function is assumed to be a straight line between 0 and -8 utils. P0, P1 and.P2 are specified in Chapter IV. "Til Ilitr I..v|'DrD}.|Iv‘D . . “INS-Hy .4 With the equation) q(u) = 1 gno‘:}-[¥E3i1 , it is easy to compute the utility attached to a Specified gain or loss. Thus, for example, suppose a respondent said No to the 1,000 dollar gain situation and Yes to the 5,000 dollar gain situation on the fair series, then the o ‘ o u 5:; n .v .7 ' l o 1 o o v a indifference pOint where 0d = ————- is, by linear interpolation, 120 near an amount equal to 3,000 dollars. Substituting into the above 119 equation gives q(u) = 150 ' ‘%3 = 595. 20 1- 'u \ This computation was made for all the possible incifierence points. Thus, for each respondent depending upon the number of indifference points present in his answers, a point was or a number of points were obtained indicating the nature of his utility function for gains or for losses.1 Using the method of least squares or Lagrange's interpolation formula, the equation '3’= ax + b}:2 wherelu'is the estimated utility and X is the amount of gain or loss, was fitted to each case in which there was at least one indifference point after replacements and transposi- tions.2 The methods of fitting and the assumptions made about the 1For those cases like Group I and A where there is no indifference points the utility curve is bounded from below by the utility or dis- utility of the odd which was answered Yes and above by the utility or disutility of the odd situation which was answered No. 2For the fitted curves on gains the most meaningful difference found between the adjusted cases and those which were consistent with the hypothesized utility function without adjustment was the number of years of school attendance. Almost 80 per cent of the respondents who had attended 8 or more years of school were consistent without adjustment, while almost ho per cent of those who had attended school for less than .7”-"-l‘m v .9. lCl location of the indifference points and other identification points are shown in AppendinD. It should be stressed that the reader can not fully understand the derivation of the utility functions without careful study of that out- line. However, the continuity of the chapter is enhanced by not inter- rupting the sequence of analysis with the outline at this point. /\ From the fitted equation u = ax + be the derivative or estimated . . . . . (lir , , ,. . . . . marginal utility'af- = a + 20x was found ior the follOWing pOints: “7' X. (1) 3,000 dollar gain and loss (approximate cost of a new car), (2) 7,300 dollar gain and loss (approximately average annual gross income of respondents), and (3) 30,000 dollar gain or loss (approximate value of a small farm). This marginal utility represents the amount of utility attached to one additional unit of wealth relative to the utility attached to one unit of wealth at the present income position. As a relative marginal 8 years required adjustment to be consistent with the hypothesis. The chi-square test was significant at 3 per cent. A similar test for the fitted curves on losses was not significant. The chi-Square test of independence between the state in which the schedule was taken and the adjusted and unadjusted cases for losses was significant at l per cent. Although six of the states had a high proportion of unadjusted cases, Kansas ha to have 50 per cent of its cases adjusted to be consistent with the hypothesis. For the cases with no indifference points there were 13 out of 105 schedules on the losses question that were changed to consistency. 0n the gains questions that were answered all No, one out of the 123 cases required adjustment. Out of the other 95 cases of weak consistency for gains 29 were adjusted in order to snow consistency with the indifference axiom. No meaningful variable could be found in the rest of the schedule to explain the difference in answers. ‘-“..‘2' x “it 5“. «(lav—r...“ “a“ u “—‘b 102 utility this quantity is interpersonally compariable, but it is not interpersonally comparable as an absolute marginal utility. It is this relative quantity which is crucial in the decision making theory of the marginal analyst. It is this interpersonal comparability which makes it possible to group individuals into the derived types. The s" estimates provided by this thesis can in no way b: useful to welfare : economics. E This procedure of grouping is intended to provide data for testing f 1 the second hypothesis presented in Chapter . The procedure is based L upon the belief that the mean and variance of a group in respect to the relative marginal utilities attached to different amounts<1fwealth correspond to (fits) the distributions of the attribute and behavior data. That is, individuals who attach relatively small marginal utilities to additional wealth will behave similarly; whereas groups of individuals who attach relatively large marginal utilities to addi— tional wealth will behave similarly, but differently than the former group. In total,30 derived types were formed on the basis of marginal utility, 15 for gains and l5 for losses. The 15 groups consist of three sets of five groups defined by the derivative at the 3,000, 7,500 and 30,000 dollar gain or loss. For the gains each of the five types contain two groups for which the marginal utility is hypothesized and three for which it is esti- mated from the utility function. These latter three groups were formed 103 on the basis of the number of individuals in eael, i.e., one large middle group with two smaller extreme groups. Each of the iive types for losses contain one group for which the marginal utility is hypothesized and four for which it is estimated ~ 1 from the utility function. These four groups contain two large mic" e groups and two smaller extreme groups. finalvsis of Relationships Between Empirical Utility *uuctiarr and Ae evant Jariaoles _-‘.—.—’¢ ’f-u-MoI-d-ccc—n‘.“ “.4... ~“. AW . u . . a . This analysis is based upon rather loosely formed hypotheses con- cerning the character of individuals who have relatively high marginal (dis) utilities for gains (losses) compared to those who have relatively 1. low marginal (dis) utilities. The characterization is necesr“"i1y ahpirical and loose because no eneral model has been constructed that r U correlates the shape of the util’""r curve with the character, diSpo— sition, and behavior of individuals. Derived Types for Gains Questions A statistical comparisor hetween the groups formed upon the basis of the derivati*es of the total utility curve at three difterent rains ($3,000, $7,500, and $30,000) showed that the 30,000 dollar derivative produced types.not only “elated to more of the other variab as but related more sifinificantly to meaningiul variables than the other two derivatives. Therefore, the analysis reported in this section is confined to the 30,000 dollar derivative. . . I II.-.» v. I Ill.-Vl|..'ln 0’5 4 ‘|.u.|» Infill. . ... . as. w . . . . r 101:. The five types compared are defined and shown with the estimate of the relative marginal utility per dollar in Table 3h. Those types v which are defined by a derivative (called derivative types) are the ca go (0 ( D U) with indifference points and differ from each other by the relative marginal utility per dollar attached to gains in wealth. The tvpes without difference points are defined by the answers on the schedule; however, the relative marginal utility per dollar for these types is hypothesized in relation to the derivative types. The respective number of individuals in each group is also shown. The convention of denoting types for the gain questions by Roman numerals will continue to be followed. TABLE 3h RELATIVE EARGINAL UTILITIES F‘R DZRIVJD TYPES AND NULBEJ 0F RESPONDEKTS AT 30,000 DOLLAR GAIN .1 Relative Type Depining Conditions harginal huMber Utility of Per Dollar Respondents I All No to Fair or to Unfair Odds 0.1h or 0.26 62 II Derivative 0.0 or 0.30 hl III Derivative 0.31 or 1.00 65 IV es to All Odds 1.56 33 V Derivative 1.01 and over )5 -* *“---‘- ~Ww¢fl “II-0‘“- In the subsequent discussion the following procedure is used. First, since Type II differs from Type I by the presence of indifference 7‘14“": .. I. .II. t. 105 points in its answers and not by marginal utilitgf, thev are paired for comparison. Second, since Type V di fie rs from Type II by th resence of indiffe ence points in its answers and not by marginal utility, they are also paired. Third, the three with indifference points or the derivative types are compared with each other. In Table 33 the average net worth, income and debt position of each type are shown. TABLE 35 'Wv—I .___. ._._._.__._._.____,‘,.l AVEIWE NET WORTH, 021088 III"0HE AND DEBT POSIT I01 S FOR THE DERIVSD TIPBS 0N 30,000 DOLLARS GAIN . Type Della Net Worth Gross Income Debt I (M.U., .lh or .26) u8,395 7,917 5,632 II (u.U., 0- .30) 33,86u 7,579 1,997 III (M.U., .3I-I.00) A2,837 8,064 2,609 Iv (h.U., 1.56) 38,959 7,159 2,716 v (M.U.,>~l.Ol) 36,720 7,357 3,256 ——7 In Table 35, Ty e I differs significantly from Type II on both net 1' worth and amount of debt. Type V does not di er significantly iron Type IV on any of the variables. Type III 1as the highest net worth of the three groups with indifference points, but Type II has the lowest net worth and debt position. None of the types differ significantly with respect to average gross income. 106 The average number of years of farming experience, age of re- Spondents, and number of respondents' dependents are given in Table 30. TABLE 36 AVERAGE LENGTH OF FARHING EKPERIENSE, AGE AND NUthi 0F RESPONDENTS' DEPENDENTS FOR FIVE DERIVED TIPES ON 30,000 DOLLAR GAIN -:k’! T: Type Years of Number of ; Farming Age Dependents 2 Experience 7 I (M.U., .lh or .26) 19.7 u6.5 3.2 E II (r.U., 0- .30) 15.6 h3.u 3.2 ' III (h.U., .31 - 1.00) 17.2 L6.6 2.6 IV (M.U., 1.56) 20.7 h9.2 2.h v (M.U., > 1.01) 17.7 h3.h 3.1 Type I individuals have significantly more farming experience than Type II individuals. However, they do not differ significantly as to age and number of dependents. Type V are significantly younger, have fewer years of farming experience and have more dependents than Type IV. Types II, III, and V do not differ significantly on any of the three variables; however, Type II individuals are relatively younger with fewer years of farming experience and with more dependents than the other two groups. Only two other variables showed significance by the chi—square test. These are the state in which the schedule was taken and 10? the respondents' type of farm (shown in Table 37 and 38 respectively). Type I is made up of about 25 per cent hichigan schedules while the other 75 per cent was distributed rather evenly over the other six states. Type II contains a high pr0portion of Michigan schedules with Ohio making up the second highest proportion. Type III is almost evenly divided with Kentucky, Ohio, and Michigan each contributing about 20 per cent and the other four states about 10 per cent each. Nichigan contributed one—third of the schedules in Type IV with Kentucky and Kansas contributing the second and third highest respec- tively. Type V is about evenly distributed over five of the states; only Iowa and Kansas contributed a smaller proportion than the other five. The proportion of schedules from a particular state is important when considering the next table on type of farming. The test of inde- pendence between the five derived types and the type of farming was significant (10 per cent) at a higher level than the same test between the derived types and the state of origin (30 per cent). This fact should be kept in mind when examining Tables 37 and 38. Although Type II is made up of a high proportion of Michigan schedules, Table 38 shows over sixty per cent are fat stock, or cash crop farmers. Type II is similar to Type I in having a high pr0portion of fat stock and cash crop farms; however, it has a higher proportion of dairy farms than Type I. Type IV has a high proportion of cash crop farmers whereas, dairy, fat stock, fat stock - cash crop, and tobacco farmers are evenly 5“.—.Jfiu. ‘m’.-.‘ “.4..- ‘n. airman! g ,0 . 108 . :mo hoe on m pmwoa Amman mpmsemnHCa .0. PP. ... ..Lm o.ooa m.® m.© e.aa a.ma H.Ha :.ma m.ma Aao.a A ..euev > o.ooa N.ma H.e H.e m.mm H.a o.m m.em Aem.a ..e.e >H o.ooa N.m m.oa m.ma m.HN N.a 6.0m i.ea Aoo.a - am. ..s.ev HHH o.ooa m.m a.e m.e m.me m.e m.ma m.m Acm. -o ..e.zv HH o.ooa m.ea e.ea m.ma m.em ®.e m.ea m.aa new. no ea. ..e.ev H mpoxwm Hapoe mwmcmm mBOH ephoz cmmflsoflm wmmflbcH oago hxodemm mane armedem Mm 2HHme modm mo ZOHBHmomEOU pm mqmde 109 .pcoo pom OH pm pddoaqedmflm ohmsvmuflzom o.ooa o m.m m.m ®.Ha m.ae e.mm m.m Aao.a A ..:.:v > o.ooa m.e e.ma e.ma a.m H.mm o.ma n.ma Aem.a ..e.ev >H o.ooa ©.e m.e e.m e.a m.mm m.aa m.em Aoo.a - Hm. ..e.ev HHH o.ooa m.m m.m n.m e.ma m.em e.mm n.ma Aom. -o ..e.mv HH o.ooa m.e m.m N.m e.a i.ew H.0m @.m flew. no ea. ..e.mv H macho zmwuu mdopu xoopm Hmpoe pmzpo ooomnoe xoopm pmm Hmhmcoo ammo pew hpflwm make I nzHeo menace 000.6m mes mom mmmwe mmeHmmm 2H ZQHBeomoem .eZHee a mo manna mm mqmde llO distributed. Fifty per cent of Type IV schedules come from Michigan and Kentucky. Contrariwise, in Type V where an almost equal proportion of schedules come from five different states, over do per cent were cash crop farmers. The second highest proportion of farms wan; fat stock farms. The nature of the five derived types for the 30,000 dollar gain may be summarized as follows. Type I (all No to fair or unfair odds) -- Individuals in this group have the highest net worth and debt position of any of the groups. They are older in comparison to the other groups but have a relatively large nunber of dependents. They are mainly cash crop and fat stock farmers. Although 03.h per cent of the all schedules taken came from east of the Mississippi, about 50 per cent of this group live east and 50 per cent west of the Mississippi. Type II (EU of O to .30 utils per dollar) -— Individuals in this group have the lowest net worth and debt position of any of the groups. They are the youngest of the groups and have a relatively large number of dependents. They are mainly cash cr0p and fat stock farmers; however, a higher proportion of them are general farmers than any other group and about 15 per cent are dairy farmers. Over 75 per cent of the individuals in this group live east of the Mississippi, 10 per cent more than for the survey as a whole. Type III (LII of .31 to 1.00 utils per dollar) -- Individuals in this group have the highest net worth of the three derivative types and 111 are about in the middle of the three on debt position. They are some- what older with more farndng experience and with fewer dependents than Type II individuals. Their types of farming are mainly cash crop and dairy. Over 65 per cent of them live east of the Mississippi which is similar to the portion of schedules taken in these geographic regions. 1 Type IV (Yes to all odds) -- These individuals have about the average net worth and a below average debt position. They are the oldest, have the most farming experience and have the fewest dependents of any L of the groups. They are engaged in almost all types of farming with a higher proportion of the fat stock-cash crop and tobacco farms than in any other gr up. A high prOportion of these individuals live in hichigan and Kentucky. Type V (RU of 1.01 utils per dollar and over) -- Individuals in this group do not differ significantly from Type II on net worth; however, they have a higher debt position than either of the other two derivative types. They are younger than the average of the groups but have about the same farming experience and number of dependents. Over 73 per cent of the farmers are engaged in cash crop or fat stock farm- ing. About 60 per cent live in the three states of Ohio, hichigan and North Dakota. The following conclusions appear to be warranted: 1) 0n the average, Type I individuals are significantly different from individuals in other groups. Their action can be rationalized in either of two ways. They may have diminishing marginal utility for ll2 wealth or near constant marginal utility with some risk aversion. A possible cause of their risk aversion could be their high debt posi- tion which might prevent them from taking anything but a fair or a more- than-fair chance. 2) Type IV is more like Type III than it is like Type V. Though this would seem to indicate that the estimate of marginal utility for i this group was high or that the estimate for V is low, an alternative explanation might be the presence of a positive preference for risk. 3) Two factors seem to distinguish the three derivative types. One is the debt position of the respondent. As the amount of debt increases, the marginal utility per dollar of wealth increases. The other factor is the type of farming engaged in by the respondent. Traditionally, cash crop and fat stock farming are considered more risky than the other types. This study substantiates this contention for, as the proportion of these two types of farms increase, the marginal utility per dollar increases. It can be inferred, therefore, that these individuals are more willing to engage in farming with risky enterprises. Dairying, fat stock-cash crop, and tobacco farming are intermediate in the amount of risk involved; a high proportion of individuals in the middle group in reSpect to marginal utility are also dairy, fat stoch- cash crop, or tobacco farmers. The general farm is usually associated with a low level of risk; it was also associated with a low marginal utility for wealth in the above discussion. 113 Derived Types on Loss Questions The statistical comparisons between groups formed upon the basis of the slope of the total utility curve at three different sizes of loss show that (l) the 7,300 and 30,000 dollar derivatives each produced types that were significantly different from each other on several of the other variables, and (2) the types formed at the 7,500 dollar derivative were consistently related to the same variables and in a fashion similar to the types formed at the 30,000 dollar derivative. -Munt"1: Amm-o .‘s‘. amt—1 A I s Therefore, in the subsequent analysis, the comparisons are limited to the types formed upon the basis of the 7,500 and 30,000 dollar derivatives. Types formed at the 7,500 dollar derivative and at the 30,000 dollar derivative will be denoted by capital letters with a subscript-.7.5 ani ~30 respectively. The defining conditions, the estimate of marginal disutility and the number of individuals for each of the derived types on losses are shown in Table 39. The types are arranged approximately by the amount of marginal disutility per dollar, i.e., Type A7.5 and A30 have the smallest whereas Type F7.5 and F30 have the largest marginal disutility per dollar. In the subsequent analysis Type B7.5 or A30 will not be included. Because of the small number of observations, no statistically signifi— cant conclusions can be drawn concerning this group. Table 39 includes these types so that it would parallel Table 3h on gains. Type E7.5 will be compared with D7,5 and D30 with E30 in the following discussion. The four derivative groups will be discussed as a group. 11a TABLE 3‘ RELATIVE MARGINAL UTILITIES F01 DERIVED TYPDS AND YUMDDR 0F RESPCNDENTS AT 7,500 AND 30,000 DOLLAR LOSSE Relative Type Defining Conditions harginal Number of Disutility Reapondents Per Dollar A7.5 Derivative 0 to .20 37 87.5 No to fair or unfair odds .1A or .26 '16 07.5 Derivative .21 to .u0 79 D7.5 De"ivative _ .hl to 1.00 72 E7.5 Yes to all odds 1.56 70 F7.£5 Derivative 1.01 to 7.00 51 A30 No to fair or unfair odds .1h or .26 16 B30 Derivative ' 0 to .DO hh 030 Derivative .ul to 1.00 69 D30 Derivative 1.00 to 2.00 60 E30 Yes to all odds 1.56 70 F30 Derivative 2.01 to 20.CO+ 58 The average net worth, gross income, and debt position for each set of derivative types are shown in Table DO. Type 37.5 has the highest net worth and income1 and the lowest debt position of any of the groups. It is more like F7.5 than like 1Average incomes are not significantly different between any pair of the types. 115 AVERAGE NET WORTK, GEOSS Income AND DEBT POSITION 0F xssPoxhzhrs m max-1117210 TYPJJS 0N Loss Dollars Type Net worth Income Debt A7.5 (M.U., 0 - .20) 38,958 8,757 h,027 c7,5 (M.U., .21 - .ho) u3,863 7,9h7 2,696 07.5 (M.U., .hl - 1.00) 39,130 7,812 0,105 37,5 (M.U., 1.58) _ 5h,078 8,011 2,083 F7,5 (M.U., 1.01 - 7.00) 32,800 7,830 2,102 B30 (h.U., o - .uo) 39,;83 8,808 2,5h8 030 (h.U., .21 - 1.00) h3,09h 7,921 u,002 D30 (M.U., 1.00 - 2.00) 36,899 7,819 3,802 330 (h.U., 1.58) 58,078 8,011 2,003 F30 (M.U., 2.01 - 2o.00+) 33,6)8 7,2h7 2,2u7 I%,5 in.debt position. Of the four derivative types, A7.s is most similar to 11.7.5 on net worth and debt position, but 07.5 has a high- er net worth and a lower debt position than either A7.5 or D7.5. Type F705 has the lowest net worth and debt position of the four. In tne lower half of Table h0, Ego again has the highest net worth and the lowest debt position of any of the groups. It is similar to F30 on debt position. 0f the four remaining types, 030 has the highest net worth and debt position. Type 330 and DSO do not differ significantly on net worth but D5O has more debt. Type D30 does not differ significantly on net worth from F30 but they are different on debt position. In general, after excluding E7.5 or E30, it appears that those individuals who have a low marginal disutility for losses have relatively high net worths and high debt positions. This conclusion is completely distorted if those individuals who took all unfair insurance odds are included. These individuals seem to be unlike any of the other groups. I A hypothesis that these individuals have a positive preference for security may explain their answers to the loss questions. Table hl shows farming experience, age, and number of dependents for each derived type for losses. For the three variables shown in Table hl there is no significant difference between the four groups based upon the derivative of the total utility function; however, type E7.5 or Ego is significantly different from them. Individuals in type E7.5 are older, have more farming experience, and have fewer dependents than any group. The same is true for the comparison between the four groups and ESQ except on the number of dependents where 030 has the largest number. The derived types on losses were also found to be significantly dependent on three other variables. These are: (l) the state in which the schedule was taken, (2) the type of farm, and (3) the concern for the two types of errors. These comparisons are shown in Tables h2, h3 and uh respectively. TABLE hl 117 FARHING EXPERIBHCE, AGE, ADD NUHBER CF DEPENDENTS OF INDIVIDUALS IN DERIVED TIPES ON 7,500 ALD 30,000 DOLLAR LOSSES Type (E'VIOU. ’ , (I'loUo, (15:01.10 ’ (M.U., (h.U., O - .20) .21 - .h0) .h1 - 1.00) 1.58) 1.01 - 7.00) 0 - .50) .hl - 1.00) 1.01 - 2.00) 1.50) 2.01 - 20.00+) Years of Ferminv experience 19.2 A57: 3 Number of Dependents .0 U) 118 .pemo pom H pm pcmoflmflcmflm we mQOHpomm neon pOM opmdvmuflnom o.ooa e.m m.o o.ma o.em e.am e.m m.on+oo.om . Ho.m ..p.ev one o.ooa o.oa a.NH H.e 0.0m m.:w o a.mm Aem.a ..p.mv 0mm o.ooa e.e m.oa m.aa a.mm m.oa e.ea e.w Aoo.m . Ho.a ..:.ev 0mm o.ooa a.ma m.e e.am m.ma H.0H m.om e.© Aoo.a - He. ..p.ev one o.ooa m.e m.ma e.ma m.na w.o e.am J.HH flog. . 0 ..:.eV 0mm o.ooa m.m a.m e.am m.em m.mm w.a m.e Aoo.e . ao.a ..e.:v u.ne o.ooa o.oa a.ma a.» 0.0m m.:m o e.mm Amm.a ..:.:v w.pm o.ooa a.e m.ma e.a m.mm a.ma m.ma o.m noo.a - an. ..D.av m.em o.ooa :.Ha H.m o.ma m.ma :.HH a.mm H.0H Ace. . Hm. ..:.:v m.eo o.ooa H.© e.aw a.wa m.eH N.N m.wa m.ma AoN. . 0 ..:.eV m.e< 090303 Hmpoe mwmcww mSOH spec: cmmfleoflm mdmflpCH oaso hzodpeom make mmmeaem Mm mammoa eeaaoa ooo.om axe oom.a ac mama mmsHemm moem mo oneHmomzoo m: mqm.12 n5.5 22.7 31.8 100.0 a .. . ... Cal-square Significant at 2 per cent. TABLE 118 CONCERN FOR TWO TYPES OF ERROR BY BBSPONDENTS WHO WEBB WEAK AND STRONG CONSISTEKT ON BOTH GAINS AND LOSSES, LOSSES ONLY, AND GAINS ONLY ‘ _ reportion Concerned AA Consistency b c d Form lst 2nd equally Total Both 26.8 30-h h2.8 100.0 Losses only 30.1 38.u 31.3 100.0 Gains only 21.1 _ b3.8 3h.8 100.0 :Chi-square significant at 20 per cent. More concerned about their taking action when should not. :More concerned about their not taking action when should. Are equally concerned about both types of error. 131 individuals who we'e consistent on either the losses only or the gains only were more concerned about not taking action when they should than about the other two possibilities. Aside from school attendance, several hypotheses explaining the relation between the form of consistency, the reSpondents financial position and his concern for the two types of errors can be given. One possible hypothesis is that those individuals who were consistent on gains only, have gone into debt to make supposedly profitable invest- ments. Their desire to get out of debt is evidenced by their being consistent on gains only and by their concern for not taking action when they should. An alternative hypothesis is that these individuals have a positive preference for risk taking, evidenced by their being con- sistent on gains only and their concern for not taking action when they should. Their relatively high financial position and debt position may be eXplained by profitable investment with borrowed funds. Still a third alternative is that they are "newly rich." Their consistency on gains or their inconsistency on losses may be rationalized by their state of confusion after becoming rich. Neither this hypothesis nor the others can be tested by this thesis, as unfortunately data on changes in financial positions were not obtained by the I.M.S. survey. The individuals who were consistent on losses only can be explained by several hypotheses also. These would be quite similar to the ones stated for gains but would be interpreted to be consistent with the loss situation. For example, these individuals may be inconsistent toward gains because of a recent change in their financial positions. This 132 change may be manifested in their greater concern for the first error in contrast to the individuals who were consistent on gains only. Involving Weak Types of Consistency Only Of the 135 cases which were only consistent with the indifference axiom, 13 were consistent on losses only which is not enough cases for analysis. Of the individuals who were consistent on both gains and losses, 70 percent answered Yes to all he odds. It appears that these individuals have an extremely high marginal (dis)utility for both gains (losses). However, with situations of more unfair odds and a greater range of gains and losses than were included in the I.M.S. schedule, this hypothesis could be tested. The other 30 per cent had combinations of all Yes or all No answers to the different odds. A more extensive schedule could also explore these cases. Comparisons by states showed no significant dependence between the consistencies on both and on gains only, on one hand, and the state in which the schedule was taken, on the other. The average net worth, gross income, and debt positions of indi- viduals who were consistent on both gains and losses and on gains only are shown in Table h9. Individuals who were consistent on gains only have the highest net worth, income, and debt position. They perhaps thought they could get themselves out of debt more easily by taking small chances of large gains than by insuring their present position and continuing in their present endeavor. I“; fdfi 133 'fi-I'v .... A V. " .‘\ “‘2’" '- ’ ‘ ‘."'.f. ‘3‘ -‘~»" - “"" ' “"‘_-7 AJniLGN NET W iTn, GLOSS LnJOUJ, AND DSJE Fbi Nn" “ “QT“‘ fl“ :LI. ‘J QHJJ‘.‘ I ON BOTH GAiNS AN? LOSSES LTD GAINS OULY Consistency Form _ ‘# Dollars E3 werth Gross ncone Debt Both u8,7oo 7,69; 2,283 1 _° n - U} '1‘" ' f. f Gains only )4,2J( 9,10c ,1 ,9 u \ {\D (X) 0\ _— The only other two significant variables were again the number of years of school attendance and their concern for the two types oD error. I ah The first of these is presented in Table 50. I ~—. ‘ r ‘ "‘r‘. 'r‘=' . ‘1 "t " "\T‘") ‘| ‘) T1 ‘rr ‘ ."“"."1 ‘ “_.El FORhS Or WLAL CONSloTnnCY 31 TH; NUnnnn Ob YEAJS Or SLnCOL ATTAMJAJOA Years of Qonsistency Kern - School Attendance Both Gains Only _h‘ _‘ _._‘ < 8 57.1 h2.9 8 76.1 23.9 8 to 12 ué.7 ' 53.3 12 N 59.2 no.8 >-l2 29.h TO 5 .v _4_‘ —~—- - ———-‘ *“w— a . . . . . _._ L Gin-square Significant at one per cent. 13a The individuals who a Wt nded school for ei5ht years had the higli- est proportion of consistency on both gains and losses, while those who attended for more than 12 years had the lowest proportion. Those who attended more than 12 years had the hirh est per cent consistent on gains only. The two groups' concern for the two types of error is shown in Table 51. Individuals who were consistent on both gains and losses are equally concerned about the two types of error. TABLE El MLCH Mi OgZTHETMDTEES(% 1M"1SYIWIII)flStmObm.LNS‘ COKSISlEhT CN BOTH HE GAIKS AND "EIE LOSSES,A1D GA.U S ODLYd -———- “3.- “....“— w—v-w “- A _ r . -. Pro or ion Concerier ConSistency rorm —- —Pw U .__ .b n .c l d 13t and equally Total Both 30.h 17.h 52.2 100.0 Gains only 25.5 hl.5 32.7 100.0 a . . . .p. __< _ Chi-Square is Si5nilicant at one per cent. More concerned about taking action when should not. O.i-iore concerned about not taking action when should. "Are equally concerned about both types of error. 0 Individuals who were consistent on gains only are more concerned about not taking action when they should. lh ese results are similar to those found for all types of consistencies above. L3 Involving_$tronc Types ofi_ponsistency Only The number of individuals who were consistent on losses only outnumber those consistent on 5gains and losses by about ,0 pm cent and those consistent on gains only by 300 per cent. The average net worth, income, and debt position for th three groups are presented in Table 52. TABLE 52 Alli LAG J11: 1131‘ WUI'ITH, GROSS THOSE-13,15,731) DTBT FOR. ST: {01.11 CC;1'TS_. T“"‘IUS 0.1-; GATDS Al I) LCXSSLS, LOSS 13 OI‘TLY AlID GA._IIS 0111.11 —.—-—~ — .“-§-. .— m - ...—b“ _‘ ‘- “ ... Consistency Form Dollars h t wbrth Gross Income Debt a: g (r! cr’ 150th )/,l\9 7,0j0 2,}{3 . 1 ’4 1' /~ 5 0’ Losses only 39,212 8,139 ),203 Gains only h0,73h 8,120 2,815 No significant diff wr nc es were found amon5 the three groups on these variables. The by-states comparison is shown in Table 53. Except for Kentucky and Ohio, the greatest proportion of the sched- ules were consistent on losses only. The ese two states had a hi5her proportion that were consistent on both. Iowa had the lowest proportion of schedules in tile consistent cate5ory for both gains and losses. The two other si ' nificant variaoles (1) school attendance and (2) concern for the two types of errors are presented in Table Sh and 55 respectively. There is an indication in Table Sh that those individuals who were consistent on both gains and losses had attended school more years than those who were consistent on either losses only or on gains only. TABLE 53 at = » C“LSISTENCY or 3 TH GATES AND LOSSES, Lee as ONLY AND GAINS ONLY WITHIN STATESa Consistency Form Both Losses Only Gains Only Total Kentucky h1.h 31.0 27.6 100.0 Ohio Sh.h 39.1 6.3 100.0 Indiana 21.6 67.6 10.8 100.0 Michigan 32.u u7.3 20.3 1o0.0 North Dakota 31.8 59.1 9.1 100.0 Iowa 12.9 61.3 25.8 100.0 Kansas 22.2 51.9 25.9 100.0 a,,. . , , . y . uni-square 15 Significant at one per cent. The proportion consistent on gains only seems to be skewed more toward fewer years of school attendance than the other two groups. The respondents' concern for the two types of error is represented in Table 59. 1 The pattern that has oeen present in the two previous tables on concern for the two types of errors is again shown in Table 73; those individuals who were consistent on both gains and losses are equally concerned while those who were consistent on losses only and gains only are more concerned about not taking action when they should. It appears, regardless of the type of consistency, that individuals who were consistent on gains only have the highest net worth, a high TABLE Sh IXSR CENT OF INDIVIDUALS WHO ATTENDED THE SPECIFIED NUJBER OF YEAfiS OF SCHOOL FOR THE STRONG CONSISTENCYa L (hansistency Form Both Losses Only Gains Only < 8 7.5 11.6 22.5 8 32.3 37.0 26.5 8 to 12 20.h 1h.u 20.h 12 33.3 25.0 214.3 >-12 6.3 11.0 6.1 Total 100.0 100.0 100.0 TABLE '55 COECERN FOR THE TWO TIPES OF EdROi BY INDIVIDUALS WHO Wfldfi STRONG CONSISTENT ON BOTH GAINS AND LESSES, LOSSES ONLY, ASH) GAINS ONLY g __ __._‘ ..___‘ __. _— g ‘ !_._- ___‘ _._ ‘7 v— ‘0 . o (3onsistency Form 1P1oportion Concerned lstD 2ndC Equallycl Total :Both 30.7 27.3 h2.0 100.0 :Losses only 21.h h2.9 35.7 100.0 Gains only 25.0 h1.7 33.3 100.0 a Y O I C ‘3 C I Uni-square Signilicant at 20 per cent. More concerned about their taking action when should not. c“ . . . , more concerned about not taking action wnen snould. d . w . - Are equally concerned aoout both types of error. 133 debt position and are concerned about not taking action when they should. In general, individuals who were consistent on losses only are not much different than those who were consistent on both gains and losses except that the first group is more concerned about not taking action when they should while the latter groups are equally con- cerned about the two errors. Summary -— Several hypotheses could be formulated that would rationalize the action of individuals in the various groups. One is that individuals who were consistent on gains only are prone to take risks and ha*e thus gotten themselves into debt but have found that their "gambling" has paid off in a high net worth and income. Another possible hypothesis is tLat these individuals, in spite of their high net worth, are heavily in debt and will take a "ganble" in their attempt to get out of ciebt rather than insure against losses. Individuals who were consistent on losses only may be extremely conservative, infrequent users of credit facilities, and fre uent visitors of insurance salesmen. Or perhaps these individuals believe their relatively lower net worth position can be protected from less more easily than it can be increased by taking chances on gains. Still another explanation for these individuals' action is that they have just experienced a change in their income or net worth position. The fact that respondents were in some cases consistent with the hypothesized utility function for gains only or for losses only must reflect upon two aSpects of this study. One of these concerns he o 0 ’ I"! ypothesis itself and tzie other concerns the technique Oi T21e hypotm sis is intended to rationalize indiViduals behavior who simultaneously 5 rhle and insure. The results may mean that some individuals do not in fact do both. Wleuher or no t tlre 73*pot11es i"ed utility function applies to these individuals is a matter ior fu‘tner Speculation. The reflection upon the technique of utility measurement could be from many sources. One is that by the time the respondent got to the gains questions after doing the loss questions, he possessed suiiicient understanding to proceed without difficultv. Another is that a person who answered the questions concerning losses was too fatigue d to answer the gains. Although many difficulties remain as far as the technique is con- "i'I cerned, significant diiierences we1 e discove red between the three forms (D of consist =ency. This could mean that the difficulty with the h3.rpot m1 sis outwe ic1s the shortcomin s of the technique. redictions From the Utility Equations When the fitted utility functions are equated to a straigut line fi function from tl1e 01 L in, the intersection points shows the (dis) utility attached to a Specified gain (loss) situation. Thus if: '1 1 c + bi“ = on where + x2 *1..-) (£1- ) and is such that P1 <:PO - P2, i.e., leaves the individual with less than PO, then solving for x produces the minimum amount of gain (loss) necessary to induce the individual to accept the unfair bet (insurance). For any deg es of fairness of odds this procedure can be followed to predict the size of gain or loss necessary to induce an individual with a specified utility function to accept the risk or insurance. For the derivative types on gains and losses, the prediction was made fo "3 unfair odds similar to the ones in the schedule. The predicted amount of gain and loss necessary to induce acceptance for the gains type and for the loss types is shown in Table 56. TAILS. E) O AHOUNT OF GAIN OR LOSS NZCESSAKY TO INDUCE ACCflPTANCE OF AN UNFAIi 0J3 _— w“.~“#¢ “w‘ ~ _.1 *— ___ -..-g- Mu“ Type Amount of Gain {dollars) . H H H <1 H H b: :- -\1 H \o C7“ (3 0 Amount of Loss Qdollarsl_ 92,340 h8,9lO 10,530 1,750 :p x] bit/'0 and 0. 0101010" 137,350 18,130 9,LOO 2,320 0) waU 0000 *TJC‘JOUJ a, Li 1 3. r . ;~' 3 33 1 n,. These are the groups Gellnefl in a pTBVlOUS Scotion of this chapter. 3“) Ilium—‘4‘. u.“ Jim 0L‘ “A“ z . ' I 141 It is apparent from Table 56 that those individuals who have a low slope or marginal (dis)utility for gains (losses) also require a larger amount of gain (loss) to induce them to take an unfair risk (insurance scheme). For individuals with a steep slope or high marginal (dis) utility, the size of gain (loss) necessary to induce them to accept an unfair situation is relatively small. By using these estinates of the size 0; V I to induce acceptance of unfair odds for those individuals who were con- sistent on both gains and losses, the correlation between the size of gain and the size of the loss was determined. Finding that the corre- lation was significant at the 10 per cent level, an equation was fitted to show the size of gain as a function of the size of the loss. The equation is X1 = 26.33809 + .28205 X2 where X1 = size of gain necessary to induce acceptance of an unfair risk X2 = size of loss necessary to induce acceptance of an unfair insurance scheme.1 The equation shows that the size of gain necessary to induce accept- U) ance of an unfair risk is at most 26 times as large as the size of lo 3 necessary to induce acceptance of an unfair insurance scheme. Reliability of Predictions The reliability of the predictions fr m the fitted utility 0 "ve could be tested by correlating it with actual behavior. The type of _J 1The coefficient on X2 is significant at the 5 per cent level. 1‘42 behavior most useful in this regar(i would be behavior in actual situ- ations in which the odds and the expected return were known. Un1ortunately the total schedule on which these questions concerning gains and loses s appeared did not contain questions on t1is kind of managerial behavior. Furthermore, in agricultural sciences, little is known about the expected returns and the odds involved in various farm enterprises. D However, an indication 01 the reliability of the prediction from the fitted curves is available for the loss questions. There were 13 questions concerning informal insurance schemes on the total schedule of which seven appear to be useful for this purpose.1 These ar (0 l) was the ere any time in the last year when you kept on hand a reserve of cash or thin s easily converted to cash, like wheat, bonds, and live- stock, in case Of unfavorable developments? 2) Na 5 there any time in the las tyear when you paid more for an item from a person you could trust, than you would have had to pay for the same item from a less reliable person? 3) Do you keep more tractor or horsepower on hand than is neCc ssary for average weather in order to handle the crop in case of poor we athe1 I; ') was there any time in the last year wlien you added c1 ops -nd live- stock enterprises for the main purpose oi get tting you1 eggs in more baskets? 5) Did you re1rain irom borrowing so as to have property to mortgage in case of trouele? 6) Was there any time in the last year when you re sed to use your money for an‘ apparently pr01itable purpose in ore er uto 'play it safe‘? 7) Was there any time in the la st year when you didn't close what appeared to be a profitable de l because the person you were dealing with might not be reliable? lsix of He questions were eliminated because they either did not apply to all respondents or U ey were mor eformal insurance schemes that could be ar'ued d-epended as much upon salesmanship as on rarginal disutility of losses for acceptance. ‘ ..tll ». l.rl. —& ‘1‘... 143 There is no estimate of the amounts of money or the odds that might be involved in these insurance schemes which would make it possible to show an exact correlation. However, if the seven items are ranked using the proportions of individuals who said'Tes'to the questions in each derivative group as an indication of the groups ranking of the Specified act, then the ranks can be compared between groups. The rank— ing of the seven items by the four derivative types on losses is shown in Table 57. )EiAi‘IiiIlJG OF scum 11:81); woe SCHMS BY row DELIVATIVS e-LOUPS 01-1 Losses Type Rank 3-0 1 2 3 h 5 6 7 030 l 6 S h 2 3 7 D“o 2 I; 'A 3 6 l 5 7 F30 l 5 h 3 2 7 The numbers in the first row of the table correspond to the numbers on the items in the list above and represent the ranking of these items by individuals in Group BBC. Three of the groups agree on the ranking of items, 1, h, and 7; however, for the other items as the marginal disutility of losses increase the ranking cnanges according to a certain pattern. This pattern can be represented by the permutation on the four Q. numbers (2635). The permutation is read: 2 goes into 6, 0 goes into 3, lid; 3 goes into S, and 5 goes into 2. 'With some variation in Groups 030 and D30, this permutation describes the change in the ranking of the seven items from Groups 830 to F30. notice that items 2 and 3 differ from items 5 and 6 in that the latter two items have a direct reference to the possibility of a loss of money. The group that would accept an unfair insurance scheme ayainst relativeir small less ranks items 5 and O 6 higher than items 2 and 3; while the pyosite is true for the 5 that requires a large loss before accepting an unfair insurance scheme. Perhaps the most that this comparison indicates is that the tech- nique used in this study to quantify utility distinguished groups of individuals who act differently in other loss situations. But even this much of an indication as to the reliability of the technique is sufficient to warrant further research with it. *1 _L ’ evaluation W This chapter has presented some data which were intended to ascertain whether or not the utility measuring technique employed provides D meaningful estimates of cardinal utility. The following statements of its effectiveness seem Justified. l) The technique vrovides some estimates of cardinal utility capable of distinguishing individuals on the basis of meaningful managerial beh vior. This behavior in respect to gains and/or losses is: (1) amount of debt, (2) type of farm, (3) net worth, (h) income and (5) concern for two ‘types of error. l! i 2) The technique does not provide estimates of cardinal utility 5’) for individuals who display no indifference points in their answers. The technique and computational methods used allow individuals ‘ to be consistent with only one 1ypothesized'utility function. The technique allows individuals to be consistent with only the gain, only the loss, or both portions of the hypothe— sized utility function. The technique did not provide its own reliability test, in that, additional gain and loss situations similar to those on the schedule were not used to test the predictions of the set of questions used in this study. c .....III.IHVIIII.JJ CHAPTEd VII SUHhAAY AND COHCLUSIUKS This chapter will review the hypotheses tested, the procedures followed and those results pertinent to the testing of the hypotheses. It will then present some conclusions which would be useful to subse- quent research in this area. In the first chapter of this thesis, man was conceived of as an animal possessing a free will and motivated by his desire for utility and naturally constituted so as to maximize this quantity. If utility is a measurable quantity there must be some means for its measurement. The first hypothesis of the thesis is that such a technique exists. The second hypothesis is that a correSpondence can be discovered relat- ing past and present characteristics of individuals to future managerial behavior via estimates of numerical utility. The procedure followed was to derive utility functions, i.e., a relationship between numerical utilities and monetary gains and losses which provided the data for testing the hypotheses. This latter inform- ation was produced by asking individuals whether or not they would accept at varying costs certain odds for gains or against losses. This schedule of questions and additional questions necessary to provide related data were asked of 529 farm managers in seven midwestern states. Only single family managerial units for farms producing over 2500 dollars gross income were interviewed. The stratified random sample was designed by the Iowa State College Statistical Laboratory. 1 / 14:.) '-._o‘- -‘—‘. -- A- . }__J :7 H .A two week school provided uniform training for the 3 interviewers. The actual numerical utility computations are based upon two Opera- tional prerequisites, (l) preference among alternative utilities and (2) indifference between a certain and an uncertain alternative (objects with probability distributions). The first phase-in analyzing the data was conducted to determine the effectiveness of the schedule and interviewers in eliciting answers. The analysis had two steps. The first and most important step was to Specify which responses were and which were not consistent with the hypothesis of measurable utility and the hypothesized utility function. The next step was to consider certain attributes and types of behavior in order to ascertain possible reasons for the different answers and degrees of consistency. t was found that relatively older people with more years of farming experience, fewer dependents,relatively high net worths, and small amounts of debts were less likely to answer the questions concerning gains or losses. Further, it was found that certain Indiana farm managers were probably misrepresented by answers on the schedules. This misrepresentation was traced to interviewer bias. In this case it was concluded thattflmxuadata should be eliminated from the subsequent analysis. One of the most outstanding reasons for the different proportions of consistent and inconsistent answers was that different interviewers had conducted the interview. It was concluded that some useful observa- tions were lost in the case of the inconsistent answers. Data added to lh8 the consistent group of answers by interviewer ias were concluded to be useful for the fitting of utility functions. The relationsnips which were discovered between estimates of numerical utility and behavior substantiate this conclusion. As more individuals answered the loss questions than the gain questions it can be concluded that the gain questions were more diffi- ‘V cult to handle. Several reasons could oe offered to explain this difficulty. One might be that the gain questions which followed the loss questions sounded similar to the loss questions and that reSpond- ents refused to answer on the ground that they "had already answered those." Another explanation is that the gain questions were worded on the schedule in such a manner that the respondents did not understand the situations as well as they did in the loss situations. A somewhat more likely explanation is that more individuals are familiar with insurance taking than they are with chance taking for property gains. Still another reason which reflects upon some of the theoretical bases of the study is that although the theory denies neither risk aversion nor ethical objections to gambling, these two factors may help explain some of the apparent difficulties encountered by the interviewers. An additional motivating factor in contrast to the utility of wealth assumed by this study is the utility attached to non-monetary gains and losses. The second phase of analysis was directed toward finding empirical relationships between the estimates of numerical utility and the character, disposition, and bekavior of farm managers. This analysis not only contributes evidence for the existence of a technique for quantifying utility, but also to the hypothesis concerning the actual correspondence between numerical utilities and predictable manaderial behavior. The predictability of managerial behavior is based upon the premise, "that men act as they have acted."1 This phase, the second one in the analysis has three steps. The first step was to divide the total group of consistent answers into types based upon the estimates of relative marginal utility J.‘ differentiated off the individual's total utility curve for wealth. This classification was based on the belief that individuals who attach the same or approximately the same relative marginal utility to a dollar of wealth at different gain or loss levels will have similar characteristics and behavior. The technique does not provide a basis for making comparisons of an absolute numerical utility. a basis for making comparisons of relative utility the unit of measure and position of origin. Thus, 3 i o e c , relative to he second step was to compare the attributes and behavior of the individuals in each type. The type of individual who answered all Yes to the losses and the gains made up a sizable group witE rather distinguishing characteris- tics. The; T are tre oldest individuals of those interviewed and have the most farming experience and the fewest dependents. On the average they have a high net worth and low debt positions relative to the other _J types. Their other behavior is commensurate with 1This is contrary to the premise "men act as which would have lead to a study of ethics rather this type of they should act" than one of behavior. It does provide “2! 7D '51:! Wampum-4;. w. n- 1" a- v- 1:0 ,1 individual; however, the fact that they accepted all the unfair insurance schemes and the unfair risk situations is surprising. This fact may reflect more heavily upon the technique than upon the disposition of the individual. Contrary to the conclusion that they attach an extremely high marginal utility to wealth may be their feel- ing that it made no difference to them what they answered. Thus, the technique and interviewing procedures may be ineffective with this type of individual. The second type of individual,which did not appear in significant numbers on the losses questions'was one who would not accept a fair bet on gains. Like the type just described, these individuals have a high net worth but, by way of contrast, have the highest debt position of any of the consistent groups. This result suggests that these individuals may have a positive preference for certain odds aside from either the probabilities involved or the utility of the gain. A hypothe- sis to this effect could be tested by a more extensive schedule of more-than-fair odds and fair odds than was included in the I.M.S. technique. The fact that for the individuals who were consistent on both gain and losses the size of gain necessary to induce acceptance of an unfair risk was at most 26 times as large as the size of loss necessary to induce acceptance of an insurance scheme substantiates the contention that the range of gains needs to be more extensive. In those cases where an estimate of utility was derived from a ufitted equation, two of the most meaningful variables found related to marginal utility for gains was the amount of debt and the type of ljl farming engaged in by the respondent. These results indicate that the eehnique and the ,rocedures followed does provide, for at least some individuals, numerical utility estimates useful in predicting mana erial behavior. It could be hypothesized that the reason a significant relationsiip was not found between net worth, income, and marginal utility was that the respondents' estimate of these two items was less accurate than their estimate of debts. This hypothesis suggests a more extensive interviewing procedure to obtain more accurate measures of these factors than was used for the I.H.S. Another factor, which may be related to answers on the gains and loss questions, is recent changes in individuals income, net worth and debt position. Unfortunately the I.M.S. schedule did not include this information. The need for such information is emphasized here as a necessity for future research in this area. For the fitted utility functions on losses, the four most meaningful variables (from the standpoint of establishing a relationship between marginal utility and behavior) were (1) net worth, (2) income, (3) type of farming, and (h) concern for the two types of error. These results indi- cate that the technique does provide meaningful estimates of marginal dis— utility for losses. Previous comparisons showed that more individuals answered and were consistent on the loss questions than on the gain questions. A third step in the analysis attempted to establish the reliability of the technique in predicting behavior. Although the technique did not include its own reliability check, one part of the total I.M.S. schedule involved informal insurance schemes. A ranking of these }._J \j‘ m' schemes showed that the various types formed on the basis of answers to the loss questions ranked the items di' ierently. The pattern mani- fested between thet 'pes warrants a hopeful conclusion that the tec dinique at least for the loss questions, possessed some reliability Ar eliabilityt st on U1e precictions would consist of' a s cond set of gain and loss situations involving di ffm ent amounts of money and different probabilities than appear on the schedule from which the utility function is estimated. Tia LC predicted answers to this second set could be statistically compared with the observed answers to lam-flu” “Av“... “.17- ascertain the con nfidence to be attached to the predictions. A strm “er way of testing the reliability of the technique would be to predict what an individual would do in specific uncertain circumstances in his particular setting. Of course, to me e this reasib ethe reS3archer would have to have the re turns and the probabilities of the outcomes involved in the enteiovLs. This would be a more expensive procedure than the first suob38ted however, it would probably be more meaningful than the hypothetical approach. The following sugges tions grow out of general experience with this study and specific analysis presented herein. 1) Further research in utility measurement is warranted on the same basis that research with other measurement tee} uniques (I. Q. tests, and personality tests) are warranted. 2) Supplemental theoretical work is needed in decisions making theory of farm management similar in character to that of mathematical statistics, i.e., greater emphasis needs to be placed upon quantitative measur‘mwNUQ and more precise definitions of the relevant variables. 3) Examination of some of the basic assumptions of utility theory in reSpect to their compatibility with the fundamental suppositions of science needs to be performed, e.g., the teleological nature of the (D theory maybe in conflict with the mechanistic approach of scienc . 21) Further effort should be given to technique construction and interviewing procedures, particularly with respect to situations involv- xtensified to include not only (D ing gains. The technique should be other situations for the various odds, but additional reliability tests. 5) When utility measurement techniques are employed they should not be a small part of a larger schedule. Information which is supplemental to the utility schedule should be or'entated to testing specific hypothe- ses and obtained in the most accurate manner available. 6) The results of this study indicate that a sample stratified on other than a geographical variable could exclude individuals from whom it is difficult to get answers on the particular questions used in the study and from whom answers would be meaningless. Another variable is the utility attached to objects other than wealth; such a variable was implied in this study by certain individuals who were distinguished by age, net worth, type of farm and debt position. Perhaps a different type of utility measuring technique could be designed to include the individuals with whom the ques'ions were unsuccessful in this study. 7) Further consideration should be given to methods of analyzing the data. Such as: 8.) Other criterion for Specifying consistent and inconsistent answers, 8.5., classifying on the basis of risk aversion as indicated by ans* ers to certain odds. 15h b)Dnermine the effect of changing the assumption of cmmtant utility over the range of the payment (stakes). c)Cm5ider the cases which rare inconsistent with the hmiiference axiom for consistency with a subjective pnbability hypothesis. tO "ul— vb 8)Eheimflications of the present study for farm management ingandemtmudon need to be considered and made available in relevant publications. IV‘RLEJSCHAJ Alchian, A. A. "The Meaning of Utility Keasurement," American Economic Review, h} (march, 1953). Bentham, Jeremy. In ccouuction to tie Principles of Eorals and __egislation. Oiiord: Clarendon Piess, 1(0). Braford, L. A. and Johnson, G; L. rarzn .ana--.ent Ana’ym ew York: John Wiley and Sons, 1953. Davidson, D. and Suppes, P. ”"periments in heasurenent oi Utility by Use of a Linear Pro'rammini local. Technical Report I40. j, Uiiice oi Naval Resea a1c11 (April 2, ljpo). 9 Edwards, W. "'tperimental on Economic Decision naRln" in halelné Situations," Seminar on the Application 0: Lathe cmatics to he Social Sciences. Unive isity oi Richigan, November, 1532. (nimeographed.) Fisher, I. l oievatical Investi ations in the T-eory oi' Value and Price IN New Haven: ale Unive1sity Press, 1,)0. ring "1 larginal Utility" anc Income Tax. In .conom c ‘ Fisher, I. A Statistical Method for m asu Testing the Justice of a Proiressi ve LESsa"s oornnn_n too iriiunuor of John Ikit as Clark. Leew lork: “...—8‘ m... “.MI-u—o-w- .————~— r7 Liacmillan Co., l/¢(. Friedman, M. and Savace, L. J. "Tie Expe ected Utility EIypothesis and the Measu1aoility of Utility," Jouinal Oi Political Economy. 00 (December, 1952). Friedman, h. and Savage, L. J. "The Ttility Analysis of Choices Involving Risk," Journal of Political Economy, 50 (August lynb). Friedman, h. and Sa avage, L. J. "The Utility of Choices Involving Ris L," Read; n; s in Pr11ce Theory, ed. G. J. Stigler and K. E. Boulding. homewood, lll.: Ricaard Irwin, Inc., 1952. Great Plains Council. Pioc3367n~s of nesea rch Conference on Risk and Uncertainty. Great Plains Council Publication No. ll. Lar;o, N.D.: North Dakota Agricultural College, 1955. Haver, C. B. "PrOgress and Problems in Decision haking Studies; The Universe of Farms Studied," Journal of Farm Economics, Proceedings No. 5 (December 1955). Jensen, H. "Progress and Problems in Decision LaIing; Studies, T113 Nature of the Study," Journal of Farm Economics, Proceedings No. 5 (Decerrioer 1395:). Jevons, W. S. Eheuiheory of Political Econogy. Uth ed. London: Macmillan and Co., 1311. Johnson, G. L., §§_§l. Unpublished Wicpo fl( ove11oer la-17), Chicago: Farm Foundation Office, '9:3. Johnson, G. L., 2: a1. Unpublished R port, C arch 23- 2a), Chicago: Farm Foundation Office, 1951. Johnson, G. L. "Learning Process ms 1h JIndividual Approach,‘ P1ocoedircs oi Resear on Conierence on Risk ano Uncertainty in —.-O--——-‘—.. ._ v ‘ A"“icultur1. 1ar c, L. D.: Hort11DaI;ota Agricultuial Experizuent Johnson, G. L. Eanagerial Concepts for Agriculturalists. Bulletin 613, Lexinpton: University o1 Lentucgy, lypa. Johnson, G. L. "More Ado About Sarles Suppositicns Regarding the Inter- state Lanagerial Study," Journal of Para Econorics, 3J (Lay 1930). Johnson, G. L. "Ne eded Developxn ents in Economic Theory," Journal of Farm Economics, 32 (November 1990). Johnson, G. L. "Pr0"ress and Problems in Decision flaking Studies; The Irieohan-oaVa"e Utility nyp0t1cois in the Interstate Ranagerial Study," Journal of Farm Economics, Proceedings No. 5 (December 19> anagement. Kentucay Bulletin 393. exington: University 5 Kentucky, 1933 Johnson, G. L. and Raver, C. D. Decision Raking Princivles in Farm Johnson, G. L and. Sn1ith, Joel. "AP1ejoinder," Journal of Farm Economics, 3U (Ieoruary 193 o). Marshall, Alfred. Principles of Economics, héh ed. London: Macmillan 00., 1947. henger, Carl. Principles of Econonics, trans. and edit. J. Dingwall and B. W. Hoselitz. Glencoe , 111.: Free Press, 1930. 1 F Neumann, J. von, and Rorjenstern. Theory of Games and Economic behavior. Princeton: Princeton UniverSity Press, iya(. .‘ Sarle, C. F. "Comment on the Eejoinder," Journal of harm Economics, ' (February laps). kl) 157 I Safikh C.F. "Eesearcn on the Dynamics of he F 1 Prooxmes" Journal of Farm Economics 38 Tebi , 9 _* $13 ...) ...: p—I “ana'erial Decision my 19:6) &nth,gbel. "Progress and Problems in Decision Studies; Some Problems afliemxm.in the Interstate Managerial Study," Journal of Farm Doormnk2, Proceedings 10. 5 (December 1955). Stbfler (L S. "The Develonment of Utilitv Theorv I and II " Journal L.) 3 J I a , ofififlitica l iconon., S7 (Octooer 1910). iimpes,fh muifinriel, W. "An Axicmatization.01 Utility Based on t21e Rotunlof Utility Difierences," hQHSTQHQUt Science, I (Ap1il- July, 193;). Suppes, P. T1e 1oleo fS“ojecti we P101~01'iir' arxiljt ilitf in D:_;§EEE; Mflchrg Te conical Acport 1o. 5, Oiiice 01' haval Resear01 (Jm ne 1, 955 Thrall,11.1h, Combs, C. H. and Davis, R. L. Decision Processes. New York: Jol1n Wiley and Sons, IQQJ. .., . . .’ Walras, Leon. ”Inverts of Pure boononics, trans. William Jafie. m“.- Homewood, Ill.: iicnard D. I1win, Inc., 1920 d‘.raliilvv1 .» MW 1 1:; .1.) LI I 'I The scheoule used in the I.M.S. survey is presented nere in toto. The particular questions which have been the main concern of this 913515 appear on pages 159. 15-0, and. 182. 1. .‘c . uni ...Iurfl. Fltldlfilfr... Elfin 1.57.921 ”10* Interview uhml““ 021.3611 U191" LIST F01 QUJL'STI CH 4,7 Nnre is a group of similar situations. Please fill in 3our ans.ers to show whetharcn'not you'd be killing to pay tn se costs to get out of g1oups in which,one person has to bear a loss. No. of people in group 113900 No. of people in group ”_g.OOO Amount 01 loss 1F£;,UCO Anount 01 loss L;;Q,UQE cost of gaming out ofb 1 2; Cost 01 getting out of i‘ 2; Yes No 1 Ye s No No. of people in group NOQ No. of people in group 20 Amount of loss 410,600 Amount of loss ' 1 :00 Cost of getting out of _. “0 Cost of getting out of 10 Yes No Yes No No. of people in group DC No. of people in group LOO Amount of loss i1 l,otU Amountnof loss l'l”,oog Cost 01 getting out of q 10 Lost 01 getting out of 1; 2; Yes No Yes No No. of people in group 20 No. of people in group DOC Amount of loss , eco Amount of loss :ilo,UtO Cost of getting out of g 10 Cost of getting out of 1 lg Yes No Yes No No. of people 1n group ('2,DOQ No. 01 people in group .L Amount 01 loss 'ggu,ooo Amount of loss . n 1 log Cost of fit ting out of g 10 Cost of ettinr out 01 . 19 Yes No Yes No No. of people in group 20 No. of people in group NO W1 _ . w \r\ Amount of loss Q poo Amount of loss a 1,0 0 _ . (TT—"""";‘."' ,1 __I,_ . Z ,1 Cost of getting out of g 4; Cost 01 getting out of a 2) Yes No Yes No No. of people in group , 2,0D9 No. of people in group [“1230 9 Amount of loss L1U2UUD Amount of oss ,a,, LQ Cost of gett1ng out of a DO Cost of getting out of 1 10 3:93 NO 1'33 NO " . - 1 v n _ . rm " ‘- no. of people in groap N No. 01 people in b1oup ( l:2 0Q ... D E} \_I” ,1 . .3- a 1. r‘ w a)». Amount 01 loss ‘ I. ltp .Aunuh.01 loss 1 p 14:1213 Cost of getting out 01 a lo cost of "Outing out 01 i no Yes No "es No . a V ’1. ’3 . " .No. of‘gieo ole 1n group _hg No. 01 people in group .. -3 Amunt oi loss l 0C ‘ Amount 01 loss 4. 1o ) F. f + I (I _._—_._” .. Cost of gettinr-out 01 . 1n) Cost of geotuugout of a a; Yes No Yes” No 160 Interview Number (‘LJ'-1":‘L v.1."1u um OFF LI'T E01 QLZSTION 58 Here is another group of situations that are similar to this one. Please filJ.jui;your answer to show whether or not you'd be willing to pay these o<3sts t;o get into a group in which one person would get the gain. 1N3. of people group lg: No. of people in group 200 ‘Value of prOperty gained 2 1 OOO Value of property gained Eéiifiiil .Amount.3nm1 pay to get in i 25 Amount you pay to get in N 25 Yes No Yes No iNo. of people in group 12000 No. of people in group 20 ‘Value of property gained 1‘52000 Value of preperty gained' 3 pug .Amount you pay to get in N NO Amount you pay to get in ; 10 Yes No Yes No '— No. of poo le in erou 22000 No. of eo lei grou 22 ‘ Value CY pgopertyogaiged L7“ GOO Value 0% piop? rtyL aiied 3:0 Egg o¢_1___ :__J_... Amount you pay to get in i 10 Amount you pay to gget in E 25 Yes No Yes No __ No. of people in group 20 No. of people in group 20 Value of property gained :_* 500 Value of property gained N ECO Amount you pay to get in i; 25 Amount you pay to get in N NO Yes No Yes No No. of people in group NO No. of people in group ____ 2N) Value of preperty gained N leOO Value of property gained N izooo 1 Amount you pay to get in :1 NO Ahount you pay to get in t 10 Yes No Yes No No. of people in group 1 000 No. of people in group . leCQ Value of preperty gained N2 2000 Value of property gained QZSzOOO Amount you pay to get in N 23 Amount you pay to get in ___ 10 Yes____ No Yes No No. of people in group 22000 No. of peeple in group N00 Value of property gained ;,0 OOO Value of property gained NIOICOQ Amount you pay to get in N NO Amount you pay to get in Q 25 Yes____ No - Yes No No. of people group NOC No. of people in group 200 Value of preperty gained llOICOO Value of property gained 2 5 COO Amount you pay to get in. i~___;§2 Amount you pay to get in § 10 Yes____ 3k)" Yesflnw No No. of peOple in grou_ ”:00 No. Q? Torple in group ’___2¥§i ValuecfiTproperty gained , lo 000 Value of property gained f;_S.OOO Amount;wn1pay to get in No Amount you pay to get in N NO Yes No .A Yes No M Intervi: w luinoer Date Income Qualijications Checued Managerial Qualifications Checked INTJNSEATE FAJN NANAGE NIAL SUNVNY 1. Now first of all, how many acres, all together: do your own? are you renting this year? (111“ "AZY“) This year how many of (IF "ANY”) How many of these the se are you actually using as: are you actually using as: crop land and rotation pasture crop land and rotation perxanent pasture pasture ___y__rent out or put out on snares perma ent pasture re man inder remainder 2. What do you consider to be the main crOp or livestock product on your farm? What did you do with it last year? What otler crops or products did you market last year? _‘__ (IF NONE THAN ONE CROP AND/CR PROD'CT WAS NALN‘TJD IN T"E *“HCnUlhu 'NAJ.) NLat preportion of your last year's total iarm income did eac h of t1ese frif! account for? (11s: UNTIL yo; or INCONZ IS Accounrar r01.) Nain product __> 1, -_--__~- ..-J‘.‘45 2nd product .1... - 5 3rd product ¥_‘ __ _‘ ___,__}5 Nth product _._ .11_11 l_, __._~__j3 5th product g__ .....1_5 6th product _ .‘¢*- ____1_Q$ 7th produc _‘ _‘ ' “ $ :3 8th product '__~___ 9th product 162 10th product ”I ,‘O 3. Now I'd like to ask you some questions about the kinds of information hat a farmer needs. What should a farmer find out before setting up a farm in a strange area for a strange family? —- Q.“ ———'—‘—— N. In general what kinds of information do you think a farmer ought to keep up with in order to operate a going farm business. ' a. In order to get the greatest profit? b. In order to ._ ‘ fl-“ get the greatest satisfaction for his entire family? — m w-..” ‘— _‘ ————— -‘—“ ‘_‘—‘ --——-- # _- ~——-. —‘ ---- ”...—un- _ !— “_._—o- ‘ *fi 5. Here is other you may have had to obtain in ore thin; s that have CON: up in tNe course Each t; pe is completely Clear, I'll try to help you (PNESENT LIST. PAUSE FOR N PLNJJTm Tc 1. PNICES: paid for and future prices. L1 nation on prices rece I‘”or T‘I\r -11; ANPl SS: Current11a1Ne Narhet outlook Corn-ho; ratio Dairy—feed ratio t prices p CTICN IAOTC S: is practic: s and items used in pr livestock production--also inior*"* LaloQU weather ariect yields. P11013111. "I in 1 1.3: v LANEWL- :‘ertilizers Sprays and insects CrOp varieties Feeding ELL“ "‘5 3. NEW D3 TZIODEVVmS Infor mati lar In practices and iten1s used FY; PT", ES: Suppl mental irrigation Antibiotics Anhydrous ammonia Chemice l weed };illeIs 1 \ NUEAN EACTOZS: with or consider in making EXANPLES: Family‘fiembers Relatives i\ 31 1111001‘8 or Other people friends a list of five types of information which enplained on ti LiS li st and ii‘ items used in i‘arN production this include 16 at one time or er to rwal decisions about 01 your farming career. explanation is not I bib-3 NIL,“ it]. \-~ v ""}CC. s, ’1 1131111 ALT-D 11111111 111 1.1.1.1.... 1.1.) s and prices t, pres-ant, ived for farm pi MC Mu cts 8 pa as Feed and supply ‘Drice Nachinery rices Wage rate Interest ates Information on the effects of all accepted cduction on rates 01 crop and ion on how soils, isease and Storage Nor}: met ‘1 I lla; Buildinx layout on on new developments or changes in in production. Neat-type hogs New feed supplements Self feeding silos Krilium Information about individuals you may have to deal decisions about a farm. Dealers and buyers Salesmen County agents imr ed WOTNGTS 1...: ‘9‘ «_.1 POLITICAL SGCIAI N3L”"7(U" bA3“(. : Inf011ation on local national and 1r"c1nat10nal governments and iormal and informal groups wnose actions affect a farm. ..... ENA r11“: Acreage‘controls Church practices , Tax rates Conservation pregrams Draft Drainage districts School districts Co-op policies a. In tNe light of your own experience in getting information to set-up and run your farm to get the most out 01 liIe , which of these five t; es of information have you found to be most im- portant to you? (ION'A INLEAVINWZAS WILL SUBSTITUTE "FOR PAUFIT" Ir PLACE ‘? "TO GET THE NOST our 01 LIF‘. ") Rank 1 Which of the remaining four do you think has been most important to you? Bang 2 ‘—‘ fi‘ Whicn of the five has be en least iiportant? dank 5 __ row of the remaining two-(11 SENT I.’ANES 01 NENAINING Tao TEFL s) wnicn do you think you've found more important in solving your farm problems? flank 3 Rank 1 _g (zaNAIJI NCAPEGC: V) All equally important Can‘t rank: Why do you feel that you can't rank them? “*0. b. The kinds of information you find important may not be equally difficult to get hold of. In the light of your own experience in getting information, then, which of these types has been the most difficult to get? Rank 1 \O Whmfilof the remaining four has been most difficult to get? Rank 2 _._. lfijch of the five would you say that you've found least difficult? j." ‘. " -Lclnfl : How of the remaining two-(lHSZiW LAIES OF RUHAINIHG THO TYPES) which do you think that you'v311"ound most difficult to get? RAKE 3 ... _W‘“ M mm: a _ _ Rewrite cause-x) __ All equally difficult Can't rank; Why do you ieel that you can' t ra ’ them? We've been talking about information needs that you may have had in making decisions about Specific problems. However, tixere are a n lung 01 other oiiiiculti s inyolv ed in maling decisions anc acquir in; information that you may also iind to oe proolems. Here is a list of some of them. (HAND CAAD TL HSSPUNDEIIT) I'd like you to tell me which of these or any other not on this list have been problems in your own experience. 1. Knowin‘ when to Chang eyour production plans. 2. Recognizing the existence of proolems. 3. Dei inin.rr the oogectiv es of your iamily. h. Knowing when you are on the "wrong track" in your attempt to reach a desired goal. V "Putting your i‘inger" on the diiiiculty wzien you l-Lnow there )0 is something wrong r whe; you know a proolem exists. 6. Just keeping up with all of the new information relating to farming that constantly comes along. '?. Getting information organized in your own mind so that you can see what it means tor you. 8. tnowing how and when to arrive at decisions (once you've eacls you to one orfianized the information) when some of it 1 conclusion and some to another. 15:3 9. Any others not on this list. 10. In figuring out what action to take on the basis of the information you have about a problem, do you sometimes look at what it will cost you and compare tais, both financially and otherwise, with the results you can expect? ' No: Why is it that you don't do this? Yes: Do you ever try to work out'the answers in W“itihg? No Yes Can you tell me some of the things you've done this for? r I 0 Do you sometimes do this figuring in your head No Yes Can you tell me some of the things you‘ve done this for? *— 1537 Here is a way for a farmer to figure out the costs and returns of expanding 3 lb litter hog enterprise to 25 litters. a. The farmer figures that his costs per litter will increase from {210 to‘ 22 ‘With the price of hegs remaining as at present, he will gross: 270 per litter. On this basis, if 113 expands his hog enterprise to 25 litters, his ne t proii t per litter will be ¢uo, or the diiLerence betwee. the €2f0 and fi'22' . would these figures tell you how many litters this farmer should raise? Yes: How many litters should he raise? Ho Don‘t know b. Here is anoth3r way to figure out t1‘.e same problem. He figures his costs and returns on e h additional litter and ac finds that each one will add or lose the efollowing amounts after costs are subtracted. Profit Loss Profit Loss 16th litter 380 — 2lst litter 31h - 11th litter $72 - 22nd litter : 7 - 18th litter $59 —' 23rd litter - $1 19th litter $h§ - 2hth litter - filS 20th litter :30 - 23th litter - :;20 would these Mi ITGS tell you how many litters this farmer SiOUld raise? Yes: How many litters should he raise? i No ! Don't know c. Which way do you figure out costs and returns in similar situations? Uses 3. Uses b. Uses both Uses another method: How would you figure it out? ‘__ 12. a. Here is the information that a farmer has for deciding whether f not to put another $250 into machiner*. (IHTEJVIENEi PRESENT CAJD) His records indicate that his average gross income per 0 L250 invested in machinery is Ehfio. The average returns above fuel and labor costs per $250 invested in machinery are @275. Is this enough information to decide whether or not a farmer should invest another f2LO in .achinery? r‘- as: For what reasons? v ho: Why not? Don't know: What difficulties are you having in figuring this out? here is another way for him to figure it out. (ILTLAVIENEm Passer: CAfiD) An analysis of records from his farm and other similar farms indicates that additional investments in machinery can be eXpected to return 255 on the dollar after the earnings of all other expeni- tures and investments are accounted for. This 2;p includes profits, in 3rcst on the machinery investment figured at be, and depreciation x.’ .L V V I figured at 10%. Is this enou3n information to decide whether or not a farmer should invest another 9250 in machinery? Yes: For what reasons? W 1k): ‘Why not? _A ‘égil _.._____.__- 169 ' I". n Don't know: hm t 3111iculties are you having in figuring this 9 13. Two methods of arrivin3 at concllsions are illustrate d by t13 examples on this card (IJLJlVL N33 Plush“ CARO) 1. In some cases we draw conclusions from experience. Thus, we may notice that in certain situations certain results always seem to follow. On the basis of this, we conclude that these results always occur in this situation. An example mi 3Lht occur in ferti- lizing a field. Tilus, if a farmer sees that the poor thin spots in a field respond to fertilize rs more than the rich Spots, he may conclude that poor thin Spots always respond more than rich spots. In other cases, we "reason out" conclusions about new situations facing us from facts and principles we know or assume to be true. For instance, a farmer may know or assume that a certain barn “rangement will 5 ve labor and then "fi3ure out" how the use of this arrangement would affect the amount of labor which would be left over from use elsewhere in his business. a. Do you use ooth, mainly one, only one, or neitler of these methods in arriving at conclusions Both Iainly one Which? OIL y on.e Linich? —-a 'BiLHSF omtrmw C" H!!! 0. Which of these thinking methods is most natural for you to use? Both One: Which? Neither Don't know HH you use one of these methods wit7r iout usin3c' the other? 0 o O 5 a Yes No Don't know H! 170 d. "hat proporti n of your thinking is like the first met :oc? (F 1-13 .5} .. 'IjlluilST) Hone *A:)out 1/2 All Lesst nan l/h etween l/2 and 3/n wDon't know how About l/h ”About 3/1 wmuch, but not all Be‘mq 3 3n l/2 and l/2~ mlore than 3/h ho anszer e. M at proportion of your thinking is like the second method? )4“ “‘{1 'r‘\‘ “(71) (R SLIJp'ui -i—JUAJ—JIQ _._-hone ___About 1/2 All ___less than l/h ___Between l/2 and 3/h —"—bon't know how mAbout l/L, _About 3/4 “much, but not 311 ____Betw; en l/2 and l/2M Whore than B/Q ._‘_Eo answer f. Could you give me anothe example of the first method of arriving at conclusions? I g. Could you give me another example of the second method? _A ‘4‘ . *m- 15. In deciding whether or not to buy a piece of land, a farmer can mafe either of two kinds of mista]:_es. He can buy land wlien he s:1ould not have. This mistake was made by many farmers aftew World Her I. On the other hand, he can make the mistake of not buyin3 land when he should have. This mistake was made by many farmers who did not buy land between 1935 and IBLLS. In making fa 3rm decisions, are you more concerned about taL:in3 action when it would have oeen better not to than you are about not taking actions when you should have, or are you equally concerned about both of these? Lore concerned aoout taking actions when shouldn't *ore concerned about not taking actions when should Equally concerned Don‘t know fl 171 17. Could you please tell me how you made up your mind about what or how much of each product to produce tais year? —"“ w‘ —— i-‘.-.___ 18. a. What important thing that you buy and use in production has had a fairly bin change in price recently? (X) ...,J b. What do you use it for? c. How did you make up your mind about how much of to use a filo-.L in producing Y , when the price of X changed? A“ w“*.- - —- _.___ _“ .___ ‘A l . ..- l9. a. What important thing that you produce for sale has had a rather drastic change in price recently? (X) b. Did the price go up or down? 6. How did you make up your mind about what to do about your pro- duction of K as a result? “In. ~m.-~‘~-—- “u. * ___- __ d. What reasons did you have for coming to this conclusion? 20. WuN3vms the last major piece of machinery that you bought Howcfid you go aoout miiing up your mind to buy it? 21. In the last two years ha f8 you attended two or more County agent's or extension specialists meetings Yes No. Meetings of farm organizations like the Farm Bureau, the Grange, and the Farmers' Union Yes Ko 22. There are ways of 35 sources we've been talking and check the appropriate Spaces for ways you usuallyu these same kinds of information? 23-2h. would you please tvk for the sources you usually use to ptting some information without us sing any of the about. would you please' ta.:e this chart to get (inruJVlhuii EAPLAIN I’ADIL l5) e this chart and check the appropriate Spaces et these different kinds of information? (11M QIVIJJJ EXPLAIN HEADI}:GS) 25. a. What do you expect the price of (IN‘fiRT NAME OF HOST ISPOWTANT COLLODIL‘Y, BIL LUDING Alfirf PRODUCTS) to be at your next market- ing time? When would that be? _‘ {.3 Cf. price of (mag PE-LOL UCl‘ 131131.51) n: a. ) Do you expect the be higher than, lower than, or the same ma “ketinc time to were at the same time last year? 3.3 they Don't know Iii ill-31‘ Still, if you had to make a Lower Same prediction now, how would _~—*.fi you figure it out? How have you arrived at this _‘ estimate? 26. a. 0. (IF NO GENERAL NODflL IS GIVEN IN b., “N THE FOLLOWING THREE dance IN c .) In general, what circumstances lead you to expect that the prices you receive will be higher than they were in previous years? In general, what circumstances lead you to ex ect that the prices you receive will be the same as they were in previous years? 4—.— _‘__i ..- In general, what circumstances lead you to expect that the prices you receive will be lower than they were in previous years? —— AA “...-.... -‘ ———-— - d. Is there any special year or group of years that you think of as typical for purposes of comparison in trying to figure out what prices to eXpect? What reasons do you have for thinking of that period as typical? ‘we buy'many'things to operate our farms. Feed, fertilizer, and seed are just some examples. In deciding when to buy things, how do you.usually judge what prices are going to be? *..*u“ 171‘; o. What are some of the things that you buy from time to time that get used up in production? Under what conditions do you assume that the prices you will be paying :or (IhSiNT NAhj OF FIRST INPUT LELTIONQD AGOVN) will be h'xher than they were? U Under what conditions do you assume that the prices you will be paying for (I"SJNT IA.J F FIIST INPUT LSXTICI i3 ABOVE) will be the same as they were? Under what cc onditions do you as su1e that the pm rices you will be payinfij for (I13 511‘ Life or ;~I1S‘I' INPUT 1:5:I 3:1 1:13 were) will be lover than they were? “ ..— No farmer opirates his Iarm wit: Iout having some contact with other people. He comes into contact with suCI peOple as farm laborers a, men who do custom work, dealers, landlords s, bankers, and so on. Do you usually have some idea as to what to expect from a person you're about to m>e e ? (INTNNVISNJ_L CODE) Has some idea: Now can you tell what to expect from a person you've just met? 1.. ”lib! ....-E’1.‘ (IF ANSNEJ IHDICATES TLAT HE DJPBXDS OH IHFCNN- ATICN FNOM OTLJNS) If you didn‘t know anyone who could give you some information about the person, then how could you tell what to expect? -‘w M'.‘ ~‘- ~--_ waits and sees: Are people so different that a man has to know new acquaintances for a whi e efore he has some idea of what he can expect from them? Yes: Are there any hings you can looa for in a person to give clues as to what to expect? N Yes‘ U‘at are some of these tflln"°? What can we figure out from them? .1. I ' No: well, than, what can you expch Lrom people you‘ve just met? _—_._. o“..‘—« -w‘ What are some of the thinds that make it possible to know what to expect from strangers? It's hard to of depends: say What does it depend on? (IF hey-«“511 warm, FOL Don't know ‘..) forecast whethe IF "UINED hAh” QUESTION hOT AhSNSNED ADEQUATELY AID NSSPONDENT IS A LALD £919. In selecting a man to operate some of your land, how would you decide whether a man would make a good tenant? ”a“..- 28. Do you think there ____ _‘.._ LOWFUP TU ”DON'T KNOW") In selecting a revular hired man, how would you r he will make a good employee? (IF "hIiED hAN" QUESTION HOT AHSNQNED ADQQUAEELY AM) NelSPOLEDEL'I‘ IS A LA...) In looking for a man to rent from, how would you decide whether a land- owner would male a good landlord? ~-‘--_. will be any changes in national, state, or local government programs and policies for farmers in the n»xt wo years? No: What are your reasons for feeling this way? o O C‘ Yes: What are your reasons for feeling this way: Change and no change equally likely: feeling this way? What are your reasons for ‘-.MJ~‘ *— Don't know: ‘Well, then, do you try to take hese things into account in your planning? VVs: EIOIJ? cur—_.- ”...—.-.“... “.-.... .- 29. a. Do you think there will be any changes in farming methods and things used in farming during the next t . No: What reasons do you have for feeling this way? Yes: Tlat reasons do you have for feeling this wav? “*m—u“ w--.“ m _ h f o For what kinds of things do you anticipate these changes _ _ .. ‘H ——*' ‘ .‘ Don't know: Well, then do you try to take these possible 9 changes into account in your planning. —_—- M.-- - v-‘ b. Assuming, for a moment, that there will be changes in 1 methods and equipment, would you be willing to be the i' in your area to try out some of these changes, or would you prefer to have some other farmer try them out before you adopt them? Willing to be first Would prejer to wait for others Depends or don’t know: What would it depend on? __ ~G--—w*-—‘o-‘ w“ --—“~ -7 _/ _‘b~ I ,IJ. Could you have used more credit profitably last year? No Yes: Did you refrain from borrowing so a to nave properW to mortgage in case of trouble? (1) m d- .~a ... (a re any ti:Le in tL la st year whe n you a t close what appeared to oe a pro itab ole deal because the person you were dealin: with might not bel eLia)le? tbs tl;ere any time in the last year when you added crops and livestock enterprises ior tne main pu; pose of getting your eggs in more baskets? Yes No "r: J" 0‘18 v "'jro i 4'7 ‘1 731 t V391" .r’wu: v :x'p‘ an’ J- :2 :rs UH J. an“. M1 n p-14 a.“ Jr»... Enron JOU reiused Do use yOUI‘ i t money ior an appaICntly pro able purpose in rder to "play it sare?" H'D Mo ...—s Do you keep more tractor or horse power on hand than is necessary ior average weather in order to handle the crop in case of poor weather? Yes 1"- O was there any time in the last year wien you pa aid more for an item from a person you could trust, tlan you would have had to pay ior the same ite n irom a 1 ss ‘eliable pers son? Yes No I‘ Do you cal‘ry liie insurance: \T 140 Yes: Do you carry additional liie ml surance to cover a debt for your family? Yes No “.- {IT-In n.0,! :Q'."r1".'¢r‘ 3.. ,r' 24,0 . 1:- \IL‘. uh. LL). How about fire insurance? Do you carry any? Yes No Was there any time in the last year when you kept on hand a reserve J" or cask1 or thin s easily converted to casn, like wheat, bonds and livestock, in case oi uniavorable developments? nan necessary to be Do you ordinaril;f 1w eep la rger ieed r Vourself ag ainst lossc due to bad weat Do you make a practice of having available more hay or pasture ground than necessary in order to protect yourself against drought? Yes “ No fl Do you carry collision insurance to cover damajes to your car or truck? a 0 Could you please give me some examples of things Jfllcll y u Or your family did last year, when you were not completely sure of the out— come, but willing to take the consequences of acting and being wrong? ‘ ___‘ ‘ _‘i L _‘u— ...n.. —-——— .‘ _- .-. now we'd like examples of thing 5 wnich you or your family decided not to (10 last year even though you ran a risk oi be ing w1ong in not acting b'e want cases in which you were willing to take tne consequences Ol being wrong and not cases where you postponed decisions until you could learn more. “m_ r. '“fi —‘—'IWKB. ”.9th 5-" '< 3400 r‘- V" R] . l I‘LU Q 4 kn. f\ 9. 130 9 Please give me some exa..ples of situations during the laSt year in which you postponed a decision to act or not to act until you coulc learn nmore. ---. ‘ Please give me some examples of situations that occurred last year in which you did not have enough in1orrati on 1or taking action and in which you ielt t3:1t what you could learn would not be worth the cost and eilort oi learning it. *“ “a.-“—.——~‘——-—o “—‘—‘ Now I'd like you to give me some examples of situations occurring last year in which you were c3rtain of t :e outcome, that is, situations in WLiCh you could act wi t;out worrying about b3ing wrong. -“. .- _—- “...... —- — —' s ..Q- - -- ....“ -‘a—n fi“. Hg.“ Wer e there any occasions last year when circumstances iorced you to malted sions and act without iniornation you If ould hav3 oeen willing ‘\ to spend time and effort to get-—if you had not been 1o1c3d to act? Ies Could you please tell me what they were? [2‘19 33-56- ‘We would like to ask you what you thir: should oe don3 in the follow- ing situation. A farmer wants to trade his combine for a tractor. There are other i’armers in the ne1rloornood W70 also want to d.eal £01 a tr coor. (110cm 17:1? 00.11.1318) a. While he's still looking around to see w“o has a traCLor to tr1de for, should he kee quiet ab out his int’e ntions so as ~e " people he r35; t went to trade with iron having plezty of time to l (K t}.“\ " ( decide on how much they would want to he . Should houldn't 181 b. WLen he finally decides who he'd like to trace wit} 1 idea for him to act as t__ou:3311m snot sure whetLer he wants to trade so tLat other faznei wLo L 13Lt also be interest" ' ' would tLinL the tractor was not d3 3 realm? .—. \o ~ - Ids u Lo ('1 1. to 1s1a‘;:e a trade for 03mpe Hi Live position by offerin3 witLout lettina o. If h. Lines out tLat his neighbor is tr the sane tra ctor, Should he imp1ove his trying to ii no out 1n1at1;i_s n21:Lb01 is his n:,2i;;11‘eo1' know wizat his offer is? d 1 6‘0 m d. If he m«3ets somL one else who wants to t1ao e ior a t1aotor out doesn‘t know about the one at h3's intereste3 :i in, is it oetter for him not to mention tLatm he knows about tit straotor? e. Is it wise for him to try to mak, the man he' s dealin3 with think t1at a combine is what he meets meet, so tLat traLes 101 other item 13 won‘t be 3iven muo h consideration? f. If h Lines tLe tractor neeos minor ropa'rs tLe owner hasn't told him about, is it better for him not to m3ntion an‘tnin" ti1a m “Lt b3 wron3 with his comb? ne so that he can maLe the traLe successi‘ully? 52. Sometiznes a man may att3L1.pL to build a greater sense oi espon51ioility in the people he's dealin; with in oz‘der to malce them more relia ale. ‘ Do you Lnow 01 a case in wLion tiis.was oo no? Mi ve you yourseli ever done this? I Yes: Who are the people that ;¢ou oo this w11and unch r what confiitions? \YL K] O .- f‘ 90 . l82 If grou were in a group of 1500 peOple in wnicn you knew one person would have to bear a loss of $10,000, would you be willing to pay'ilO in oroer to get out of the group, and, d1us, avoid the risk of having to bear this loss? (iJAD S‘i ALILBIF On SXJUT OF QUJSTI Y5 A53 iTAID T0 RESPONDfiNT FOi NIH TO 131.111.: OUL) If you know that one person out of a group of 1,400 would gust a property worth 11-5,COC, at no further cost to him,1 IIould you piece of ‘ present income to become a member oi be willing to pay ng out oi your that :roup. 93 F4 1 _J -*\ m was vmj‘um 3 7:“!“1 10 ll. ' EU. ' (ICEAD S'l‘ Tifiill CH SIEET OF @351101513131) 11151.11.) 1-1111111l . n__11 TO FlLL DUI) (.0 k1) 59. a. bid you grow up on a farm? All of childhood Spent on farm Part of childhood spent on farm None of childhood Spent on farm b. What are the names of the How long did Did the; give you any schools you've attended? you go these? training in agriculture? Yes No __7 A‘ _‘ Yes 30 _ ‘ Yes K0 __ _A Yes No What was the last grade of school you Have you had any additional training, vocational training? Lo Yes: What was it? _._; fi-‘-W‘ * w~ ’—-‘-‘ ——— completed? “.m.———..—‘—. such as short courses or Nov lonr did it run? k) .0 *“ma— e. Did rou ever belong to: l 7' n . x n ~v‘ -.— a u-d uluoz Yes no The Future Farmers of America? fies No A "P 1" "1.2) "fl “' 1 ‘ ".3 :3" 'id "1‘ 7‘ 1r 4 ,3" n"? o0. ls tnis tnc only lalfl you we opclatc lo- Joulscli. Yes: How many years have you run this place? __ do: How many years have you operated farms for yourself? ‘__ How many years have you run this place? 61. Were you ever out of farming for a While? No Yes: For how long? What kinds of work did you do du “ Have you ever lived in a city? ring this tire? a. “—‘M—“ 18k No ‘ Yes: What kinds of work did you do during that period? *-w-M- —-—-—.-_.- 62. Do you ordinarily do any work off tne farm for income during the year? No Yes: Do you have regular year—round work, or do you just work 0:3 the farm parts of the year? All year: Is it a full day's work? Full day Part day Part of the year: What part of the year do you work? Do you work a full day or just part of the day? Full day Part day What proportion of your to tal gross income from all sources came from iarning operations last year? (lflmjiVIEUU: P l iSENT CAJLD) Le 35 than 1/2 “About 1/2 __kore than 5/4 :Al)out l/u “Between l/2 and 3/4 “Don't know how “Between l/k and l/2 ;__About 3/u ~much, but not ... all _N 0 answer 63. a. we'd appreciate knowing who also lives here, their approximate ages, and whether they're dependent on you? Eel ationsngtpptofl Respondent Axe Dede (INEEN IJ'J:R CLEO n IF SO) nabPOTU’NT —-‘—‘ ‘1.— u b. Are there any other persons not living with you to whom you contribute financial support? No Yes: How many? H v: .(IF nubrunoagf PEAS ANY ClILUAJN AT ALL) Have any of your children belonxed to h-H or FrA? Yes O F—d 61. Did you use any hired labor in running your farm last year? Yes: Did tkey work for you year round or part time? Year round: How mc' ny full time workers did you ham Part ti rne: How many were there? , On the average, how many days did t e average part—time worker work lor you? 65. What was your average gross farm income in the last three years? D | .... 65. We'd like to establish an estimate 0‘ your net worth. a. Could.you plea ase give me your best estini1tes of the value of your ssets at t11- be r“1min" of the year. We want estimates of the actual values, not the book values for accounting purposes. The point is, w11at were these items worth to you. Value of your land and buildings Value of your livestock Value of yourn achiner¢' and equipment Value of your feed and crops Cash on hand _. ‘ Value of your stocks, bonds, and other investments Amount of money owed to you U) Value of your other asset __...‘ 1'1 ’1 (loan) a a- b. Now, how about your financial obligations at the beginning of the year? What was the amount of: Your real estate debt Your Your Your Your Your short-te~m notes other notes accounts payable (money you owe) household installment debts r installment debts not covered in (D 0th. short term notes Your other debts (TOTAL) ~~‘~:‘-1”«*.;n W24). WURLL'J STATE COULTY TOVISHIP ENTER T15 FOLLOWING ’— er rm“, 1 7‘1~’* IN LLtJIJMWli DATE I“”‘ ; INTERVIEW NUMBER OF THINGS PRODUCTION FACTORS NEW DEVEL— HUMAN FACTORS POLITICAL, RELIGIOUS FACTORS PEOPLE VQQ Ag GOVERNV figHFRS’ MENT ORCANIF ‘ ZATIONS PEOPLE ATIVES PAST PRICES AND PRICE TRENDS CURRENT PRICES AND CHANGES IN PRICES PRICE OUTLOOK PAST PRICES AND THEIR TRENDS CURRENT PRICES CHANGES IN COSTS PRICE OUTLOOK EXISTING VARIETIES OF CROPS BI LIVESTOCK EXISTING METHODS OF PRODUCING CROPS SI CLIMATE, SOIL, AND DISEASE CONDITIONS NEw INVENTIONS, DEVELOPMENTS, AND PEOPLE YOU HAVE TO DEAL WITH IN RUN- PEOPLE WHOSE REAC- TIONS MAY BE IMPOR‘ TANT TO YOU IN RUN- CHANCES FOR DEPRES- SION OR PROSPERITY OF LOCAL GROUPS THAT MAY ACTIONS OF GROUPS AFFECTING ING FARM BUREAU, FEDERAL, STATE, AND LOCAL GOV'T ACTIONS ROUTE DRIVERS PROFESF SIONAL FARM BANKERS DEALERS, TIONS, LENDING ACENTS BUYERS PUBLICA- TIONS OF EXP STAT EXT. PUBLICA- TIONS OF FARM ORGANI- ZATI ONS FORMAL MAIL AD- NEWS- VERTISING PAPERS RADIO TELE- VISION AUCTIONS -—:'I INTERVIEW NUVIBER 188 PAS T EXPERI- ENCE ERROR ON WHO LE OPER AT ION NNNNN em STAIN OLIMITED 'ENCE OF KNOWN TO SCALE OTI-ERS BE TRUE TRIAL AND EXPERI- IOBSERvINGIREASOA“NG KEEPING WRITTEN RECORDS PRICES PAST PRICES AND PRICE TRENDS OF THINGS SOLD CURRENT PRICES AND CHANGES IN PRICES PRICE OUTLOOK PAST PRICES AAD THEIR TRENDS OF THINGS CURRENT PRICES AND CHANGES IN COSTS BOUGHT PRICE OUTLOOK PRODUCTION EXISTING VARIETES OF CROPS 8| LIVESTOCK FACTORS EXISTNG METHODS OF PRWCING CROPS 8 LIVESTOCK CLIMATE, SOIL, AND DISEASE CONDITIONS NEW DEVEL- OPMENTS NEw INVENTIONS, DEVELOPMENTS, AND DISCOVERIES HUMAN FACTORS PEOPLE YOU HAVE TO DEAL WITH IN RUN- NING YOUR FARM PECPLE WHOSE REAC- TIONS MAY BE IMPOR- TANT TO YOU IN RUN- W POLITICAL, SOCIAL. RELIGIOUS FACTOR S CHANCES FOR DEPRES- SION 0R PROSPERITY ACTIONS BI ATTITUDES OF LOCAL IPFORMAL GROUPS THAT MAY AFFECT YQR FARM ACTIONS OF NON-GOVT GRQPS AFFECTING FARM-N ING IE.G., FARM BUREAU, AMERICAN LEGION. ETC.) FEDERAL, STATE, AND LOCAL GOV'T ACTIONS AFFECTING FARMING 1.29.. I'm-3. "(Qs‘g‘;!‘f¢‘fi fl—rffTT HEIDI}; B The table below (Table 50) shows the distribution of the 66 . o 1'. questions used in the I.M.S. survey over tne Six Yield schedules. The *1 numbers in the first column corre5pond to the question numbers of tie schedule in Appendix A. The six columns of numbers in the body of the table Show the order of the questions on the field schedules. The last c:olumn indicates whether or not an information card was used Our- lI'I f. Lhe iniornation on tne card is J J- ' - S Ulono ing interviewing with the Cu. I (L shown with the question in Appendix A. A I1 If!!! HIP." III’T' ‘Ifl‘fl— DISTRIBUTTCT.’ CL‘ QUESHIUTIB OTI TIL: :‘lxI‘LU SULLLLEULQ ._ Dion CirL {gilHS' 'I‘I lULl INIuIIIItISr SclLe iule Question L‘ILLIIIILIer I JI'IIIN'LI, 3 1I L: W I LL 1 O I I I) J D :9 'I" o L \Q I\J C) I" K4 Cd a ‘I F I I I \ , V. .I. “ , , 13 2e 3 , L 1;, 3 ,1 1L lC L1 I ‘ 17 J 18 Y 13 2O 2? ' 21 L L 2 P. L} 3’ 22 24 23 25 2; 2L; E j :1.) 2 ,- l2 In 1N» I», / l] LO 2 7 2 j 2 ,2 :I. 1,; 28 1M 29 26 ‘M 31. L I; SL5) 1O if 1: L5 52 L) C l? l'I 3O 2 L 33 7 I it 15 21 2? IL 8 “ l9 1? ‘ 2} §; 9 9 2O 2C 2m 36 1O 1O 21 21 e; 7) F ll ll 2 2 T) 2 z ,> TIC 1;: \ .Lj 2'4. ill, '2 I R 1M 1D “ 22 :2 2U 27 1: lb 2 I) 21> 2 ,9 j 16 15 27 27 3“ jl i: 12 25 2; 2G 27 2'," 27 2 O‘ 2 l" 1? LI; 2E 28 5U lIg, la Aé 29 29 29 2; 1; 1; L7 3O 3O 31 41 1; 1, LI,” El 31 32 52 1.”) l6 O9 )2 32 53 :5 l7 1? 5U ll ll ' pl ll ll j 2 l] ILL} ,5 ll ll ;,; L; 1?, ll g, ll ll ; :3 ll 1.- 57 l? 51 14 5b 15 ii 5: 59 33 33 3A 3H 3; in 60 BA BA 3: 33 3A 3; 61 35 35 36 36 3: 36 62 36 36 37 37 36 57 63 37 37 33 58 37 38 X CL 38 59 :39 39 65 39 3) LO LO 39 AG 66 LIO IIO ILl Ll HO I‘Ll lf’O 191 APPENDIX C WUJK SHEET Loss Gain Loss Gain O 2: TE) “10* 2; II __ ‘10 2?: 1O IO 2: 2K? 501 ' _l,OOO 51990 10.990 E34000 l—2 No ~ 1 IO “cm—o ‘flo—u—I All_Yes QELOOO 2” COO 34L _+ ... IIEQLQQC ....JI 50,OOO . ‘. - Inconsistent ‘0‘ No Answer — Reason v ,u.~l.1t of III‘ Ill..- 192 APPENDIX D The methods of fitting the utility functions and the assumptions I made about the location of the indifference points and other identifi— cation points are Rive 'n the following outline. :3 I4 I. Hethod of least squares --‘ I ‘ O ” I "n O O T i A. Equation derived u31ng iour given pelnts ; I 1. Tiger: indifference points I , - a) In different intervals i O) Two in the sane interval and one in another interval 9 Q (1) two at extreme values of the interval (2) one at the center of interval c) Three in the same interval (1) more—than—fair point at lower extreme of interval (2) fair point at center of interval (3) unfair point at upper ex reme of interval B. E;uation derived using three given points and one assumed 1. One indifference point for each of the fair and more-than- fair odds a) In different intervals b) In the same interval at its extreme values 2. The origin 3. One assumed point at £0,000 dollar gain (or loss) and q(u) = 2/3 (155‘) <8). II. Kethod of Larranre interpolation A. O Equation derive d using three given points (all Yes answers to the more -than-Lair odds) 1. Two indifference points a) In difLe Tent intervals b) In the same interval at its ext ene values 2. The origin Equation derived using two given points and one assumed (all Yes to the more-than-fair odds) 1. One indifference point at the center oL the interval for the fair odds 2. One assumed point at 50,000 dollar gain (or loss) and =23 -——-———-— (A) 3. The oririn Equation deriveL iusing two given points and one assumed all Ho answers to the unfair odds) A One indiLLer nce point at tile cente rof the interval for [...-J the more-than-fair Odds. 2. One assumed point at RO,OOO dollar gain (or loss) and / q(u) = 2/3 (Lg-é) Is) Eouation derived usinv two given points and one assumed (allY es answers to tile um rc—than-fair and fair odds) 1. One indifference point at the center of tne interval for the unfair odds. 19h 2. One assumed point at 50,000 dollar gain (or loss) and l " 0‘ Q q(u) = 3/2 (---—-) (u) 01 For those cases in which there was no indifference point after re- placements and tranSpositions, the following assumptions were made con- cerning the slope of the utility curve: I) tor the cases of all les answers to the more-than-fair odds and No answers to the fair and the unfair odds, the slope at any value was assumed to be NP A C L C. C2 0| v rim—rs: “Flee 11.1, = the (dis)utility if the indifference point existed JILL at 50,000 dollar gain (loss) on more-than-fair odds. LL f = the ( is)utility if the indifference point existed at 30,000 dollar gain (loss) on fair odds 2) For the cases of all Yes answers to 'he more-than-fair and the 4 fair odds and all No anscers to the unlair odds, the slepe at any value was assumed to be 1_ ( E + dz) 2 ;0,000 Where um the (dis)utility if the indifference point existed L at 50,000 dollar gain (loss) on fair odds uUV = the (dis)utility if the indifference point existed at 50,000 dollar gain (loss) on unfair odds for the cases of all Yes answers to all three odds, the slope ‘ 1 . . . “U? was assumed to be m wnere LC =-v:77ff L1-C /U’ ‘e'Ukl’ \l and uU“ = the (dis)utility if the indifference point existed at ‘1 00,000 gain (loss) on the unfair ends. Uemco-293 Mg STATE UNIVE HIM IIIHHHIIIWIW“WIT!“ 1293 03062 1951 "Tlllllflr