Division of Ag riculturol Sciences UNIVERSITY OF CALIFORNIA FACTORS RELATED TO ACTION PRICE PREMIUMS- FRESH THOMPSON SEEDLESS AND TOKAY GRAPES Ivan M. Lee and H. B. Richardson CALIFORNIA AGRICULTURAL EXPERIMENT STATION GIANNINI FOUNDATION OF AGRICULTURAL ECONOMICS Mimeographed Report No. W9 1 60 ! ' -' pNiVERs^7er CA[?Fjiftt Ty ^^^^ DAVIS LIBRARY UNIVERSil Y OF CALIFORNIA DAVIS JAN i^i 1954 i TABLE OF CONTENTS Page Introduction 1 Description of Data 2 Method of Analysis 8 Summary of Results I9 Concluding Remarks , ^3 List of Exhibits Exhibit 1 Schedules Used for Recording Observations on Lots of Fresh Grapes Sold on the New York Auction Markets 3 2 Code Designation, Variables Entering Analyses of Factors Related to Price Premiums 12 List of Tables Table 1 Regression Coefficients and Supplementary Measures, Thompson Seedless Seasonal Equations, 1901 . 21 2 Regression Coefficients and Supplementary Measures, Thompson Seedless Daily Equations, 1951 27 3 Net Regression Coefficients Converted to Dollar Units, Selected Seasonal Equations, Thompson Seedless and Tokay .... 32 h Regression Coefficients and St^pplementary Measures, Thompson Seedless Seasonal Equations, 1952 3lj $ Regression Coefficients and Supplementary Measures, Thompson Seedless Daily Equations, 1952 36 6 Regression Coefficients and Supplementary Measures, Tokay Seasonal Equations, 1952 . 39 7 Regression Coefficients and Supplementary Measures, Tokay Daily Equations, 1952 Ul 21 .IT •3 ■•4. FACTORS REUTED TO AUCTION PRICE PREMIUMS- FRESH THOMPSON SEEDLESS AND TOKAY GRAPESi/ Ivan M. Lee-' and H. B. Richardson-' Introduction Growers of grapes for fresh market exercise a certain amount of control over the physical characteristics of the product marketed through the combina- tion of cultural practices employed. For example, berry size and the configu- ration of bxmches can be controlled to a certain extent by appropriate thinning and pruning practices. Likewise, timing of picking operations enters to influ- ence such characteristics as berry size, color, and Balling-acid ratio. After the grapes leave the vineyard, the packing operation provides further opportu- nity to improve the physical appearance of a lug of fruit through appropriate sorting, trimming, and handling in the packing process. Such characteristics as firmness of pack, uniformity in color and size, and the elimination of de- tracting bunch features can be introduced at this stage of the production proc- ess. This report offers some tentative quantitative measures of the relations between selected qualitative characteristics and price premiums for fresh Thomp- son and Tokay grapes in the New York auction market during the 1951 and 19$2 seasons . The information which forms the basis for the analyses summarized in this report was, to a significant extent, a by-product of a program initiated by the Extension Service in response to requests by growers. An essential feature of the Extension program involved the careful inspection, supplemented by photos, of selected lots of fresh grapes in order to carry back to the growers a picture of the fruit and pack characteristics which were, in the opinion of the observer, detracting from the salability of the fruit in this market. In the process of 1/ The research on which this report is based represents a joint undertaking of the Giannini Foundation and the California Agricultural Extension Service. The jvinior author was the observer responsible for the collection of data in the New York auction market. The senior author was responsible for the analy- sis of the data and the preparation of this report. 2/ Assistant Professor of Agricultural Economics and Assistant Agricultural Economist in the Experiment Station and on the Giannini Foundation, Berkeley, California. 3/ Extension Specialist in Viticulture, Davis, California. -Aftfrfmoo 9riv+ rig'-oidi L--;fejiTsm .torrbo-rer '»r!,+ "^f^rfo J.'nr'^vr «5 .C'fO[OT'TIa6ai aoiitmn >fioY ^^n&M srid- lo aoloflo -ftoo risfi'l lol esqsig ^^^lbe~^ ' noaqjiioriT lo agnrEa-' ^)nl l-^io :iicY noiS o,* ha^om ti&ruo'l find boR «*lil sfio oaswj-sd ,5c;vX bns ic;' *ns>3 ■om Xs^o^ sn^* i&sx dots ni il Tol iel^x/o ns aA. ni isjitsm > sdi ' BRTT 0360Tr:'> tpvo lo noxrfoel''?. siicf ui d-rfiie'-r ©mos &ovi% Bs?r rioxriw 10* --'sl sii: t.-trretsqqs s-r ? -'Vlovni sSnsiaiJb gniqqJtris tcanoX sricf .liruj 30 oB'XBdo no a.-v icfo donj-i ebn&^iiin^ts io^m 16 d-oft ax loi -tolnx 9fiJ ai beaingoo--^'! \^StoHtpt» df>&6 «tOn svBri ^ar^s-taxb Sfiiqqine otf bsJ-sIa'i -6l3i £ aabiVoiq J^'^'-" W9tl ^avsjlof dasil oi . ' rfiiW -.bab^oos 16 ii&o isq 5 osi!* ^es-I iol S2v'l ni g(ix*ni;d»Sp yrf^iiuo .fnBHoqmi azs^i \l»ytJ uoii^Mf^i ec\S ai bitia g^w riSxrf^ I'S lis Y.XXsbjtiv te.*;; ■ ;^tfQJaieJ-ni . bnsi i -tTiJ^ 3fliiub lool'i ncsi-foiJs ::-xoY • no WfeeXIoo aisA s^sb ui/O no -lavo iio bobiooa-. ste"? anoijevtoadQ \Xti5i nj! »3rt6dB&3 ' , .+oif«5 BiriJ «b Bioe Siol easlbaeS ftrt33n'o:ij' ^^i^- io )n^o -t^.f .eXioo Gdtt noii^Wtolfii tS^^I nl »0{: yXul grrirtniged BTta&T svi.t jvxios aBi7 JejJtsr" 3E?Xb99S noBqniori" -.OC ^^fii/tj iKhirtig^?':: ' ■ V imotlln^h ni i^AtBtn eiit no etevr axsMol' i^^IW* iftoi'iaq s'siina i-.xitJ r>a£9^ S^'^i brid SnrxwC .boxisq eiri<^ Ic IXeri -ieiifiL sri* gnxiyb a^oJ. XB-A 'Ijod lo *as>o i<=>q t>jl x^tssiil ■^■i?v no bshiossi essni ' ■ [ ^e. fasbuiDni t8c»lL'b9fio3 X^nxliio on* lo noi^^^qsni nA \i gnxTCftl iioX doss* lol babToost asv.' rioxriw nox^famtolitX la BflxJi s.ii .r -.?V bagftEns ea-JT Moa ad o* .Jix/Ttl erict tt^b ^^^^ ^nibpoe'xq soin=tc; <-i 0U& 9Ai no t83ir" 'levd 6^ OX moil x< ' ■ gnl^v ,8cloX \6 -nt 101 benfeqo ^XibXk^si s31£s* «toX dOB? fil agui odi lo ^noo toq X iiso^b. i -isfid eiii !>.'. 3r g3 lo §nia3qo ari* 9iOls»d a-t3\ua £»vX-+tJ^30-iq xd r.cjic^eqS i .<-3SbIo gfe'T ;loX ri;*B9 5srii agixi beneqo e^Lvit Id noWosqsni rtA lo ~hh di ax hsxlxosqa eoxja- ~rio efnw nr. nr.tr>. ■: zadi . . • .'3 ,©si's vYtsd no •a^oi arfi In . . ■[ -do oi esiT iq alriif : . 'roj-o"e'i sni 1:r' v^rs 'hi c ■ ■ 9vjt8«rlDno3 B Tol . :.y\csx bebiecV sVkH oi ba'tsbianoo sd 9rfJ lo nof.+' "i ?ri't ji^oc' '^r ■' •-.Jieoita '*h?-oi 9ifb '^-ii-* -aib eriJ moA. 4d einp?'.^ '^M Mfi^ ''.s^"-* 7. as the vdthin-lug uniformity Tdth respect to the individual characteristics and the general appearance and condition of the fruit as it reaches the auction floor are here reflected. It is clear that the omnibus character of this fac- tor limits materially the information which it supplies for meaningful analysis. Yet, it seemed plausible to proceed on the assumption that the characteristics recorded under fruit quality might well exert an influence on price independent of the individual factors observed separately, although the characteilstics re- sponsible for such an influence are not escplicitly identified. Attention is called to the fact that several of the characteristics observed in the 1951 study were either omitted or altered in the 1952 study. In part, these changes were instituted in an attempt to obtain observations on additional pack and fruit characteristics for analysis and, at the same time, keep the length of the schedule vlthin manageable proportions. The time for observation on the auction floor was limited. Furthermore, completion of the schedules, as has been indicated, was only a part of the observer's job. It was the time limi- tation which motivated the substitution of the judgmental classification with respect to berry size in 1952 for the weight measuronents employed in 1951. It is recognized that, from the standpoint of the phase of the study reported here, this shift is probably in the wrong direction. However, it is difficult to as- sess the loss of information involved in this substitution in the present frame- work. As noted earlier, the objectivity sought through weight measurement may well be misleading due to the sampling methods employed. Aside from the observed characteristics of the fruit or pack, certain addi- tional factors have been assembled for inclusion in the analysis. Important among these are mode of shipment (freight versus express) and brand name under which the fruit is sold. These factors, of course, are not unrelated to the pack and fruit characteristics observed. Mode of shipment would be expected to affect the condition of fruit upon arrival in distant eastern markets and, to this ex- tent, is probably partially reflected in the fruit quality and stem condition characteristics. A particular brand name, likewise, probably attains its status in the market through the reputation it acquires largely through the character- istics of the fruit packed under that name. Recognition of these interrelation- ships, however, does not preclude the possibility that mode of shipment and brand may exert an influence on price not explainable by the characteristics of the fruit or pack observed. It will be observed that mode of shipment information was recorded on the schedule in each year. The brand information was recorded directly on the schedule in 1952. In 1951, the brand information came from sri^ KsrtoESi'i J j as ALtrA ,h9i iJ:; . Slaw frrt* rsbr-: ^0 ^9 n r 1 flrli 8. auction slips made available to us and easily identifiable by lot from the iden- tifying infonnation recorded in the heading of the schedule at the time the fruit was inspected. The absence of information concerning the particular fruit and pack char- acteristics reflected by the fruit quality, mode of shipment, and brand observa- tions is a notable inadequacy of our data. This will be particularly apparent in the case of the latter two factors which, it develops, are dominant variables in several of the formulations summarized in the next section. Had infonnation been collected on the important separate characteristics reflected by these three factors, a more meaningful interpretation of the results with respect to these factors might have been possible. On the other hand, there rosy well have re- mained net effects due to these factors. In this connection, it is remarked that it was not the intent that the schedule should specify all of the separate char- acteristics in detail. Those selected for observation reflect that orientation of the study toward characteristics which are, in part, controllable through ap- propriate cultural and, to a lesser extent, packing practices. Position in sale represents another factor which was recorded on the 1952 schedules. It is the practice on the auction market to rotate position in sale among receivers from day to day through the season. It was observed during the 1951 survey that the first position seemed to be preferred by receivers, suggest- ing the possibility of an associated price advantage. Information on position in sale was collected in 1952 to investigate the validity of this supposition. Method of Analysis In the 1951 Thompson Seedless study, just under 1,U00 schedules were taken on lots of grapes marketed during the 8-week period beginning July 30, The l,iiOO lots observed represented something over 90 per cent of the lots marketed on the New York auction during this period. In 1952, around 2,800 Thompson Seed- less sched\iles were obtained during the 19 consecutive weeks beginning June 30 and some 300 Tokay schedules during the latter half of this season when Tok^ marketings vrere active. In the latter year, schedules were obtained on almost 100 per cent of both Thompsons and Tokays marketed during the period studied. The object of the analysis was to obtain some approximate indications of the quantitative effects of the several factors observed on price in the auction market. Conventional multiple regression methods were employed in analyzing -ei no. iitn'i If f 9d His? eir.. oa b bx S: ineninrob o-jb leqol- ; zioioBi ov/cr isJJ-si sfii lo 98=0 aricr ni ' f>pJo&X'if-» ." -.*-iOQuu erf* no t ^ s»rii .'iR vlxooqa hii/oria eJjjbsrios iorfi .* aa-v ji ■• •^nxjfoRa ,>tne^r.9 ^^9^ti0^ " . lucflx/o 9usx"i ; . nei isflw I! • . ■ ■ -"tq&i sl^ia fii no . : yt Qi ie>!iftm notioi/5 srii no Qoitos-. , ?:t?>v iff-^fvi Yd ba-f'X5>?«ntq ©cf <,''i a^m-^'ei noii'iacq J"3nii 9rii ■Jf; ftBvbB s3x"iq bs^r.rooass n? lo yd . ■ grf i" ■ .t:.3 airti 16 yilbilsv 90^ aiesiiasvn; £ ni fc. ss^ else ox axaY-ff--^A -sM a ■ ;-;33 004] ci •rgbnxr .. , ^xf^s 3ae-Ib9C'o nogcp.:>aT 15^1 srf* nl nn.f'^ocf boi-i&q jIsptt^B ^rid' gnfTCfb ■•"m s-cisii 1-^ 8ic- ■ "iXsri le^tjsi. 9ricf snxix/b aexx'barioa \- iq nc 1 • • • -a e^Of • ■ . 9. these net effects.-^/ However, preliminary analysis was necessaiy in order to bring the problem Tfithin the regression framevrork. The chief difficulty here is obviously with respect to the nonquantitative character of the observations on most variables. Where a classification with respect to a nonquantitative variable on the schedule contained more than two classes, it was necessary to dichotomize the classification. With more than two classes, the appropriate numerical values to be assigned to the classes are not known. Dichotomization can obviously be accomplished by discarding classes, by combining classes, or by a combination of the two. Preliminary analyses with respect to price and each individual qualitative characteristic served as the basis for the dichotomi- zation adopted in this study. For the characteristics on the schedule containing one criterion of classification and more than two classes, dichotomization was accomplished by combining classes. In cases of two criteria of classification, dichotomization was accomplished by suppressing one criterion. For example, in the case of color on the 1951 schedule, "green" and "light green" were combined forming one class, with •'amber/yellow" and "straw/yellow green" forming the other class. In the case of stem condition on the same schedule, the color criterion was suppressed. In this case, also, the very few lots placed under the "dry" criterion were simply discarded leaving "pliable" and "fresh" as the basis for the final classification- It is evident that, in the process of combination, a part of the informa- tion contained on the original schedule was sacrificed. The preliminary analyses revealed, however, that the loss of information was not as great as might be as- sumed from an inspection of the form of the schedule. This is due to the fact that, for each of the characteristics observed, most of the lots were placed in two or, at most, three cells. Study of the frequency distributions aided in the dichotomization in another way. It permitted the formation of some reasonable basis for ordering the criteria within each characteristic for combination into classes. In most cases, the ordering on the schedule is appropriate. However, in the case of color in the 1952 Thompson Seedless schedule, this procedure in- dicated fairly clearly that the "straw green" criterion belonged with the "green" and "light green" rather than with "amber/yellow" as ordered on the schedule. 5/ A similar study on selected vegetables in the Boston wholesale market was made by TJaugh in the late 1920 's. In that study quantitative measurements were available on most variables entering the analysis. See Frederick V. Waugh, xy^^^^ ^ Detemin ant of Vegetable Prices . New York, Columbia University Press, ion erfi. 0* oi y7pp^iei)9ii" '2'^.' »3aBl5 '^v^ Ofuli Mora benxBiaoo sXxfbarioa eriJ • nox. jrioin .rw.-vT>! ion a^gasJ:- od sgixlsv IsaiT ,2PE8sin i-5 gaif ■•.-'■sUciKOooft ad xXeuQivdo nsi f^^4,,^^„,jr ■ iia sivxJ-aiilBi/p i ■pMi ttl „^ j.:,,.,^*..,, .• _ , ^ . ^ T 9ri^ r%o poiixt'cioa /cs^s lo.- 9> ^• 999X1906 'T»niraae-iq,9riT .•*>--.•.,.. .pa-jG.,, • • ■ no. *»9iiifiS^nc9 ^i.^ - fc,. • As- f»rf^-lo 'rtoija ■ _ . _ . ^-'■■';r-■-T'' Arii "f>t'J-'^ .yiJ'^^'> 9^'£ri^ ■*^^«^;.^-• -J-- , . ' - "OS I0 t.tr— .^{fiv ledjipm , '-' ■ / .■'^Ip© ji^:^* f ni tlCS bps . ..:.?bpo ai z?^ "•^i'ip: i:^r-or.-:j r»lj^-^^•<^>^?^ y?:A^Ss.ixe ■.y^::ii-iii?p> 3i?^Ii5fi^; :?jiScsr?|C.:Vi!-f ;:flf i3ii2T5^ii.r£> •^ ^^Pf-'^t'S-^i^^'v. '^^^^^ -iVy . : i.fS!a^^^;^^- EXHIBIT 2 Code Designation, Variables Entering Analyses of Factors Related to Prxce Premiums Lot price as per cent of daily average Code Fruit quality Code Ballxng-acid ratio Code Stem condition A. Thompson Seedless, 19^1 55.0- 59.9 1 6o.O- 64.9 2 65.0- 69.9 3 70.0- 74.9 4 75.0> 79.9 5 80.0- 84.9 6 85.0- 89.9 7 90.0- 94.9 8 95.0- 99.9 9 100.0-104.9 10 105.0-109.9 11 110.0-114.9 12 115.0-119.9 13 120.0-124.9 14 125.0-129.9 15 130.0-134.9 16 135.0-139.9 17 Poor Low good . Medium good J i Top good Excellent } 3 Color Straw/yellow greenl Amber/yellow .1 Lxght green \ Green J Berry size (gram weight per berry) 1.7-1.9 2.0-2.2 2.3-2.5 2.6-2.8 1 2 3 4 14.0-15.9 16.0-17.9 18.0-19.9 20.0-21.9 22.0-23.9 24.0-25.9 26.0-27.9 28.0-29.9 30.0-31.9 32.0-33.9 34.0-35.9 36.0-37.9 38.0-39.9 40.0-41.9 42.0-43.9 44.0-45.9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Pliable Fresh ^8 Mode of shipment Freight Express — X — 9 Brand Wonpremium Premium (Continued on next page.) TSq-'-TSf'/d I 1 0 r:-? S'O-S-5 (Sj.3aj r.c.-SfT! ---.ji:- vujpex.\7\6jycy-v j --i tjCOT, • 5 5 59'0-5c:- ^1 re I' I p V 3 ! i S t _i 1 X-- X i I Exhibit 2 continued. ^10 Daj-ly voliime of sales Code i hi Daily per cent of lots rated quality 2 1 1 Code \2 Daily average price '! I 1 i Y ^13 thousand lugs dollars 2.0- 3.9 1 0 - 4.9 1 2.10-2.49 1 1 Monday 1 k.O- 5.9 6.0- 7.9 2 3 5.0- 9.9 10.0-14.9 2 3 2 . 50-2 . 99 3.00-3.49 2 3 Tuesday 2 8.0- 9.9 h 15.0-19.9 4 3.50-3.99 1 4 Wednesday 3 10.0-11.9 5 20.0-24.9 5 ✓ 4.00-4 49 i ^ XXiUx oLLcxy u 12.0-13.9 14.0-15.9 6 7 25.0-29.9 30.0-34.9 6 7 4.50-4.99 5.00-5.49 6 7 Friday 5 16.0-17.9 8 35-0-39.9 8 5.50-5.99 8 18.0-19.9 9 40.0-44.9 9 6.00-6.49 9 20.0-21.9 10 45.0-49.9 10 6.50-6.99 10 22.0-23.9 11 7.00-7.49 11 24.0-25.9 12 55.0-59.9 12 26.0-27.9 13 60.0-64.9 13 28.0-29.9 14 65.0-69.9 14 30.0-31.9 1 15 70 0-74 Q 75.0-79.9 80.0-84.9 85.0-89.9 90.0-94.9 16 17 18 19 (Continued on next page.) ■T " ■ ~ ~ J . — — r :7yr--rr-r~VJ t 1 1 \ ! ~--r--' '". > ;..T.,-^ i J 1 t ' ■ i < i 1 > ■ i t- . ' ( j t ) 1 .] 1 ; If' I i 1 1 \ 4 i t i 't i tr j i i '! * 1 I • :?.'0-3.J:.'? ; -orb 1 T3 • TS • : „■ > ■ . ■ »> --' TJ ■ TO ;■ 0 i i 1 1 rS ■■•L>.0-$-.-i^?- !. 1 ' J 1 i e' i 1 f 1 yOra- od- \ 1 ^ TO 3 i " T J 1 i 1 1 q-CTT?T.!' 1 1, !^-!9rJ s. : cij jo^-e is-^: ., 3 Exhibit 2 continued. h Lot price as per cent of 1 Pnrlo 1 U UU-C Fruit quality Code Balling- acid ratio Bunch description L/Ocie B. Thompson Seedless, 1952 60.0- 64.9 65.0- 69.9 70 .0- 74.9 1 2 3 Poor 1 Low good > Mediiim goodj Top good 1 Excellent j 1 14.0-15.9 16.0-17.9 18.0-19.9 1 2 3 Tight Loose X 2 75.0- 79.9 80 0- 8i|- Q 4 2 20.0-21.9 22.0-23.9 4 c h jyioae 01 65.0- 89.9 shipment 6 6 S Color yu.u- 7 26.0-27.9 r-i 7 Freight 1 95.0- 99.9 8 9 Amber/yellow Straw green 1 Light green/ Green J 1 28.0-29.9 30.0-31.9 8 9 Express 2 105.0-109.9 110.0- 111*. 9 10 11 2 32.0-33.9 34.0-35.9 10 11 Brand 115.0-119.9 12 36.0-37.9 12 Nonpremiiam 120.0-121^.9 13 \ 38.0-39.9 13 1 125.0-129.9 l4 Berry size 40.0-41.9 14 Premium 2 130.0-134.9 135.0-139.9 15 16 Small I Medium i 1 42.0-43.9 44.0-45.9 15 16 ll<-0.0-l41t-.9 145.0-11*9.9 17 18 Large 2 150.0-154.9 19 155.0-159.9 20 (Continued on next page.) i i i 1 Ik T 1 1 s 1 •t ft .i '™ '^s aj r" ■ T9 • " : Sxesnj ' ! j cici'o- ao'c- i ! Tv-; '.'V- >«1-*<" , . 1 t i 1 Ad'O- AS'' 1 iO'd- Xr'h- 1 ■ ! ■ I 1 i- i ! !, .■■ . '• ^' ■"c b'J.Ycc' • i A ! ) ! . 1 i i i i i 1 1 i T I'C-iUfrTar XT ;c. 3g*0-3T"o 9 1 ' 1 r. I -il - ^ •* " ■ V- -- - ^- .5 ! r-H^W*- - ^^^^ Exhibit 2 continued. ^10 Daily volume of sales Code hi Daily per cent Of lots rated quality 2 1 Code X 12 Daily averaee nrice thousand lugs dollars 2.0- 3.9 1 0 - i+.9 1 2.10-2.49 1 Monday 1 h.O- 5.Q 6.0- 7.9 2 3 S 0- 0 Q 10.0-14.9 p 9 •iD 9 QQ 00- 4q 0 eL •3 J Tuesday 2 8.0- 9.9 k I'l 0-10 0 0 en •} QQ jOu-j.yy k Wednesday 3 10.0-11.9 5 20.0.24.9 5 ✓ 4.00-4 49 k 12.0-13.9 1^+. 0-15. 9 6 7 2S 0-2Q Q 30.0-34.Q 7 1 k ^n_ii GO S 00- s 4q 7 1 Friday 5 ifi n_i7 Q 1\J • V— J- f my 0 0 0 5.50-5.99 8 18.0-19.9 9 40.0-44.9 Q ft on_ft ko w . vvj— vj .^y 0 y 20.0-21.9 10 45.0-49.9 10 6.50-6.99 10 22.0-23.9 11 50.0-54.9 11 7.00-7.49 11 2l^.0-25.9 12 55.0-59.9 12 26.0-27.9 13 60.0-64.9 13 28.0-29.9 11^ 65.0-69.9 14 Of) n 'SI G .u- ji , y 75.0-79.9 80.0-84.9 85.0-89.9 90.0-94.9 15 16 17 18 19 (Continued on next page.) 1 .... ! i i 1 i cf ' ysT 'J : J li t f r .1^ 0-Ao d r .vO'O-llt-ii' t5 Tr ; j3 TS Ji SO'O-ST'ci TO 1 Jtf 0 io O ■ 1 .i ■ • 1 i TO *0-T St' & \ I i. ■ rh r i per \ ox, "Ofe / i 1: XT TO , .T.pnl^nqtrJi. i ! t .J 3 5 Exhibit 2 continued. Lot price as per cent of daily average Code Fruit quality Code Balling-acid ratio Code X 7 Bunch description Code C. Tokay, 1932 75.0- 79.9 8o.O- 81+. 9 85.0- 89.9 90.0- 9^.9 95.0- 99.9 100.0-104.9 105.0-109.9 110.0-11U.9 115.0-119.9 120.0-121^.9 125.0-129.9 130.0-134.9 1 2 3 k 5 6 7 8 9 10 11 12 Poor Low good > Medium good Top good L Excellent J Color Pink 1 25 per centj' PuiTple T 50 per centj' Berry size Small Medium Large 1 2 28.0-29.9 30.0-31.9 32.0-33.9 34.0-35.9 36.0-37.9 38.0-39.9 40.0-41.9 42.0-43.9 44.0-45.9 46.0-47.9 48.0-49.9 50.0-51.9 52.0-53.9 54.0-55.9 56.0-57.9 58.0-59.9 60.0-61.9 62.0-63.9 64.0-65.9 66.0-67.9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Tight Loose ^8 Mode of shipment Freight Express X 9 Brand Wonpremium Premiiim 1 2 1 2 1 2 (Continued on next page.) 0'0-llfc ■ -jU-c JOO'vO-IGlfB ?>i^ • • - A> • ;• I- 4 ..5 r — j lob Sooor ! ■hsz. '' foe Bticfe flf 'c ■ ■ ■ ' • 1 1 1 J t' 73 T5 1 X 1 1 ■1 1 V • 1 ? is 1 ^ * j ^•0-35 ■(> ! : i -■ - .r. ... t 1 ■ ; 1 ! i ; } ! Exhibit 2 continued. Y ^11 Y Daily per cent ^12 Y Daily volume of lots rated Daily 13 of sales Code quality 2 Code average price Code Day of week Code thousand lugs dollars 0 - 0.9 1 0 - 4.9 1 1.60-1.99 1 Monday 1 1.0- 1.9 2 5.0- 9.9 2 2.00-2.49 2 Tuesdav 2 2.0- 2.9 1 A /N 1 Ji A 3 ■3 J 0 tjo 0 CO ■3 - Wednesday 3 IS 0-lQ 9 4 3.0- 3.9 4 20.0-24.9 5 3.00-3.49 4 ' Thursday 4 4.0- 4.9 5 25.0-29.9 6 3.50-3.59 5 Friday 5 5.0- 5.9 6 30.0-34.9 7 35.0-39.9 8 6.0- 6.9 7 40.0-44.9 1 9 7.0- 7.9 8 45.0-49.9 10 8.0- 8.9 9 50.0-54.9 11 9.0- 9.9 10 55.0-59.9 12 60.0-64.9 13 10 0-10 Q 65.0-69.9 14 IT All ri 70.0-74.9 15 75.0-79.9 16 80.0-84.9 17 85.0-89.9 18 90.0-94.9 19 . . ..__J (Continued on next page.) i j t*0 ' J-Ojt'ci i ) I ! 1 t . ... i t ! 1 •1 1 r , ' i i i 1 .1 ■ 1 •i T 'y i i 1 .4 1 1 t 1 j 1 1 1 ■ 1 1 , 1 1 1 t i ■ 1 i ■ ■ ■ A '0 'd P'O- Q*?* ; > ^ J. J *^ ! ! 1 I ' - J I - T \ i i f 0- -1^ o Si? O'SO i * 3-0- 1 i 00- i l^d f i • ■ ' i ! v; , - < • > ' M,cTTi-.3grA i 1 TO"0-If'c ; ! ■ 1 1 ' 1 1- T ■ ^1 1 1 ■ 5 ; 5 ! 1 0 - O'c i i ! i * ! . .. cfojjex." . , ( i 1 - nog& - ! . f)fs5w OT, Hopp f f X. J I- 1 18. The notation used to identify the individual characteristics in the analy- ses summarized in the following section is as follows: ^1 = Lot price per lug as per cent of unweighted daily average price over lots 2 Fruit aualitv 3 Color X, h Berry size X^ s X, 6 Stem condition Xr, 7 Bunch description Xn 8 Mods of* shinrnflni'- ^10 Daily volume of sales ^11 s Daily per cent of lots rated quality 2 ^12 Daily average price ^13 Day of week Uniform notation is employed in the three sets of data to facilitate conparison of the results, although certain variables do not enter all analyses. In Exhibit 2 it will be observed that the variables X^* X^, X^, X^, Xg, and X^ represent dichotomized variables which take on values of either 1 or 2 in each analysis in which they appear. Dichotomization has been accomplished in the first four of these by suppression of certain criteria and/or the combination of appro- priate classifications appearing on the schedules. The mode of shipment variable represents freight and express coded 1 and 2, respectively. The brand classifi- cation was constructed with the intent of reflecting roughly premium versus non- premium brands. A brand was classified as premium if the price which it conraianded exceeded the lot average price on the corresponding day more frequently than it fell below the average. The definition of brand classes was based on the entire set of lots observed. Premium and nonpremium brands were assigned codes 2 and 1, respectively. Due to the substitution of judgmental observations on berry size in 1952 for the weight measurements employed in 195l» dichotomization of this variable was necessary in 19^2. eoriq 93S19VJS '^IxsJ? •bostxlgxpwnu . fo inso isq. riS ^i/I- 'leq ssx^q. ?o ■ =--■ . ejoi i?vo o-i .tGi bios-inxIXscI irta.mqxria . lo sho?.! = . ;^X. .^saTcLens • lis leins Jcfl ob. e^ifJcxTfiv nxsiiso rigworiJXs eSiXirsan ^d^ rises It i S 'io i -iBsiiic To a^ulsv no- 33teJ rioxri.'r s^tidsi.iBV bpsxraoiorloib ^neas-rqe-i j^X Si/Wi &ci^-- nl fcsriei.fqmooDs n95«J 3s;l "noicfBSiffi o.+ oris xQ * .usaqqs \;erid- riolrfw nx sxa^ci^nB -o'xqqfe' To noxtfinxdfflo.-) srfi -^oXfans si^silia nisiiao '5''. iio is ^siqaua ■^jd.a.isriJ lo -xirol 9.£cfejrii3V j-Of'nqJcda 'io ©boni eriT. .aoluberioa erli.no anbsseqqB snoid-sDilxsafilo s&ci-i^y -ni&ZBLo bfisid erfT . VAOVxd.osqse-r ^$ bm L baboo aaonqxe bns irigxs-^ B+nr-aeicsfj -r.on 303197 r^isaeaq YXrfgt-'ts'x gnxioelT&'i lo J-na^ni erii x!*xw bs^oxn^snccj. as'r nci^so bafafi^raiioo :ii rfoxrhr aoliq axU Ic rPi.'Xii?'5*tq bs bsiliaae-Co sr"- bnstcf .. A .abusid nau Lmaiq d-x'nfiriJ- ■^I^^fiaxtp81'jt e-son: Y.Bb->}aifcno«5aeT'xoc.. -atid. no- soitq sas-xevs rfoX prf* bsbe&oxe etiUm '?tii no bsacd sstt- asasBlo bflsfi "Jo iif>i:.txnJ:l:?i> sriT .?3siii»v^ srfc} ivclr>d {I-'>. t.L- b£i6 5'3eboo bsngisau s*C9^ ebflsid myititstqnon- bns rayxmei^! .bsvisero a^oX 1.^ i^e 94!i3- --C^led no anoicJ-evisedd Iod-n5n^bir.t;. Ip- no:jd..\-''.'*edu2 ©rid o^. ©jjC- ..■^Xeri^osqaei, 8xdi" 'io nfti:}£'si;it:Dd-odoi:b. jJ^^X ni, oi*-'i ^xebi'i"? n^co'ixivt ^jsbnoM «3V^>b e/id oi 5 rl^isoisli I eiediRDu pnirnaiaBs xd frf^l'i.o erfj iiiric*- no jJiaoqqra s>iiT .yisi^xdifi ■•iibsJiiiiibB .otuboooiq BxriT .Yisvci v/^fi-J amtfxmoiq eoitq s-.^ejnao'iaq rtc aoi^'iXi&^osii^riD svxJ-fijxXEop leievea arfi lo sau?63 diarfj^ ii9V9W;H .s^ldieyslq smese ^95t ?»nj- lo eYsb in^isllib ao inai&llrb -lib ftrii- nss-jriocf qiiJanoi^sXsi ©rfc» to labto loxi^is 'xol exasd xaoirq s el^Jil &d6i&qi?ci .r.cafiST j-ldJ- lo'^ .bsj-soibrrx smbeooiq eri.t ni b$-*£ioqiODnx e/ab ine'rsl .^aeo rioae Pii? 1o ysb dnse •ro'i nsMs^T^ebnu nsscf svsri seaiilnn?. o^ai^qotqqs 1o asa-sifenfi oini nolioifcoiiai sriit riguo'id* 09*153 r^sevni: ^'^'^-"^ - e-rocfi nsvig od Ilbv a/rrisj- s^sailT .B'/otf-jsl "io anojcjnnxdson bobo?) bits am*'^ d-of/boiq i^):/-^sl3 srii nx sitHi/d tjsa.'Ta-fs'i&'rq e.l^ gnfais'iqqs nx fcxs \r>iii as isloanx i^Ino '.to ^ol isli/oxi'xsq e sasriDiirq srf^ oct' iooqeei ri^iirr snoiaioob 'aioYi^d 11 .eaatq yrld 2ja©3sijs airfrf' tSg^-i bs-^siqaib lo nojrd3?qe.ii no baa^q et° nQlio'r:B xtx *xuil ©fit lo eoi^aiis^os'tsrio ssldsviasdo fl9eonxR noii5»- xrx sstiootq no Hutl edi lo slsa rif^owied ■^al edT .sssaoiq gnlbbid srfJ- ai noxaesxqxa dTxoii :> -noapet rix ij .^-rodJ? Y'^^vi^elsi ex a'lemoan^.f? t{<^ sasrijymq bni5 ^ooIl ncx^o^os sri'd- -S's &d ^'pm aeqs-sg lo )o£ loiwoi^ifiq s o* aoitouai -isrrtciaaoo cfsriJ- saciqqija ^^idy ?.Ja-5bionI ^orci* J-ioris v:X'3vj'-"'''^Xc>i 3 ni noiJoue, erii ni loivsrisd te^d ai b&ir/aXl lo Qon&id .slqcRBXi? lo'^ .XRx;&£/mx *on eia v:J-ivJ:iian9n lo faniif ?.Xrii ani^osXlS'i vsfiaoqcj?* £ i^^ifta oc" foevteado nssd avsri ^iitsjup nl gnoi.is -^ainnin -^XiGuTion ^ilu^l Si-ii no noiSB-ne^dO .soxJ-aXie-tosTcsjlo ins^ioqaix t»?XuDicht6q ci ^oaqesi riixw qcsolc 10I etsiii srfsiipebfi asw a"^sb ti/oI oi asxrii vtcrij asees essd^ nx baise^aii/a ;f»>!i)!{ii! Ic To/vsiiad -saxiq f^dt ai b?.'36-ri9'X 9d cyi ^oubo'iq toiielat ©rii lo aoftsiseqqs ©rfJ batiupsTt sew ?3iif io ^nuotoK ©bibs srict ixfOcfA ..Ixl'tI larl^o ^viJ-cXsi hmrfd ahii . bec+oC'Ttco /i39d bi;:i x^n^-xtixl';!-' Y^-^'Xex^p ©r't • ri- ?v r-hr-' n'" ---': ' 20. The characteristics included on the schedule for observation represent those which growers, packers, and production specialists considered important as quaUty identifying factors. From an economic point of view, the most meaningful defini- tion of quality is to be found in the expression of consumer preferences in the market. Hence, if the commonly held concepts of fruit quality are economically meaningful, one would expect this to be reflected in the price structure in the market. From this point of view, this analysis may be regarded as an attempt to assemble some evidence which might be useful for appraising the popularly held views regarding the relative importance of various fruit and pack characteristics. In the presentation of results which follows, the 1951 Thompson analyses are discussed in somewhat more detail than the 1952 Thompsons or Tokays. The reader will have no difficulty in fitting the results of the analyses of the latter year into the framework of the discussion of the 1951 results. Thompson S eedless, 1951 .— Selected equations from the 1951 Thompson Seedless analysis are exhibited in Table 1. In presenting the results, the practice has been adopted of presenting regression coefficients in three forms: in standard- ized units (p), in coded units (b), and in price percentage premium iinits (b'). The beta coefficients (p) permit direct comparisons in standardized units of the coefficients in a given eauation.l/ The b coefficients are those coming directly from the analyses. The b' coefficients are presented to facilitate interpreta- tion directly in terms of the original units of the dependent variable, that is, per cent of average price. Coefficients of only three of the variables reflecting fruit characteris- tics appear statistically significant. These variables are fruit quality (X^), size (X^), and stem condition (X^). The positive relation with Xg indicates that, on the average, lots recorded as "excellent" and "top good" commanded a premium over lots placed in the lower fruit quality category. The positive co- efficient of X|^ indicates that larger berries commanded premiums over smaller berries while the relation with X^ suggests that "pliable" stems are discounted in favor of "fresh" stems. No ambiguity should exist with respect to the 7/ The beta coefficients are the regression coefficients which result when each variable is expressed in units of its standard deviation. It is in this sense that the units are standardized. In a descriptive interpretation of our results, the beta coefficients in a given equation lend themselves to direct comparison since their magnitudes are not affected by the unit of measurement of the original variable. eri* nx CP3r.eT:9l9*q -n-'mwenoo lo noiBBa^qx© exij ni brmol ed oi el v.JiXr At . 06 86 babrxei^i e-d vSiTi 8: I'va.; ~di mo': .a^Dsif^ . , fens Jiinl axjol : • eviifilei erii §■ oxvtn-^'tr- Bcif ^ztluzi ■ • T^rrttnessiq nJ .1 elds', . iM-oix- +, ijnnol esT.i. - - ' ost^si snJ:*n?3?nq "Jo beJnobs ; ^iXxosl OS bcineaein otb s^neiolHsoo ' • . *c.ri* ^e^i£x.i6v in:. sdi 'i ■ -fxo esU lo ssn^r •^oxiq {Jilaap iiml sis s^Xdsx^ex ' rx rm) .fa yXXftrict.- • vt7oib«i ^y. Aibt ru>si^l^-i s>vx . , • - ■ • . ■ "»>'^-.'^ rjo*" bns "^nsXXsdx©" as foebtoost . isvs srl* cio ,Jcd* i?vo ijmyxmsnq babnemmco soiTx- TABLE 1 Regression Coefficients and Supplementary Measures, Thompson Seedless Seasonal Equations, 1951a/ (X dependent) n = 2k6 ^ Constant Independent variables Equation term h ^6 ^8 ^10 i ^1 1 i2_ 2 R R 6 0.184 0.096 0.354 0.208 1 ( 1 0.982 D 1.271 0.611 0.570 1.970 1.160 .5486 .7343 v% t D 0.30 06 9.0? p .00 t If. 17 4.10 2.06 5 Q4 ^ 7ft P 0.219 U . 3'+i 0 .224 0.111 -O.082 b 1.280 O.o2o 0.625 1.897 1.249 0.172 .5607 .7414 6 ka 3.13 3.12 9.48 6.24 0 86 t 4.25 4.25 2.28 5.76 4.09 2.57 P 0.237 0.161 0 107 ^ • U.9U -0.009 3 0.636 b 1.391 0.535 0.635 1.963 1.212 0.138 -0.081 .5670 .7445 b' 6.96 2.68 3.18 9.82 6.06 0.69 -0.4o _t ^.55 3.46 2.33 5.96 3.98 2.00 1.87 0.226 0.186 0.097 0.346 0.215 -0.072 k 1.695 b 1.326 1 0.620 0.579 1.924 1.196 -0.211 .5537 .7365 1 b' 6.63 3.10 2.90 I 9.62 5.98 -1.06 1 _t [ h.3k 1 4.17 2.10 I 5.80 3.90 1.65 1 Identified in Exhibit 2, pp. 12-17. Approximate t-ratio at 5-per cent level of significance is I.96. t\3 1 • i 1 i "1 ! MM*'- - i ■ *i ^ *i — - - 1 1 ^ ; i j 1 ! 1 I i 22. interpretation of the beriy size relation but, in the case of and X^, this in- terpretation may not be so cleaj: cut. This is particularly true in the case of X^. As was indicated in the previous section, there remains considerable ambi- guity concerning the specific fruit and pack characteristics or combination of characteristics which served as the basis for the classification of a lot of fruit vith respect to X^. In part, the classification with respect to fruit quality was based on some weighted combination of the other characteristics ap- pearing on the face of the schedule.-^ The significance of X^ in these equations, however, suggests that additional fruit or pack characteristics are reflected in Xj which are not recorded specifically on the schedule. It is not difficult to conceive of specific additional factors which could have entered in this determination. Within-lug uniformity of factors, such as berry size and color, is not reflected by the observations of individual charac- teristics and could be reflected in this variable. In addition, there are such attributes as general condition and general appearance of the fruit which are not called for explicitly on the schedule. These attributes may reflect maturity at time of picking, lag between time of packing and sale, and treatment in ship- ment, giving rise to bruising, decay, or overripeness—conditions T*iich may seri- ously affect the keeping qualities of the fruit. More useful results from the standpoint of appraisal of shipping practices might well have been forthcoming from identifying these individual characteristics and observing each separately. As has been indicated, however, the fruit and pack characteristics more directly identifiable vdth cultural practices at the grower level and packaging processes at the packer level were of primary interest in this study. The ambiguity with respect to the stem condition variable is also -worth noting. The question here concerns what specific undesirable attribute is re- flected by "pliable" stems. The degree of shatter would be expected to be higher with "pliable" than with "fresh" stems. On the other hand, it may be a combina- tion of fruit characteristics typically associated with "pliable" stems which is significant here. This set of traits might serve, for example, to forewarn the buyer of inferior keeping qualities of the fruit. In addition to fruit quality, size, and stem condition, other variables ap- pear in Table 1 to be significantly related to percentage price. Notable among 8/ The multiple correlation with X„ dependent and X,, X, , X^, and X, inde- pendent is 0.67. ^ ^ ^ 5 6 rl. ■ erii ni ^iud nt. . «ri+ ^!^ ^u- trnx-^^sq ei f uo tsets ca e<< ton w 1 ei^^ri J . ^ at'oi^raiq ©s: ' sew bA '-.tcfsni?; i;ie^o6i6ffo :)I06q bns i-itnO: ohioeqa art* 3nf oifloa rro 098jBd as?? y.d"i. .o/jb sxlos ;^riJ no -jlXsoili nooet ion srrft riaitfw -trs '.siG^jsl "icv x^iaiic - I-nixiiiV? '.r, axffa ni b5»t8;; •.^Tfbai lo anoi J-Bvt&e.+ 59X^97 don ex < ?oioo bns esis Y^scf ,.,=.1* bnr ' bfie gnxJinR'i '^r- '--^rt nsswc tanhioiq 10 saixi is - " r'':>.:fi'7 snoxixfaiiOD — eeshorrHis-, .-d ^gnieiir'id esxy gnjyx3 t tne.-?? /•s-i Xi/l«*av 9ioM 0 39±*iX)3Xf'p gAxq???! ptI* ioell* • ]_ npsicf s.-/;;'! Li£'w ^rfgiff -i'^ "io Xa;. lo in. ;^ .-, • . . U . ' ' "eXdex' 4 23. these are mode of shipment (Xg) and brand (X^). Here, likewise, the direction of the relations seems reasonable although the magnitude of the effects, consid- ered net of the fruit characteristics included in the equations, is in each case somewhat larger than might be e:q)ected. These results indicate that, for a given combination of fruit quality, berry size, and stem condition, express shipments command a price premium over freight shipments. Similarly, the significant brand relation suggests a premium associated 7d.th brand name as such. Again, however, it is not clear precisely what is reflected by these relations. It is plausible to suppose that mode of shipment is related to condition of fruit on arrival at distant markets. Brand, also, may reflect a combination of desirable fruit char- acteristics not adequately reflected by the remaining variables introduced ex- plicitly. The correlation between each of these factors and fruit quality simports 9 / this supposition.- The small number and perhaps highly selective lugs in each lot that were opened for inspection may well have led to the situation in which the observations recorded have not reflected accurately the fruit and pack char- acteristics of the lots sold. Whatever the specific causes, the existence of significant relations between percentage price and these two factors in this year cannot be discarded. The conclusion that the net effects which appear for mode of shipment and brand have no foundation in the quality of the fruit and pack is perhaps premature, but this possibility cannot be excluded on the basis of these results. The uncertainty which remains with regard to the correct interpretation of the estimated relation between price premiums and these two variables empha- sizes the necessity for identifying and incorporating in the analysis the dis- tinguishing characteristics which make mode of shipment and brand the dominant variables in the "explanation" of price premiums. Aside from the attributes identifiable with individual lots of fruit, other measures reflecting more directly the market conditions on a given market day might well enter to influence price premiums. It is this reasoning which lies behind the introduction of selected "market variables" into our analysis of price premiums associated with individual lots. Daily average price level (X.^^^) such variable. The average level of price over lots on any given day may be taken to represent, in a sense, the strength of the market on that day. Introduction 9/ The simple correlation coefficients of fruit quality with mode of shipment and brand are, respectively, 0.5l and 0.43. noUss'iife prfl v9®iv;9>Jxi ^eioH ..(^X) ba^^d bf\e. (gX) in&mqrria iLo sbon &ib ©a-.d* &$8o does ffjc 3i: ,&nciU&i}p9 srf^ nx tebyXon^ soxiaiisdroEOfiriD ilinl adi 'io ien bs-ia .-Tsvs^ori t«iK^iA .dou3 86 &ma ba&td rf*i?r ba^elooE-i.^ auxme'rq £ '•■liOs^y^s noiJ-Eaei gidxei .aoiS&lT>t QZodS id bBiaaLle-i ex .iaiiTT v;i^8i:ooiq lesXo cToa ai it i& Isvii-tfi no d-xiril to ttoltibaoo oJ bsdsX©-? ai ^nsmqiria Ic; aboai *Bfii saoqqffs -isrfo .Jiuil ^Xd^ixasb lo noiisfiMfiCo b do9l"X3'X vgra iCbXg ,bnnia .sdsjIiBin dns^sib -xe linx aelifiiiBV snvtoisinei arijr, -^jd bsd-oeXlei ici^^^si/psbs *i'>f5 aoxdsxied'Ofi sttocfoua xJ-xIsi-'p ^xtntl bns 3^o.fosl saerfJ- lo rises n^swj-ecf noiis.i&itoo sriT .^X^xollq rri: p.siul ©viJ-osIsa ,, qsrfisq bne isdmun iJ-ftins ariT -.noxcMeoqqua sxnd !i ^cci-fixfdxa srict ocf b?i svari XXaw \33ffi nox^r)9qanJ: lol beflsqo sisw .+srt* ioX lo &oaej::J.X9 edi ^ceavB^ oiliosqa sdi ^ovsisdW .bloa aloX etii lo BoiS^in^ics T;3y a-rrf^-^ nc s-toiael owi sBPri;)- fane eotzq sgiBinso-isq naewi-sd acoiiJBXs'c ^nsoHirtaxs risxrivf aioaVi© ^sn srii iedi uol-^^a .otioo sriT .bsbisoalb ad ioanco si >;3Rq bns ^xi/tl erf* lo y*xXr.i/p eri;f fli noi&Rbnvol on ©vsd bns-xd bnB ^nsmqxria 'lo 4?3ahJ "^o aj-asd sAi no bsbylox* ^d ^onnBO yctJiidisaoq ?J;il eic.T: gnirfoeXlsi as-xx/acsic aeil rioirl-ff fjninoao Ji exri* ax JI .a.Bwiaia'xq etijt'iq ^'onsuViai oi •seim ilc-tyr :: ■?r;i:'rq lo si-eYXsns ix'o ctni' "BsXdBiif^' *'?'v(-"^.-r!" fcSvtnf^Xps lo rsoxioi/b'^iJr?i «>d* hfJXii*'? -• r>.£ ( .X) XdV^l ctiltq issBtstB ■ ■-. . ribnX rl^xvr b- • -.rrrrsii 5 Y«a ^£sb nsTxs -vins M e*oX i9"vo sfoJnq lo XsreX I'gsiavft exlT .oXdsiisv iin^rs ,^4»0 bar, X^.O ,yi9Yit+P$q2*''i: ,£»t2- r .. 2h. of this variable explicitly is to recognize the possibility that percentage pre- miums for preferred combinations of fruit characteristics may be different in magnitude under circumstances which give rise to a strong market as opposed to conditions giving rise to a weak market. The coefficient of 1^^, when added to equation 1, was negative but was not statistically significant. The conclusion would seem to follow from the lack of significance that price premiums (and dis- counts) remained fairly uniform percentagewise over the range of price variation represented by our 19^1 Thompson data. Daily volume of sales (X^q) and daily per cent of lots rated qu^ity 2 (1^^) represent the other "market variables." Volume of sales reflects the supply of fruit on the market. The premiums paid for a given combination of fruit charac- teristics might well be different on days of heavy volume and on days of light volume. Equation 2 in Table 1 reflects the effect of the introduction of this variable. The significant positive coefficient of X^^ in this equation suggests that an increase in volume of sales was associated with an increase in percent- age premiums for a given combination of the remaining characteristics incorpo- rated in the equation. It likewise indicates that percentage discounts are lower on heavy volume days. Due to the significant negative relation between volume of sales and daily level of price (-0.U8), the possibility must be recog- nized that the relation of premiums to volume is, to a certain extent, a reflec- tion of the premium relation with price level. With the introduction of price level in equation 3, the coefficient of X^^ remains relatively stable, but sta- tistical significance of X^q is not established in the formulation including X^^* Daily per cent of lots rated quality 2 (X^^, ) has been added in equation 3 to reflect the daily composition with respect to the qualitative characteristics of fruit sold on the auction market. The choice of this measure for this purpose is not considered completely appropriate in view of the uncertainty as to the combination of fruit characteristics reflected. On the other hand, it would ap- pear to be the most appropriate single variable available to us reflecting "desir- able" combinations of fruit characteristics associated with lots. Although the negative relation between X^^^ and percentage premium would appear reasonable, it is noted that our results also indicate that the relation is not statistically significant. The possibility of joint relations betvreen selected individual characteris- tics and a variable reflecting strength of the market was also given explicit recognition in our analysis. This allows for the possibility that net effect on nl ;tn3i!?llib pd 'f^m 9 5^5 iio^oeierfc Ji^ - ' ^aio iiBul'iaioo bo awxia ■ gnoii-i fi o,t sail avJig /iairfw tio ii^x.^u; s?i: norajjiofi -^^^ svl^ta^^ci 3; iC^xXairp fcsJst eJ'oI lo */T«3 'fls (qjX) aeisa to u .fiinl to ttoi*sni:dfno5 ffsvia 6 -lol jjxsi emt'tusiq .?xiT ..iorfism eri* no ^loTi . eysfo no baB ^aiitlcsv yvfisrl lo 3YJ5b - ' aoWaiiai ,; lo nox,t;}irboT*H.!: arf* 1 rii aioeil*! i sidsT fix ^ noxj ■ ' .eiaulov SIB B^at'oop.ib ©gB^neo-rsq isri^ aa-taoibni; 5Bxws:*t: il .nol^st/ptv srii al b&)n': ,+sy.a xjr ,(8ii.O-) ^i^i-f^l i'' i -'Isb bos aaXfia ar: ■ Lxi: babbB n£»9.f} z&ii ( , J,) "5 X-i'^Isi/p bears i ^-^r I lo insa tSK-; iO^ attrspsffi 3iciv+ 1<> s^if^nly s^^: .^^jitem noi^sjt"^ ecid n-o bXc« ^itrfl OCT 3S Y^ai:5#-ts.om; srf* ^^o n^tv nt e^^^iiiqc'sqc^ ^Xs^sXqr cc b»-i&tisrfo;> .ton ei -rr. oXxfOfl- *± tbnsrf -xadc^o ©lii n.0 .bsiooX'isT ajsji^aJne^MidAn liirzl lo no-i- ? , -ofc" gni^s^^Ilai aw o4- sir^RLlnvs- eXdsxisv 8l§nx.T eisJ-iqc^q-Ti iaonr f*ti& ad o.t ispq arii riguoilvt i .\ .sJoI rf^^r.T' beiBirnoeafi 8s-i^ei:t9*w?TX?rf5 ihs-iX lo ano^isralcffflos "olds T.e-i tft^eqB h/jtrcw cmf..tn»iq ass^fits^D^aq bns ^^.X f-siO'v.fenf :icij- lei sTrxJ-ii^f • i--, ..m ? s»riJ '■'{id bir . I3i ttx. ■ iO 1 »rii ai aoXcfpiitjv o.+siBqss ae aflfxe^ Soubcnc ... • j ' ■•V* ^iti.f T<> ic&'i^s bshfc/s 9tii lo ^Jiijss^ra s^^rSxnTrr r> . . . xngis ion oftvr iciQloli'iooo '■rf': -OSS': l?rf* V ■'c'f.aaoq © i^ bdaingo^s'i J'ja eEci isl euri* ■. . ^ ■ aorfixW ,9ao Ic: ^ »f:.'OYr YJ'-c-£.f3j:s3'5q ziriT ►Jlesvr eri* 1?> -.T'o t,r,^f f.-* LO«q s Xtfla^r^^sfl" - ■ ^Icfi^coqroi • ■ -ioti ^stfi , .. ^ ;. . . . ... , , ^.TO'i.^..? ■? -tn ^. •■ v'-'W dili to sixeh ■ -iB^i fc"'-"-' -'- ^^ aYsb t!.. . '-^-r-, >f-foT wpK ;t e>cii lo " - . /b .••■» ■ ' •••rf -^m s>Js^' --i^ rt'c fyJ'^I ■'.doiisric - , /:dfrt-^- ..'^v '••'■*■ '■ .-^-fvr ' r-'.' iXq ms&s^, bli/ow rioi-rfw Tfoil.! -u.. , gniqesjt oJ' idgisvr tsirvser'. . 26. placed in the retail display bin on heavy shopping days moves relatively more quickly than fruit displayed on other days. To the extent that there exists some uniformity in time patterns of shopping habits in the area served by the New York auction market, this would be expected to be reflected in the charac- ter of the demand for fresh fruit on this market. Other factors traceable to consumer behavior in the retail market may exist which, in turn, find expression in the behavior of auction buyers on different days of the week. Two formulations which give explicit recognition to day of the week were considered in this study: (1) the introduction of day of the week as a variable in the seasonal analysis and (2) separate analyses for each dsy of the week. The first formulation involves assigning numerical values to different days of the week. One possible technique, of course, is to assign the numbers, one through five, to the days, Monday through Friday, consecutively. This procedure is obviously extremely arbitrary since there is no basis for assuming that the relationship between days of the yreek is reflected by either the order or the magnitudes thus assigned. Recognizing the arbitrariness of this procedure, day of the week was nevertheless introduced into the seasonal analysis in an effort to detect systematic shifts associated with this variable where they existed. Whether or not day of the week proved significant, the data were subjected to separate daily analyses, and main reliance was placed on the separate analyses in exploring the possible changes in patterns of relations associated with day of the week. The reduction in degrees of freedom in the daily analyses made for somewhat larger standard errors of the coefficients. Although the loss of pre- cision could have been partially offset by augmenting the daily samples in the Thompson Seedless analyses, this procedure was not adopted. Day of the week, introduced as a separate variable in the seasonal analysis of 1951 Thompson Seedless, did not prove significant. This suggests only that the systematic linear effect specified by our model is not supported by our data. Selected equations from the separate daily analyses are summarized in Table 2. The equations presented include only those variables which were significant in the seasonal results. The variables Xy X^, and X^^ proved not to be significant in the daily analyses as was the case in the seasonal analysis. The results by day of the week lend themselves easily to specific compari- sons by the reader. The differences which appear in the coefficients of a given variable on different days of the week may be considered suggestive, although the interpretation of these differences is difficult vlthout more information about the market served by the New York auction. It is appropriate to point out also C o^ -^''--'i^-^i> asfi'Iew Isorisno/n gnxna^sae asvlovr.: a-- r !r .r,-* /^ot''^ lo •• ■•>•»' ■-. 3fle^*si9'x lo- 8C"ii>.t '-.4 • - rfBi-io. eldteacq ari+ •t't jx.o f,i 3--*:q^. arivf 8it.^cr«s!>iin8tf.»- V.* •■f^i^asq n?*ad srreri bii-o.-. ■ ■' .b$i.iqob& ion 'j^.v.. . , ,^daY..[€»s»S nosq. 10 3d-S93§0C. a-xri? .vJnsn 131113^2 ■ '?vxyu{ ionjMfc)- ,;?a5lhe?3 nosqiiTOriX -i s. iat z m o& ion bsvoiq j^^X bm t^X - ^' ieiiav .^SLubsi lenov . - i^' D-,?."v£:Jrte lencessa- arid- - v , aM^lBiift- xX-x«B 9tii r." n n^ilxoaqa o>l ^XXas?* asvXssiKnti iiat -i .-. ' ' '"•■'^d aiXi-as-i sriT ^ao'i.B lat s>ipBi ' .'.•-•^^ii l;^-.:;' . • ■ ' lo nc r • ' ' ■ ' Jifo Jii^oq ©.tain -iX Js.- .nof#3i;E Vvvv.--c'-..J frpv^ti's : ^ ■. e.?* TABLE 2 Regression Coefficients and Supplementary Measures, Thompson Seedless Daxly Equations, 19512/ (X, dependent) Constant term Independent variables Equation n ^2 \ "6 ^8 ! X 9 X 10 R t-_ — . Monday 1 P 0.584 O.OlfO 0.028 0.217 0.019 1 2.^^78 b b' t 3.^32 17.16 3.03 0.134 0.67 0.36 0.183 0.92 0.21 1.214 6.07 1.35 0.105 0.52 0.14 .5424 P 0.1^63 0.C82 0.051 0.294 0.055 0.311 2 h5 -1.07^ b 2.718 0.275 0.332 1.648 0.306 0.418 .6282 .75^ b' 13.59 1.38 1.66 8.24 1.53 0.21 2.56 0.81 o.4i 1.98 0.44 2.96 Tuesday P 0.336 0.226 - ■ - 0.059 1 0.392 0.167 3 50 -0.329 b b' t 1.892 9.46 3.05 0.836 4.18 2.67 0.323 1.62 0.68 2.122 10.61 3.25 0.905 4.52 1.47 .7364 .8405 P 0.329 0.213 0.068 0.405 0.181 0.093 k 50 -1.223 b 1.852 0.788 0.373 2.188 0.980 0.159 .7444 .8418 b' 9.26 3.94 1.86 10.94 4.90 0.80 t 2.99 2.50 0.78 3.36 1.59 1.16 1 (Continued on next page.) 1 i' i 1 '1 , . ■ i * J- i ■ -J • • ! j j 0-J8T- — ^ 1 — 1 i ■ t ! - v-h ] i * -• ■ ' 1 i t , ^ 1 1 '■ \). rvj - i i t- - i - — 1^ i 1 . ! ! 1 ,:: 1 ;r 1 i ' ■ \- ■ 3 ■■■ - 1 ^ j > ' ' " i i j .! . .!trpTon i ^ .. '-...5 . ■ ^ ' 1. ^1.1.: .. B Table 2 continued. Constant Independent variables 1 X X X 2 Equation n term 2 \ "8 9 10 R R Wednesday ? 0,245 0.240 0.206 0.297 0.274 5 52 -1.424 b 1.583 0.856 1.437 1.798 1.671 .6132 .7557 b' 7.92 4.28 7.18 8.99 8.36 t 2.36 2.27 1.84 2,53 2.47 P 0.263 0,219 0.191 0.251 0.316 0.120 6 52 -2.706 b 1.699 0.782 1.337 1.523 1.923 0.310 .6256 .7587 b' 8.50 3.91 6.68 7.62 9.62 1.55 t 2.52 2.06 1.71 2.05 2.73 1.22 Thursday P 0.129 0.229 0.006 0.409 0.241 7 46 0.451 b 0.835 0.758 0.037 2.450 1.467 .5773 .7242 b' 4.18 3.79 0.18 12.25 7.34 t 1.05 1.90 0.05 2.65 1.71 P 0.116 0.267 0.037 0.419 0.231 -0.137 6 1.728 b 0.747 0.886 0.231 2.507 1.4o8 -0.323 .5940 .7291 b' 3.74 4.43 1.16 12.54 7.04 -1.62 t 0.94 2.17 0.32 2.73 1.65 1.27 (Continued on next peige.) CD i r i f 1 P i I 0-^ \ • • •"•——'t' f 0' i5 -..Ts:.iJt<.. j j'si. j.. j'?tn,9^.i -o*353. i . q:s]!,T._ j -o't3.' ... . j i : , ; 1 i .i 1 • . - >'03i . i ,, , ■ J ) 0-aes ...... .■ • '-^^^^^ 0-S5T; ..Q'TSC ■ [•- - 1 ... :.: : ' i 1 i p i ; O'SOP i. ! i 1 . ' j t j 0 j fej-in I • • s 7 ^ " 1 ! -^-^ , 1 ■;,cpT6 3 ccnr Ttr.rrsrr* Table 2 continued. Equation n Constant term. \ Independent variables R ^2 Fri ^6 day ^8 s 9 I i 53 I 1.881 1 b b' t 0.021 0.101 0.50 0.19 0.352 0.968 1^.84 3-60 0.278 1.366 6.83 2.82 0.226 1.072 5.36 1.74 T i 0.299 1.410 7.05 2.37 .5862 10 53 8 b t 1 0.020 0.095 0.48 0.17 0.350 0.965 j 4.82 1 3.36 i 0.278 1.365 6.82 2.79 0.228 1.080 5.4o 1.64 0.300 1.410 7.05 2.34 -0.004 -0.008 -0.04 0.04 .5862 .7297 a/ Variables identified m Exhibit 2, pp. 12-I7. Approximate t-ratio at 5-per cent level of significant, p hp r i ^ i 1 ■ ^'^^ 1 — . i i -0 Ot^ 'O'OOg 1 X^'^ 30. that With the exception of X,, the maximum differences in the daily coefficients would not appear to be statistically significant when the magnitudes of the standard errors are taken into account. Our daily results with respect to two variables might be noted in particu- lar. First, it would appear that the general appearance of fruit as reflected by X^ is significant in influencing percentage premiums only in the early part of the week. In fact, this variable seems to be the dominant factor on the Monday market. It is possible that the dominance of on Mond^ may reflect the effect of advancement in fruit maturity due to the added delays associated wxth the week-end market holiday. Second, stem condition (X.) appears to be a statistically significant variable only on Friday, although ?he influence of thxs variable on Wednesday cannot be ignored, it is possible that this Friday relation may also be attributable to the week-end market holiday. It seems a plausible assumption that "fresh" stems as opposed to "pliable" stems represent from among our variables the strongest indicator of keeping quality of the fruit Its significance on Friday may reflect an attempt by buyers in the auction mar- ket to protect themselves against losses through deterioration before the fruit reaches the retail outlet early in the following week. Returning to the seasonal analysis, a direct interpretation of the b' coeffi- cients of equation 3 (Table 1) may serve to clarify the nature of our results. The coefficient of X^, for example, indicates that, for a given combination of other characteristics represented in this equation, lots classified as quality 1 sold at a price which, on the average, was around 7 percentage points less than quality 2 fruit. Percentage points in this esse are with reference to daily av- erage price. That is, if quality 2 fruit sold at 103 per cent of the average price for the day, quality 1 fruit possessing the same combination of other characteristics sold at 96 per cent of the average price for the day. The b- coefficients of the other dichotomized variables are interpreted in the same way. With respect to berry size, weight meacurements enter directly into our 1951 equations. The percentage coefficient here indicates that, for given combina- tions of the remaining factors, a decrease of 0.3 gram in weight per berry was associated, on the average, with a decrease of 2.7 percentage points in price. Interpretation of the coefficients of X^^ and X^^ in terms of the original units of these variables can be easily made by referring to Exhibit 2. Extension of our interpretation to monetary units may also be accomplished by applying the percentage coefficients to different daily average price levels through the season. For purposes of illustration, it would seem appropriate to -tsm r dhjil eAt vAlis^L ,^aiq993l -io*s3ibni ^a^snoiie erf 5 ^^^'^'^ , ggg; • ■ -• hauo^^ ?x{cP «Vt^ r^-thk^^ s^ttq s cTs bioa ■ ■'; 9ni38 £^iiJ ■ ■ .soJ'iq nx . .tr»3T0q V. a Ail : ■ -ra no ,be»8±o<«8« 31. use the average price over days covered by our study as the basis for our calcu- lations. The average price for the period covered by our 19^1 study was $3.70 per lug. A straightforward application of the above-indicated percentage pre- mium associated with fruit quality to this price base indicates a price premium of around 26 cents for quality 2 over quality 1 fruit. The absolute price pre- miums corresponding to the coefficients of equation 3 in the 1951 analysis and each of the 19^2 analyses are shown in Table 3. In each case the dollar price premiums are based on the average price over dgys covered by ovir data. Obviously, the absolute price premium resulting from the linear percentage premium relation increases as the daily average price to which it is applied increases. Thompson Seedless, 19$2 . — As has been indicated, certain differences exist between the data which form the basis for the 1951 and 1952 analyses in Thompsons. The period observed in 1952 was about twice as long. Hence, the 1952 data in- clude observations of early season Thompsons from the Coachella and Imperial areas as well as the later season fruit marketed in volume from Fresno and Tulare. The fruit characteristics associated with marketing practices in different areas are thus more fully represented, and the amplitude of the variation in price level over the season is more marked in the latter year. With respect to information recorded, also, differences appear in the two years. Stem condition was not ob- served in 1952, and qualitative judgments were recorded on berry size rather than actual weight measurements. Taste was not recorded as such in 1952 while box condition and labeling were observed. As has been indicated, however, observa- tions on these latter factors were considered to provide little or no information for analysis so their inclusion or exclusion is of no particular consequence. With regard to procedure of analysis, it was possible to take into account position in sale in 1952. This was done in preliminary formulations which in- cluded position in sale as a separate variable. Lots sold in the first position were assigned the value 2 while other lots were assigned the value 1. This pro- cedure revealed no systematic relation between position in sale and percentage pronium. Other preliminary formulations excluded those lots occupying the first position from the analysis. This procedure resxilted in a slightly improved fit as measured by the multiple correlation coefficient. This suggests the possi- bility that price behavior of fruit sold early in the day reflected, to a certain extent, a process of "feeling out the market," and that the inclusion of these lots might tend to obscure the \inder lying relations between price premiums and fruit characteristics. Although the evidence provided by preliminary analyses was not considered conclusive on this point, the results here sxmmarieed are based .11- Itni^wustq -^sM'xq R aii-^s'-^ifani:- ?.38d s-i.»i5«,i* i3njr betilqqe ai ri' .'.oiife' &ol zv\ agfiTsv?i -.x/i£.b ©fi^-.as aeasei^.^ir ■^Cti '^^^d' Sci^i iMiv ^©C'fieH .aaoX as stox-ri c^>.'Qdt5 best .S??X Ai bsvicarfp boiteq sdT:- ^^■i&htT 'm& cir^aa-x'u moil sol-Iov nr ba^ejJism j.tx)"i^ nosfise 7?JtaX. ari* as .XXoif s& Sftstg . Xc.Y5.i ani-iq rjx 4'iU3ii3V-«rf* 'lo ©Ki/j iXqina arjj btis ib©cfn?aatq9't YiXXt/^ oiom ax/dd.o'is -fio .h)ft^ AOiiiki')3 in&-*3- 9rii nx iBeqqs ee^n8^9llib ,.oafs ^feoMos©*!;- eX/rlf? 5?^X xUVrfDifQ -es bamooe^ ^on eaw siegT. - e jnaiP^Ttvass^m- -trigiew Xsi,\ta^ -svp^atfo tt5^va'/»ori »bsii. n9=;«d tzH a>V. .bevreacf'S gn/Xftrffi.?.^ bns aoi:Jfib«a'i Koli^tnto'lffJ: .'iii it? eXivti: C- Qbxvmq Ov} b&tafjisrto-.-! e-is;" a^ctfoel- "X'?,td-BX aBsxii no an-scd' -rel'riolHw sn/rivrBXirono'l xiBnlt.Hb-rq aX »mh 33« axflT .$(l9.f. nt&Us nX aaX*i s,oq no f^iaacf jbiU sd* ni bXoa aioj ^eXdsi'isv gJeieq^e s ss ©Xsa «i noXJxaoq bebi/Io 'Oiq eiriT- .X wsXev. aric* ft©fi.3i«B8 eis*^- r.*oX feri.to. sX-b^r ^- firXpv. sri-J tif^nsisse s>i*5.*r ctgec^naot^'cf iMja &i"sB noi^-xcoq no5*-^^i?«3<^ «joX eeorfc? bsibyXi)'?!:^ snoxitsXitmiol; ■v;T:2r.ijnil!j»iq. ?<>.-^ia »i3iux:n->-xq HI bevo-TQitsiJ- YX-3%tIa c ni b?jl-?«HtH'^.s'nffe9oii>si^»'Wca ^Xqi.iXtmi ad* X'^ bexjssam ac bne- f.mi.rXvTtg'iq sajhiqi nssiw-tecf- arfo tJ-sio^, sniYX-icifc^jf? r.d.t • tyoado o.j <)na:^- tf'.^.ia zioi b?5scf 91.9 bsifxtBiKcjijg {)nr?r,i siXit/sc*'^ '^'rfcj (^r-'-foq- airij- no 9-«;AEy5[t>.TOo. bs-ssfaxsaoa *ofl esir Table 3 Net Regression Coefficients Converted to Dollar Units, Selected Seasonal Equations, Thompson Seedless and Tokaya/ Equation^/ Average price Independent variables X 2 X 3 X h X 5 X 6 X 8 X 9 1 10 X 11 X X j 12 1 13 Thompson Seedless — 1951 3 $3.70 0.258 0.099 0.118 0.363 0.224 0.026 -0.015 Thompson Seedless — 1952 3 $3.62 0.223 0.151 0.280 0.018 0.381 o.i+86 -0.045 -0.078 Tokay— 1952 3 $2.57 0.128 0.155 0.4U0 -0.019 0.033 a/ Variables identified in Exhibit 2, pp. 12-17. b/ Refers to equation number in original tables in which seasonal resiilts appear. I ! i ! 1 ! I t I i I i -0*1" To ^ ! -■■■■•1 r J T i I 'r 1 ■: 1 ■ '.! ' 0" ■: ■.t •1 . ■ .- 1 1 O'Ocid ! I j o'Tj-g 0'3e3 1 O'SSi^ _ .L _ i LJ or „ "pJ-TCf. j s 1 3 i ^ X 1 X '< ] i ro ■ ^ ; X ■ r -JT X 3«j6cpeg geseouffj Ecfnij^ioDs' .Tposabrjoa gcoqjees ffug iojcff2iff\ 33. en the set of observations which excludes those lots "which fell in the first position in sale. It remains questionable whether the results would have dif- fered materially had the first position been included. Selected equations from the 1952 seasonal analysis are summarized in Table li. A somewhat lower multiple correlation is characteristic of the equations fitted to the 19^2 data. Although the absence of precise measurement on an im- portant variable like berry size may offer a partial explanation for this differ- ence, the possibility that this lower correlation is attributable to the longer period observed cannot be discarded. It would not be surprising, for example, if the individual characteristics influencing price premiums entered our rela- tions to a different degree at different points in the season. Furthermore, as the period of observation is lengthened, the possibilities for the omission of significant variables in our framework may well multiply. Turning to the effects of the several factors observed, the similarities of the 1952 and 1951 results are perhaps more noteworthy than their differences. With respect to factors identifiable with lots, brand name (X^) and mode of ship- ment (Xg) are dominant, exhibiting highly significant effects on price premiums in each formulation. With regard to characteristics associated vdth the fruit as such, berry size (X^) appears again as the most significant factor, and fruit quality (X^) displays a significant effect consistently in each formxilation. The effects attributable to fruit characteristics in the 1952 results differ from 1951 primarily in that color (X^) and Balling-acid ratio (X^) appear significant or nearly significant in each of the formulations for the latter year. It is possible, due to the overlapping of marketings from mere areas of origin during the longer period observed in 1952, that auction bi^rers were confronted with a wider range of combinations of fruit characteristics on given d^s than was the case in the shorter period observed in 1951. If so, price premiums associated with selected factors might be expected to be reflected more strongly in the 1952 data. Color and Balling-acid ratio, both being associated with maturity of fruit, might well exhibit somewhat stronger effects under these circumstances. The pos- sibility cannot be overlooked, of course, that color in 1952 carries the role carried by stem condition in 1951. This supposition is not supported, however, by the rather low correlation between these two characteristics exhibited by the 1951 data (0.18). The signs of the coefficients of each of the above variables are consistent with those exhibited in the 1951 analysis and ?dth a priori expectations. View- ing the beta coefficients as approximate indications of relative importance, the iQiit sdi ni I-CdjI riairfw aJ'oX 03 cd* gssfoi'Ioxe rloirfw snf^Wsvt&s'fc d'sa arid- n*? -lib yvari falirow es+ii/^S'X s»f{* Tsri^ads? ©XnsHOlJseifr afJX''.c.p'T .?»Ie8 ni co^^iaoq WiU ns no iassr.^mzsetn esiocnq lo asnsads erid- ii%L")dilk .siab 55^X sri^ o* b&^il'). -lollib siJii io'i noii^nelqxs IsiJrtsq e i5»'i'io •ifim ssxa "VTi&d e^fll elcfeiiBV ct/jSvtic- 1 i9§ncl Slit at BldsJudii! iiR 3x noicff^snoo •isxral airi^ ^srf* ^^ii/xcfiscoq &ri* t&oa^ tolqnrs^ts -col tgrrxaxaqiua srf ion hLuaw il .b8bi£5f;,fb erf ^onnso b9vt??do boi-c&q -bXsi Ti/o bsfis&m nmui^neiq solnq sn^ansx/Xlni: 3oi;v+3iisJ"06i';d? icubtvtbni eAi J.^ 3ff ,9iO!irTorf*iu'? .no3esE ert# nx s.tnloq .tn^ielTt if' i-r. eorg^b imn'^Yi.j-.b e o* arioxd- lo noias ;mo srf^ 10 '2 ^iet&iiidisz'^q jb-^nerictsnsl -^x noiJ'i;'J"i?"5d<5 lo fortiisq arid .Beans'Xvt'i'ixb ixsiit* fjeri^* y^Sto-aoi^ci pioni sqaritsq 91e e^Xxraoi Z^^L bns S5^X 'io -qiAz 1(h about bns („X) ocisn bnsid tsJ^nX ri»+ir«- sXdsXll^nabx aiotoel i-09a??i d-tii'/ a.TS/xEdtq &Diiq no sio^xlf .tnsoxlxngxa Y^xlBirf a^ix^idxrixe tJ-nsnrcmb eis (^^X) cfn?'/- iiwil edi dtfxT bais i:305Sfi 33 Wai'jsd-SG'rorin o* .%7s83i dil'xW .noxJfiXxfrmol rial's? nr iXxiil dfiB s'lo^iosi ^npoi'iciTa xa cJ-aoirt erii 3b nxsgs s'SGeqqe (^jX) esia Y'^i-d ^linxfs ei? -^riT ,r?:-iir>X{/im.o?. rios? ni vl+ns^cianorj *3alla dneox'Ungxa s e'^aXqair. (^,X) xiXXiSxfc- ajGil -rgllxb ed-Iifae-i S59X ariJ" nx EnxiaxTa.tosisriD J cj-xl sXrfRctudiii^tf: aJoe'tls d-nrsaxlingig issqqa (^X) oiisi pios-snxXXsa baa: ajnil aaftxJ-si'SRffl lo anxqqsXi'.^vo otii ci ei/b jPldxaeoq 6 d^rs? bo^t^o1'}aoo siovr 3it>Yifd noxctD/Jc ^arii 9XPS rfix'-v t3-£0't'i lo xii'^vizr. d^xt? J>e^f:i::>'>';3B gnied ri^od ^ox^ei bit^s-gnxXXpQ bns loIoD .?.*jsb -ai^q .aaorteieau/oixo saedJ 'i?brw cioell? neanot-Js -tsdw^ffios cfidxdx? IXsw ^rigtra sXoi srii 89i'ne?> S^li'X n.t -loXoo J-sd.j- t?a-ti;o?> lo ,b9>!ooiT9Vo sd donnso x.+tXtdxa tisvovyod .bsiioqqira ion noXto9qx9 iioiiq b ri^xw bfiB sXaxXsns Ic?X srii nx bsJctldxdxs ssosii ricfrw TABLE k Regression Coefficients and Supplementary Measures, Thompson Seedless Seasonal Equatxcns, 19525:/ (X^^ dependent) Constant term X 2 X 3 J. X X 5 X 8 X 9 Y A 12 V" A 13 R IZjQ Ucl u X vJli P O.l'*? U .xUd. 0.25I* 0.335 • 1 -A793 b b' t 1.201 6.00 2.19 0.893 l*.lt6 2.08 1.775 8.88 3.1*8 0.1'*7 0.7I* 2.1*8 1.970 9.85 i*.56 2.723 13.62 6.23 .403** .D ( v^ 0' Jf¥- ^ .. ' .silt ■ -rO 33]t -0 - 1 ■ ^i,mH,ii"m .1 J ^ . ... - i.-i. jj jjij — ^.-^^ "■ • 1 p j'39j : •o:too • ^ V. D ** U i- ^ O'SiT •530 ,. 1 : i p— 10 T ..lo .» S'EO ' ' ' ..r* r; .( ' n * ! •♦"j XT '512 J3'05 -1 ! — -> .0-T33 o'qA(? •0-3S0 -o't3s ! i ■5 -09 1 - " 1 O'Ar " O He ■T '^^■^ i j'sor . o'ghl ; ' I,"ci>J . 1 1 »i v.- ; i 2*'3sr4:TO,i.T ^A^^iJ V - . ' •■'X ,; h .-V, — 1? Si/ 35. main difference in the results is to be found in the interchange of position of mode of shipment and brand in the two years. Extending the analysis to include variables reflecting market conditions as opposed to lot characteristics, it appears that both price level and day of the week are systematically related to percentage premiums in 1902. It will be recalled that neither of these variables was significant in 1951. The negative sign of the price level variable (X^g) suggests that, given a particular combi- nation of lot characteristics, the percentage premium is smaller (or the percent- age discount is larger), the higher the level of price prevailing on the market.—^ The remaining variable which appears significant is day of the week. The nega- tive sign of this coefficient suggests a ^stematic net decrease in percentage premiums as the week progresses. Stated directly in terms of percentage pre- miums, our results indicate that a lot of fruit exhibiting a given combination of the other factors commanded, on the average, a percentage premium arovuid 2 points less on Tuesday than on Monday and similarly for each successive day as the -week progressed. The arbitrariness involved in the introduction of day of the week in the vray in which it has been introduced here was recognized in con- nection with the 1951 analysis. Its significance in this formulation suggests the existence of a systematic relation with percentage premium. But a more thor- ough familiarity with marketing practices beyond the auction level might suggest a different relation than that exhibited by our analysis. Product terns of X^^^ "^^^ ^2* \* ^8* ^9 "^^^^ successively to equa- tion 3. In no case was there a significant additional effect associated with the product term. Selected results from the 1952 daily analyses are presented in Table 5. The equation presented for each day includes only variables identifiable with lots 11/ The proper interpretation of the significance of X-j^. is not clear. If there were no relation between lot prices (numerator of Xj^) and X-j^g* ^ negative relation between X^ and X^g would be expected since X^^ denominator of I.^. In this case, X^^ ^ linear function of prices over all lots on a given day. The negative relation with X^ may arise primarily from the relation of X-j^2 to the distribution of quality of lots on which X^g based. A correlation of 0,U2 between X^^ ^9 lends some sv5)port to this view. This suggests that some other base price for X^ might have been preferable. A daily price base urtiich adequately reflects seasonal variation but is not derived directly from lot prices on the corresponding day might serve this purpose somewhat better. Introduction of X-j^2» formulation, would not seem an objectionable procedure, but cau- tion must be exercised in interpreting the resulting coefficient. .5^ ^aoMifcnos. ,3-9?i;i58i gfi^tpsnst 35j:d6-4:iRV^ ?bttXbnl.o? E-to-viXfiae prfcf anjcbrtoixa ''-•^.iajitBSJ 9rfi no aniXiBATeTq ooitq I0 levsX ani '5e;i;gxrf sfi^-.f-iagtsX ai > •fnjuooalh .ligs -Bg-^n 9dT^ 4:^&dw Ici \6b zt iaaoi'lirigxs aiBeqqs-aairir-.oXcisx-XvW gniniBiRSi s-riT .93Eirr92isq nt jiss if. oab Jsn oiJ■5m^^^3\? r, Ed-asi^crs ^fleioxHaoo 'sixlrf- lo ngis ©vii .•;:a'xq.. asr.ina.o'xeq lo. aane^ .ni.-xXioeiXt) &9*£j8 .Beeasiao-jq jIsieF r-d^ sss a-itariinskiq \Gi? sviAUsoDcra. riose nol x+'i^ i^ii-i-tB itsbnoM no nsrU •ycEfccsi/T no esal'sinxcq q*a«g§,y3 nQ|.tj5lifnrrcl .eXji^ •ni.ap.acai'il^^ a^I tSia'fXR.'id Xcl^X sfrf* lUiw n^Xiaen -•j.orfi .©.lom tvS .jra;iK9iq. S'33^iisoT*q d[>*Xw noxcf-^XsT oXd-^giejaYa s lo aonsisixs !?flcf .•t53-35*^s irigxaf Xeva r noxicae adt bnoxsd ^ssoirfaeTiq gnXv+ainr^ai fid!-i:vr-x*i*JsXXxms"i rigac nJ- TgX^viftasDOiie bstfabu- ei9\T fens ,gX . , : d&ifv ^.^X-tdrem^^ ^ovb&iH aviisasA s ,^j.X fcas (j-X lo .io.^R't£«(Bttn) a&oXiq doi aaaw^sd nciJ-^Xe-x on sieTir eisrii . lo /io,t*ninon»b sriJ- aX ^vX apnxa bs^oaqxa ad bXaow ^p: bns ,.X nsaswtsd iioi^cXs-x rSf^vxa s no gJvW. XXx: 'levc^- ssoXrxq lo miiom'i issnlX s ax ,.z'\s^dbo 3xdi nl . ,.X <> ,X lo noi.tsXtj'i strfd- racsil-YlXifiarL-iq ?axas 'iani rfiXw noiii is'f. svXJ-sgoa sriT , 'lo noiie.ie%'iOo A .b9Eoi^flI .lajf-tsd iprl'-femc-j 9?C'q-JiK!.,8 tdj s^visa cWaXiis -Ysb aniijiTQqasi'iToo- eriij -ne . .■ '-r;^^^.?■^Tq eidsirtiioetdo. n-H.-fiTsea-d-on bfiT^'ST. ^noi*/>X'ofln ^ '"iO T 'bid 0 ■ If ' 0"50 3'd5t ; ; '91 i i i 1 • 1 i i L ! P u . ... } .-'^ j S'^ ■ r • • ■ . ■ /dd9 ■ 1 ! 1 'A5.TJ ; ! 1 ! T -1- it- ■y*fCT 5 'Ot, 10 -sa 5*059 T '80i -• -52 S'AT I'ods d'dd i 7'dc39 ; 'fl^ 1 i ■ . 1 1 ! ■ .* X." 1 >r' ..A A -1 , J (X q: ■•-) Table 5 continued. 1 Constant i Independent variables 1 I X ' \ X X 2 Equation 1 n term 1 2 1 3 \ 5 i 9 R ! R Thursday P 0.087 0.020 i 0.307 0.181 0.295 0.398 k 58 -3.827 b 0.600 0.157 2.079 0.166 1.875 2.698 .5155 .6772 b' 3.00 .78 10.40 0.83 9.38 13.49 t oM 0.20 1.81 1.70 2.79 3.66 Friday o.ik9 0.005 0.258 0.098 0.079 0.454 5 30 -3.803 b 1.251 0.01^5 2.059 0.110 0.633 3.802 .5726 .6790 b' 6.26 0.22 10.30 0.55 3.16 19.01 t 0.73 0.03 1.12 0.57 0.42 2.85 a/ Variables identified in Exhibit 2, pp. 12-17. Approximate t-ratios at 5-per cent level are: Monday, Wednesday, and Thursday, 2.02; Tuesday, 2.04; and Friday, 2.07. r •- ' J 1 _j ^jv. .. . „- „. 1 i i. 1 ■ i f I t - i ■ ! ; ! : 30' 1 1 ^. ! i f' K . , 1 ! 1 -r T' i i ! p _ f; t ■ ■.■O'J'L.y ' . 5;Gi."; >< • Jr-. ■ s i 1 ' • i . 1 : 1 T ■ ■" — r— . "".V. ;.v —-y" * 't 1 i f " --s • ■ ., • 1» i ' ' ' " r"^ o " ' j l" „ ■., 38. and is directly comparable Tdth the seasonal equation 1 in Table k* The vari- ables represented in the 1952 daily analyses differ from 1951 in that color (X^) and Balling-acid ratio (X^) are included in the latter year Trhile stem con- dition (Xg) and daily volume of sales (X^q) do not appear. The first two vari- ables do not appear in the 1951 daily analyses because neither appears significant in that year. The variable X^q has not been introduced into the 1952 daily analy- ses because of its apparent lack of significance in the seasonal analysis. A comparison of the 1951 and 1952 daily resvilts reveals something less than a high degree of consistency in these two years. Brand name (X^) is the only variable uniformly significant through the week in 1952. For mode of shipment (Xg), TPrhich appears as a highly significant variable on four days in 195l> sta- tistical significance is established in 1952 on only two days — Monday and Thiirs- day. With respect to fruit characteristics, berry size (X^)> which appears highly significant in four daily analyses in 1951, does not appear significant 12/ on any day in 1952. — Color (X^) and Balling-acid ratio (X^) each appear signi- ficant on a single day — Monday and Tuesday, respectively. The significance of fruit quality (X^) in the early part of the week suggested by the 1951 daily analyses is not supported by the 1952 results. Fruit quality is not significant on any day in the latter year. With regard to the direction of the relations which appear significant, the results in 1952, as in 195l» are in line with a priori expections. And, like 195l» the differences in the coefficients on different days of the week cannot be considered significant when the standard errors are taken into account. Tokay Grapes, 1952 . — Selected results of the Tokay seasonal analysis are summarized in Table 6. Only five variables appear significant in the seasonal results, and these variables remained uniformly significant in formulations which introduced the remaining variables into the analysis. Brand name (X^) appears clearly to dominate the factors influencing percentage premiums in Tokays. Mode of shipment (Xg), which appeared rather consistently in$)ortant in Thonqpsons, does not appear in the Tokays since the few lots of fruit shipped express were excluded from the analysis. Among the fruit characteristics, color (X^) and berry size (Xj^) are clearly significant. Other formulations introducing fruit 12/ It is noted that relatively high intercorrelations between berry size and fruit quality on each day in 1952 serve to increase the standard errors of the coefficients of these variables, thus, reducing the statistical significance in each case. -xtsV' owi i'a^r'l 9rit .iR^qq,-;- &cf. ob- (q^X) aelfia' iLd smi/IoV yXxgB bhe ('^1) ndiJ-ib &cL^'i"C-itt{^i:?i r'XGC'q<5(? ';.5dct-j:s»n ^guJi ii^cf aac-'^fT.rra -^jlxsb' it^L adi tit TiaioFBeB ior? ob ael'rfs "XXsifa- vliG'^ 'i'Si 9ri* o.^ni bKoi/bo'^ini ns^J son asrf ->rX e^IdsMcv oh''^ ,t&^v .01 *.^3£!iqM8 lo ih^m TO"? ni Ji'ssVeri* rigwotrtd jftsoitxnsxa ^5Xlirfo'^^m; ©XtfeiisV -)si5 .'li^^I ftJ: 8-\^eb' fyol no eJcfaiisv ^nsiiiHii'^xB "VlXrlgirf b cb aie^qqa dsirf-r t(gX) ^aiu/iT bu« Tj8brrdM-~3i{flb ospi ylrt'o n"y S2QX ni bs>rfsiIds*Bp si edngcixlingis I.n.oii&iS rfiiteoi'iiflSia •^•i^kiqqG .ton asob tIc!l'X nx saa'^ienB T^Xisb -xuoi til JitaoiixJxiia x^fiaxii -xngcg -icsqqs iio^4& (^.X) ox;fc*l fai6s-S«i-^Xb5 bns (^X> loXoO ^Sl.^tjj^/ ^b -^jiis £}o lo ^sncaJblxngxa r-ri'? .•yleyx.-toj^qap^ .-^sbsstfl," Y'SbrfciiS— ei%cda s fto ^riaoxl •^LtfiS 15? X «irii -^d b©*«3i5aU8 5i9:?tT oliS lo Jiaq %1^9 s.ij ni: (-X) -iiilBSsp ^hJti cfftsslliffsxa ls>n ?,£ x-i'j'Xr.t'p d-xiriU .aiXoAsi- S59X sri^ ^cf beiioqque J'on ai ssaYl^ns «d'f t^risoilxnsts i^eqqs doinvj-' anoiis-Xfi eAs id nox^ofi^iifi sri.* oi fcis§'$>t iiiiW •SiUi ;bnA .•e'noxd't>5>qxe i^oiiq e d^ifr ?irt£X ni ai's tl5!A^ ois siStis biabMiQ ^riJ nariw iosoiliftSt^ fcfitabxenoo t-rf Jjjridgfloa 6rl« ox c^nTo'ii.fnaxB -iaeqqs e^Xdexi^v evil tXrtC eXrfsT ni besx-isniira/fe' clixiiw moitniUm-to'l. ai icisirilagiz ^raio'lxiv b&nxsmsi a^XS^/'ieV f^ssr^t fcns tSiii/ss'l ateeqctsr (^-X) c»fften basid .-aia^/fgns 'ifif o#ni ssXcfeiisv ^ninxsiDoi' srid- b3oi/bQ7*a^- dbcM ;e-\^r:}{ot ox asi/.-rsss-iq s^jaJiisoi&q gnxonoirXini sio-Joisl ©ri^ ©J^niraob cri ylTifcffXr* tanosqiadriT ni Jnsi'icoqmi ■'{.Cindisiafiooi lediai bsiesqqs rfoxriw i^gX) insurqirie Id eisw sBsiqks bsqqjrjia iitnl io aj^oX wsi srii 5bnx3 8Ys>loT edi rix issqqf i6n ^3Qb bnc (^5) ^cloij • 3".es " ■ " ' b- - ^0'3A :■ -0-013 ■ •0 "OTg i—J < ir^-r i 1 ■j ! i i I i -S'iK'" r 1 1 • ■3-er ^■>•3T••i ! 0'c)cft j T'SOQ , Ti'Jf • I ! ^->-' ' -j -li e '- 5 ! ii . . .1 • ■ ■ rios 3'30J i ' ■ O'TOT : ■ •J35r i 1 * ■ ■ ' •, 1 p 1 . ■ 0".93" O'Jt.,- S'jr - , ..... ■ ■ :rST' ■ , — ^ — 1 , , ; -u3J^ ■ 1.1 . — r ov . , X'"- ■ ^- ! a X ■ ! X ' X 1 V V*- I — rvc-rr-'— — Xc>rg'X ^sa'sonS'j ]£'3w4! '^f^'-'?* JcSS^* ho. quality (X^), bunch description (X^), and Balling-acid ratio (X^) indicated that none of these variables were significantly related to price premiums in Tokays, With regard to the variables introduced to reflect market conditions, it will be observed that per cent quality 2 (X^^^) and day of week (X^^) are signi- ficant. The negative sign of the coefficient of X^^ suggests that, for a given COTibination of other factors, the price premium decreases with an increase in the proportion of high quality fruit on the market. The positive sign of the co- efficient of Xj^2 suggests a systematic increase in the percentage premium, given a particular combination of the remaining characteristics, as the week progresses. It will be recalled that the i^stematic relation Tdth day of week suggested by the 1952 Thompson analysis was in the opposite direction. Daily volume of sales (X^q) snd daily average price (X^2) "^^^^ ^ilso introduced into the Tok^ analysis, but no support was found for a significant relation with these variables. As in the Thompson analyses, product terms involving X^g with X^, X|^, and X^ were introduced successively into our Tokay analysis but gave no statistically signi- ficant results. Daily Tokay results are summarized in Table 7. The basic equations in the daily analyses contain those variables which proved significant in the seasonal analysis with two exceptions. Berry size (X^) was excluded on Monday and color (X^) on Friday. In each case, these variables have been excluded because vir- tually all of the observations with respect to these variables on the day in- volved have fallen in a single classification. This means that observations on these variables provided no information for analysis on the days indicated. The most noteworthy result observed in the Tokay daily analyses is the uni- formly dominant effect of brand name on price premiums throughout the week. As in other analyses where brand appears as a dominant variable, the interpretation of this relation is troublesome. It does not sean reasonable to suppose that brand preference is not somehow supported by favorable experience with fruit sold under the preferred label. If favorable experience is the explanation, it should be possible to discover and measure the factors which are responsible. With measurement of individual characteristics associated with preferred brands, more complete and certainly more useful results would be expected. Our results in Table 7 suggest that color and berry size do not enter sig- nificantly into price premiums in the early part of the week. As has been indi- cated, the Monday data provide no information on berry size, and color is not significant on this day. On Tuesday neither variable is significant. On Wednes- day color is significant while size is not. On Thursday both variables are .3-^:^oT at 80HriHr?'-q ..Dxi;i o-t bs^slsi yld-fffioilraglB siew esIdsiiBV eae/i^ lo snoti i:a ei£ (£(•>•) -^«^9'^ bos (jj^X) S -^JilEifp Jnso tsq tBcii bsvisscfo Iliw -oo 9ri(t lo ngis svxiiaoq ehT ♦^^iliem on'* no iltnl v^xUssp d^lii lo nox^t'xoqcnq s»riJ npvi:^ ^iittfjOTiq sg6*f!9ons^i ariJ- ai asssioni oi:i£ca3j-a-'£e s s^e^^^ifs ^j-X lo ^Jnvx^illf? .*,938f.n-^o'r4 >f99T(f 3rf* e* jaox^sxisJ-cBTsrlo sninxsasi »ri* lo noxd^sxiidmoa ifili»3xiisq R Y.d b'-t^aea^tfo Mssw lo yeb rtii'T noirisl?-! oiisni':^^'^ 9^^^ jsii:^ bsXiBDet sd Xlitr Jl 8-51*3 lo eiMf.Cov x-fJ'-'iCI .ftr>i*oeTxb scfxsoqqo srfi ni sbw alsYienfi noaqroofii' b5Cx silo- tSia-^If-rts ^Aol srii oini hsr^uboiiax oals eisw (jj-X) eoiiq &^&t&vs. \Li£.b bits, (q^X) .«!eIdfXiBv 980X!^ riirr rtoxJ-JsIi>'x j-nsoilinsxs s lol brujol aisw cf-coqqifa on ctod s-fs>«f bns ^^J rfixw ^jj-X aniVlovnx anna* iouboiq .ssCYiSnr, no aqooriT srfj- at .c*Xi;8ei Ji'isoil L^rij as anoxjejjp© oised ?fiT .\ old^T ni bssxisraaif/e 3-is 3*Xi/38i vsjIoT ^XxsG iGfioeeaa ni ctnGDiiiogi.^ bsvoiq risxrfw aeldsrisv eaorii nisdnoo 393YiB«s ^±sb -:o£oo bns icsbnoM no bafcxiloxo s.rr (^X) ©sie arfi no asIdTsxisv easfirf oi Jfs&qe9t xl*.a7 sfioiiBViasdo ori* 1 -i" no enoitsviesdo J'sri* easeoi zhlf .nol-tsoxliaaaXo aXgnxr. c ni hsXXbI svsii faevXov .be*-'.ofbnx a-sjBb ^di no aiexXsnB lol noLtecnDlni on bebivoiq aaXdRXisv saari* -itu' eri.t Bx 398xXfina vXxeI) tjs:!!©! eri* nJ: b^visedo ^tiuae'r ya^-rowB.ton i&om sd? .>{9'^w a»i* v.;fod3Uoir{i zmuboBiq ©oiiq no soisn bne'id lo ioella JngnxnKO \.Xiinol noiJj3*9 iqisJni: od-J j^XdRincv- tasnxmob s gi; atesqqB hnsid sisrfw aoa>iXsn& -i?ftJo ni. .Jarf^ SBoqqtra o.t aXdenossei iopea ion asoi 'inoa&XduoiJ- el aol(i.<^J-S't ^h'.i 'lo ixtnl ri*iw sons.tteqxs »f.d)3TL'»rGl Y.rt boJ-'irqqys woriscacc J'on si sons'islsiq bnsis hxisoAB St ,aox*J5nsXqx3 sri* si sowiteqxB eXds-iovBl II .XadsX baitciletq edi tebtw riJxW .^jXdii^aoqas-i ©is rioxdisr ortoiss'; sxij- 9'iU3£9m bns levosexb oi sXdx^aoq sd aiofs I'ibnsid fcgl9-iq riiiw bs^sxooaas aoij^xi^^JoB'Xsrfo XeubJCVxbx-iX lo cJnaiZCix/Bsain .£>t».jo9qX3 9d fcXiiOT a^li/aya jxil^tsu enom ylnxsjiso bns s^^9Xqf^^o -gi' *on ob osiz v^iie^ hns loXoo iarii' iesggtfB T sideT ni a^X^issi ttO -ibni rt9ed Brsd sA .ji^aw o-dr -to dx-sq \:Iie3 -^Hi nx aKtrineiq eoXiq odnx i£l^nBoxlxn .*oa 9X ToXoo bns tSsla YTi<»d n'> iroxiBfinolnx on sbtvoiq siab x^batM ^dS ^bst&G :7rnO .*n«oili;n% ia ax sXdexnev lari-t/sn Ycbeairt" nO .ycb axd;t no ^nsoilinsis fkiB aslf-j cifiv rijod •^faeixrxlT nO . Jon ax sisia aXxdir ^aKoxlxn^ia si xr^too Y^b TABLE 7 Regression Coefficients and Supplementary Measures, Tokay Daily Equations, 19525/ (X dependent) Equation n Constant term independent vari abies K 1 X 9 1 11 2 R R 1 30 1.479 b b' 0.313 1.375 6.88 1 1.15 0.091 0.365 1.82 0.46 i r 0.341 1.337 6.68 1.82 -0.225 -0.104 -0.52 0.98 .2846 .4125 i 2 1^9 -1.503 ... b b' t j 0.109 0.719 3.60 0.63 Tuesday 0.112 0.647 3.24 1.06 Wednesd 1 0.069 0.486 2.43 0.56 1 0.649 3.595 17.98 5.06 -0.145 -0.109 0.54 1.27 1 .6014 .7450 3 45 -3.168 b b' t 0.183 1.280 6.40 1.56 0.227 1.327 6.64 2.69 i ay 0.20 0.163 0.82 f 0.23 1 0.624 3.604 18.02 5.90 -0.032 -0.060 -0.30 0.39 .7721 .8619 (Continued on next page.) (COUi:im«ig on uexf Acrup 1 ■ '^'^(^ •3-53 - ' "1 i 0-85 1 T9*os : -0'?:. T'SSP T ■ iSi. -O'OQO •.U5T rq3 ■ 0-'5SS O'SO o"2o (VTTS "T — 5 1 i 0 ' : 1? J*85 T-33A i 0'3i*T X J XojW^ D»rjJv Tr;<=ir!yr fcas* Td?Si\ Table 7 continued. Constant Independent variables ! X X i ^ X 2 Equation n term 1 2 3 1 k 1 9 11 R R Thursday P 1 0.189 0.178 0.226 0.606 -0.333 k 55 -2.911 b 1.3^ 1.906 3.234 -0.273 .6776 .8030 b' 6.7k k.jk 9.53 16.17 -1.36 t 1.62 2.13 2.01 6.68 3.54 Friday 0.121 0.309 0.431 -0.321 5 19 1.471 b 0.512 1.724 1.886 -0.276 .4729 .5677 b' 2.56 8.62 9.43 -1.38 t 0.28 1.22 0.93 1.38 a/ Variables identified in Exhibit 2, pp. 12-17. Approximate t-ratxos at 5-per cent level are: Monday, 2.06; Tuesday, Wednesday, and Thursday, 2.02; and Friday, 214. f --r i i T ■ ; T"' i ! 1 t 1 ( 3 'A5Tt t 1 i i , ■' (» 1 1 ; S'OT .. _ j^. 1 ! i ^+ i i i ! 1 Uju.».. .i.^.'...,.— : .... — . f V > t • * significant. The Friday results are somen^at iroaker than those for the other days. No information is provided on color on this day while the small number of observations and the relatively high intercorrelation between size and brand combine to increase the standard errors of the coefficients of the latter vari- ables. Interpreting these results as they appear, the conclusion would seem to follow that color and berry size influence price premiums on the auction market only in the middle of the week. Although it is possible to visualize market situations into which these results would fit logically, more information is needed before it can be known what characterizes the market served by the New York auction. Per cent quality 2, significant in the seasonal analysis of Tokays, appears significant only on Thursday in the daily analyses. There appears no strong a priori basis for expecting such a result. Concluding Remarks The 'results of the analyses summarized in this report suggest some of the factors related to auction price premiums. Among the specific fruit characteris- tics considered, berry size appears clearly to be related to price premiums in both Thompsons and Tokays. The statistical significance of the relations with such factors as color, stem condition, and Balling-acid ratio in selected formu- lations suggests that the possible effect of these characteristics on price pre- miums cannot be dismissed. The combination of fruit characteristics represented by our variable labeled fruit quality appears significant in Thompson Seedless premiums. This suggests a net effect attributable to this variable, but clear- cut interpretation is not possible due to the absence of a clear specification of the fruit characteristics involved. The results with respect to other variables identifiable with individual lots are also suggestive. The importance of brand name in all three sets of data and of mode of shipment in Thompsons in both years is clearly evident. The significance of these variables suggests the existence of net effects on price premiums of brand name and mode of shipment as such. On the other hand, it is recognized that these net relations may reflect in part the effect of a combina- tion of frviit characteristics typicsilly associated vdth these two factors. If this is the case, a more complete specification of the associated fruit charac- teristics might well alter our results with respect to brand and mode of shipment. Selected "market variables" and day of the week were introduced in an at- tempt to explore more fully the relations of fruit characteristics with p'ice rtjcf^ff Us/us ■ 9fj.t eXlrJw- Ysfc e^riJ^no- 7.c;io»i ao bebivoiq aoxJ-gitnolftj. ot; .avsri, fen«nl bi?& £\.sj:J*;-np4j^5«f?d fK)i,jBl^TXooTe.^f(i rigid- -^Isvx^siei sdcf bns. spojlJ^y'iea.do lo 04.' cross MLKin'noiBjjIpa-^o . 9f!i ^^ssiiiq^ x®0* '''^ a-yXussi'j, s^'.srii.. an^tfs-iqna^rjl .a.sXcfA .tMipoi ftoijfoy.^ sriJ. at- amir^Taifi apjciq aonet'XlnJ: asie "fX'iscf briB 10X9.0 ^srii -.•roxLo'i, ^,9^&r,.5stic,v5iv of ©Xdiaaoq 3^ rfgi/ori^IA i*;!^. 1o sXbtim arid-; flx.\IfiQ si nox.!smii?5;- .moiislsi exLt 'lo sprtcox J.irg ia Xst'J:isx^aia9a9'iq;5'i aoXieitsJ'OB'^firiD' iix/i'i ic xioi^anidmoo sdT .b3.38xffl^i:fc ed io;inx>o snu/iin assfbesa noaqpMriX n.x ona.cu'tjLig is aissqqe V-iX^^i^P ^it^i^ bc»is?teX pldciisv. v-^o YY d-to.d ni annac^aioriT nx lii&mqine "io sbojn lo bfir; s«t/Bb saxiq no e^los'ils d-an Is *9rt&J'axx!3 arid' eJ-aag^i/s saXdsXiev easrid lo ©on^oxlxngxB ".i tx tbri£5rf *i9dio erid nO -nous ss insmqxrla lo sbom bps esen bms.ad 'io s^ntjlm'fq -s«j;-?(noo 5 Jpa^ls i^aq pi ioai^si xm: snoi.t.'s f 9'i dsn es&ri^ ^erii b^lSjx[sos^^ li .iioj'jsi Gwt 9£,ddi fiirv bgisiooaas yXXsoXq^i e3|d'?.i"X«'>Js^''t*J^^^ dxinl Ip aoid- - >.-76rio Jitnl bs^lfiiooesjs srid -lo notdsoxlxoeqa j^olqmoo ^lom s ^saso alJ si: eirld . lo eboiT. bns brijetd So&qzot ciiirff z4Jjja9t 700 latfls ILsyr M^isr 2oi?2XT:ei.d -J!? ■ aeo0faoiinx sisw Ms&w s.ij 'io Ysfc bae ^'s^ldel-iBV ^^s^is^n'.' .bad'osle..-; premiums. Our results in selected formulations suggest that these variables or related measvires merit attention in an analysis of this kind. The separate daily analyses summarized are considered purely exploratory in nature. More complete information about the characteristics of the market served by the New York auction would seem essential to a full appraisal of the daily results. The analyses summarized bear only on the "explanation" of price premiums. Clearly, the premium associated with a given combination of fruit and pack char- acteristics represents only a part of the information which the grower or packer needs for his economic decisions. Cost considerations are equally important. The "production" of berry size, for example, involves costs. Direct costs asso- ciated with thinning and girdling are involved as well as real costs measurable in terms of yield sacrificed in the process of producing larger berries. In his decision, the grower must compare his "e^qpected" net returns from a smaller volume of large berries with his "expected" returns from a larger volume of small berries. Similar considerations are clearly involved in connection with other fruit and pack characteristics where the "production" of premium characteristics involves additional costs to the grower or packer. A rational economic decision with regard to the production and marketing of fresh grapes likewise cannot dismiss seasonality in price variation. Berry color. Balling-acid ratio, and, to some extent, size are related to maturity of berry at time of harvest. It follows that one way to alter these characteris- tics is to alter time of harvest. Yet, it is clear that the economic decision with respect to time of harvest must involve a joint consideration of the "ex- pected" premiums associated with desirable qualitative characteristics and the "expected" price advantage associated with seasonal behavior of prices in the market. The resvilts which have been presented are viewed as suggesting some of the factors related to price premiums. The quantitative relations presented are recognized as being of limited direct applicability. Reference has been made to some of the reasons for this. In?)ortant among them is the absence of speci- fication and analysis of the more important separate characteristics apparently reflected by the fruit quality, mode of shipment, and brand classifications. With respect to fruit quality, for example, it is not clear just how the grower or packer proceeds to alter the characteristics of the fruit or pack to conform to the preferred quality classification. Similarly, it would appear hazardous, without further study, to view the coefficients of mode of shipment and brand as measures of "net" effects of these variables. 0TOM n-^ Ticker .rtlqKs* ^a^ew^ fc^TPibJ^snoo f>'u. bB^l't.'',smj.z tu^xlscis \Jli©h -03a»* Biaoa .tssixT .edaoo Qs»vfovnjc t^Xq^asxs lol iS>sia iftiacf lo '^rroliomboiq" sriT cil .39iii9d -legT^f artiowboiq sseooiq srf^ rti b-^'aixtoea blsiy to Qcnsi aX i9ilJ-o .'icfiw ftoiioannoo n.r bovXovni: xLiselo stf- anofietabXsftoo •X'-Iiral3 .gsiTiaif .i9'Ao&Q -xo is-^voaa Qct X6/iox.txbbi3 B'^vXo rni. ga.£.j3Jh£n; brji miiovbi-.tse9'xq n&ed srsri riolrfvir zJ^Xuaei sriT <^-r3 b9i'iies9iq anoXisX&t ovx^Kd'iift'jyp £»ifT .zmimoiq 9oXiq a* b«»2j.qq6 ya^iib bsfisnil I0 §fT>9d be^xngoosT; -iooqa l-o 9cin93ds sidj si -nsnj ^roMs .tns.tioqml .eXrlct -rvi anoesai sd^t smoa o>i xJjnsrjaqqK ^isxiaii&Joaitsrfo giricqps ^nstioqgnX aion sd& TLo eia\tXsi-is bns noxJ-sDil .snoxJeaxliaasXa bap-id bae »i-fi9(sqi.i3 i** ?ihom ^Y-^ii^f^'^P J'iy'il sdd id bsiosXla's jBTolfJoc ocJ- jjoisq 10 ^iu'it 9ricf c:>cj8iisj-on8f{r ©ri* toils 3b9900'£q tejiauq to ,auf>b'sas6ii iseqae bXifow Ji .vX'ipXUnxS .r.oiHoi'Uz'daJz ^^xXeyp beTie'iaiq ert* o:> A related point concerns the question of measurement of the qualitative factors. If quantitative measurements of fruit and pack characteristics re- flecting the important qualitative factors could be developed, the quantita- tive estimates of relations with price premiums might prove more useful. Under this procedure, there would be no difficulty in identifying the characteristic observed. The measurement data also lend themselves readily to the analytical procedures employed in this study. Furthermore, the evaluation of costs of pro- ducing premiim fruit might be made somewhat more precise were measurable charac- teristics involved. On the other hand, without further study, it is by no means clear that quantitative measurements of such factors as color and stem condition can be developed which will serve well for analyzing the relations of these fac- tors to price premiums. It is clear, of course, that relations derived from a study of the New York auction may not apply to fruit sold in other auction markets or to fruit not mar- keted through auction channels. The prevailing opinion in the industry appears to be that the New York auction is a "quality conscious" market. Although our results would not suggest a high degree of quality consciousness in this market, the possibility remains that the relations with qualitative factors might be somewhat different in other markets. -iii i^n&L'p eHi .beqoXovob ecf Wxfoo s-ioiosl &»viisiiJ6ifp ^nrntjoq.ni: ©rii §n.xJ-s«»X!S: Isr-Ad-xfsnB '■■•i \(ihp'?~.. Bevloaatsri^ &xi^>I osis steb st/'ecjsiiiestntf srfT .b^'VT.&eno -otitfirio sIcfiSW/ajBtj-H 3i6'v e3j:o!?tq Jerft^enioa abac? iiigxrj iiutl cK/i^isi j grjloi^t Dnasai on xqqs yi^strrjix sfl* nX noxnxqc gnirXf.'sve'sc . 3X&.-.iiB!lo miiovs d'giJKruif b^iff^ •I'jo .iatrcrJjUA ♦ts^i'xjyr '^f.voiosfroo -V/^ilsyp" £ ax aoiioua H-ioX wept {kI^ ?(i