Risk and Equity in Agricultural Cooperatives

Articl,es Risk and Equity in Agpicultural Cooperatives Clou.tdlo.Parlitlarnrlnt ud, Zai Lg.mtgrtt Thts rescarch exarnlnes the elfect of rlsk on the proportton of equtty held by agrtcultural coopcratlvE. Thc measured components of rtsk are bustneg rlsk and the flnanctal rtsk that ts dependent on the proportton of debt ln the coopcratlve's capltal structure. The emptrlcal rcsults tndlcatc the proportlon of equlty ls lnrrcrsety related to flnancld rlsk and posltlrrcly related to bustness rlsk. Thesc rlsk effects are cstlmatcd to dtffcr bascd on the commodity handl€d by the cooperathrc. No sfnlncant relatlon bctvecn the proportlon of equlty and whether or not the ctx)Per' attv! operatcs on a pooltng basls ls estlmated. Banks, lenders, and financld managers look at the proportton of equlty ln the balance sheet ln order to obtaln lnformatlon about a firm's soundness and solvency. A firm wtth low equlty capltal may have dtfficulty borrowlng ln order to grow and expand. Althougfir lt ls posstble to establlsh a firm fin-anced totally by equtty, tt ts hardly feaslble to finance a firm entlrely wlth borrowed capltal. Equlty capltal thus plays an lmportant role ln the lnvestment and growth strategles of both lnvestor-owned firms and cooperatlves. Equlty capttal provtdes a cushlon or a buffer that can absorb swlngs ln earnlngs. Wlth llttle equlty, a firm faclng a large loss as a result of fluctuattng earnlngs performance may be unable to meet lts obllgatlo_nsand may ultlmately be forced tnto bankmptcy and dlssolutlon. If a firm holds a sufficlent amount of equlty, however, losses can be absorbed by equtty capltal and the firm can contlnue operatlng, although lts owners wlll be that much poorer. Thus, firms faced wlth hlgh vadablllty of earnlngs, or hlgh rlsk, are expected to malntatn a hlgher proportlon of equlty ln order to absorb extreme downswlngs ln performance (Bredey and Myers). Claudta Parltament ls assoclate proJessor, Department oJ Agrtcultural rrnd Applted Economlcs, UntuerstQtoJ Mlnnesota,St. Pcul. Zut Lcrman !s sentor lectttrer, Department oJ Agrtcultural Economtcs and. Management, Hebreut Untuerstty, Rehouot,Israel. The authors grateJullg acknowledge the asslstance oJ Dana Huseby wtth the deuelopmentoJthe databose and,preltmtnary analyses. The authors also urtsh to thaikYacou Tsur, Frcnk Smtth, ondtlruo anongmous revlewetsJorthelr helpJul comments. Thlsstudy ls partoJaBARD-JundedproJectJortheeualuatTonoJtheperJormance oJwperatlves.B.I{RD[s a U.S.llsroelBtnatlonolAgrlcultural ResearchandDeuel' opment Foundatlon. CoopERATroN 2 JoURNALoF AcRTcULTURAL 1993 Thls reasontng on the role of equlty ls usual$ derreloped tn finance theory ln appllcatlon to lnvestor-owned firms, but tt may be equally valld for the user-owned cooperatlves. Insofar as equlty capttal provldes a measure of protectton agatnst adverse buslness outcomes, the proportlon of equtty ls erpected to depend on the rtsk faced by a cooperatlve. Although tt may be accepted that equlty holdtngs vary among cooperattves, the effect of varlous rlsk-related factors on oooperatlve equlty has yet to be examtned. Unllke lnvestor-owned firms. cooperatlves do not necessarlly seek the obJecttve of maxlmlzlng the rate of return on equlty, nor ls the notlon of equtty ln a cooperatlve ldentlcal to that ln an lnvestor-owned firm. It ls therefore relevant to exaurlne to what o<tent the behavlor of equlty ln cooperattves ls slmllar to that ln lnvestor-owned firms. The purpose of thts research ts to determlne whether tradltlonal measures of rlsk affect the proportlon of equtty ln the capttal structure of agrlcultural cooperatlves. After a brtef revlew of the components of rlsk and the sources of equlty capltal for cooperatlves, the methodologl wtll be outllned, the data descrlbed, and the results summarlzed. Buslness and Flnancial Risk Rlsk can be defined ln the business context as the uncertainty of future outcomes that artses fromvarlatlons ln the economlc envlronment (Brealey and Myers). Rlsk ls alfected by a varlety of factors that tnclude demand varlabtllty, changes ln the efficlency of labor, changlng qualtty of managerlal declslons, the degree of operattng and financlal lwerage, lnterest rate fluctuatlons, weather, pest tnfestatlon, natural dlsasters, and policy or technologlcal changes. All these factors comblne to produce vartabllity ln earnlngs, whtch is an accepted measure of rlsk for firms. The portlon of rlsk that depends on the uncertatnty of operatlng lncome ls usually referred to as buslness rlsk (Brlgham and Gapenskl). Another rlsk component, financlal rlsk, depends on the frrm's financlal leverage, 1.e., the proportlon of equlty and debt ln the firm's capttal structure. The financtal expense assoclated wlth borrowlng tncreases the varlablltty of resldual earntngs beyond the vartablllty assoclated wlth the firm's buslness rtsk (Brealey and Myers). A nrm's overall or total rlsk, as represented by the obsened varlabtlity of earnlngs, ls thus the sum of the buslness rtsk, lndependent of the firm's financlal structure, and the financtal rlsk that is dependent on the firm's proportlon of debt and equlty (Brlgham and Gapenskl). Flrms factng dlfferent levels of bustness rlsk can malntaln the same total or overall risk by controlllng thelr fuianctal rlsk through the adjustment of thelr proportlons of equlty and debt. A htgher proportion of equlty translates into a lower financtal rlsk component of the firm's total rlsk. Thus, utllltles wtth relatlvely safe operatlon outcomes tend to borrow more and accept a hlgher furanclal rlsk tJran, say, pharmaceutlcal corporatlons, whosebuslness rlsk ls much hlgher. Rtsk may also be alfected by structural factors that can be expltcttly ldenttfied and examtned. One of these structural factors ls slze. It ts usually assumed that larger slze confers a measure of safety or stabllity to a firm Risk and Equity in CooperativeslParlbmmt and l*-rmon 3 (Brlgham and Gapenskt). Banks and other credltors may have greater conffdence ln the repayment capaclty of large firms, assumlng they are more dlverslftedand representlessof a credlt rtsk (Sporleder,Mallck, and Tough). Therefore large cooperatlvesmay be able to borrow proportlonately more than small cooperatlvesand functlon wtth a lower proportlon of equlty capttal. Equfty Capttal ln Cooperatives Cooperatlves, llke lnvestor-owned firms, obtaln equtty capltal through dtrect lnvestment by members and from retalned earnings (Cobta and Brewer). The lnttlal funds for startlng a cooperative are tradltlonally ralsed by dtrect contrlbutlon from members through the purchase of shares. Yet dlrect lnvestment generates the smallest pereentage of equtty among cooperatlves (Kane). Cooperatlves on the whole are unable to ralse equtty easlly through the sale of stock because the retums to cooperatlve owners are based on patronage, not investment (Schrader). As a result, there ts no market mechanlsm to ralse equity through the sale of stock on an ongolng basls (Staatz). The alternatlve to ralslng equlty through dlrect contrlbution from members and owners ls provided by the accumulation of equlty through the retentlon of earntngs. This ls a common mechanlsm among lnvestor-owned fir:rrs and cooperatlves, whereby a portlon of net lncome ts added to the equlty capltal rather than patd out ln cash patronage refunds or dlvldends. Desplte apparent slmllarlttes. equlty capltal ln cooperatlves ls eonceptually dlfferent from equlty ln lnvestor-owned firms. First, not all retained equtty components ln cooperatlves are llnked to net earnlngs. A unique component of equlty retentlon among cooperatlves ls provided by the perunlt capltd retalns: the cooperatlve makes deductions from payments due to members based on thelr volume of buslness and not on net profit and retalns these amounts as part of the equlty capttal account. Second, the equlty of cooperatlves, contrary to that of lnvestor-owned firms, cannot be all vlewed as perrnanent capttal. Part of the retalned eamlngs in a cooperatlve are allocated to members based on patronage. These allocated retained earntngs and the per-unit retalns are eventually distrlbuted back to the members through a revolving fund or other equlty redemptlon program (Cobla, Royer, and Ingalsbe). Thus, allocated patronage refunds can be vlewed as a pool of "deferred cash dtvldends" that the cooperatlve temporartly employs as a component of lts equtty capltal. These dlstlncttve factors may produce dlfferences ln the proportion of equlty tn the capltal structure of cooperatlves as a functlon of rlsk compared wlth lnvestor-owned firms. Thus, the relattonship between the proportion of equtty and the variability of earntngs, as hypotheslzed above, may not hold for cooperatlves. On the other hand, it can be argued that cooperatlves operate tn the s:une economlc envlronment as lnvestor-owned firms and in practlce are apparently judged by the same crlteria by thelr lenders. Thls envlronment may force the cooperatlves to assume the same patterns of financlal behavlor as lnvestor-owned firms, desplte the lntrlnstc differences tn the notton of equity. The present analysts ln effect tests whether 4 JoUR.NALoF AcRIcULTUML CooPERATtoN 1993 the unlque features of cooperatlveequtty are overwelghedby the presslures of a common buslness envlronment. Analysls The theoretlcal conslderattons suggest that the proportlon ofequftycapltal held by a firm ls a functlon of rlsk-related factors: ESITA = fbuslness rlsk, flnanclal rlsk, slze) (I ) where EQ|TA ls the ratlo of equtty to total assets. Flnanclal theory lndlcates that EQ|TA ls an lncreaslng functton of buslness rlsk (the hlgher the buslness rlsk, the greater the equlty cushlon that the firm employs), a decreaslng functlon of financld rlsk (hlgher financtal rlsk tmplles htgher. borrowlng, hence a lower proportlon of equtty), and a decreaslng functlon of slze (larger firms can afford to accept a lower level of equlty than smaller firms of comparable bustness rlsk). Data The model (l) ls esttmated uslng cross-secttonal tlme-serles data from annual financtal statements for fifty-nlne agrlcultural cooperatlves for the years 1973 to 1987. The cooperatlves lncluded tn the analysls are those that responded to a reguest for financlal statement data sent to nonbargalntng cooperatlves llsted by the USDA Agrlcultural Cooperatlve Servlce (Jermolowlcz and Kennedy). They represent a selectlon of dtfferent lndustrles and cover a wlde spectrum of slzes as measured by total assets or sales: the average total assets of the sample cooperatlves for 1973-87 range from $4 mllllon to $1.4 bllllon, and the average sales for the same perlod range from $13 mllllon to $3.1 btllton. The avallablllty of sufficlently long tlme serles for each cooperatlve (15 years of data) makes lt posstble to estlmate the rlsk measures as standard devlatlon of earnlngs over tlme. Thts ls a partlcular strength of the data used tn thls research and dtstlnguishes the present analysts from prevlous studles of factors affectlng equtty capltal. Variables The dependent varlable EQITA ln model (l) ts calculated as the ratlo of total equlty to total assets for each cooperatlve ln each year. The use of thls ratlo tnstead of the actual equtty capttal controls for the strong posttlve correlatlon between equlty and slze and allows comparlson for cooperatlves of dtfferent slze. Cooperatlve slze ls then measured by the cooperatlve's totd sdes ln each year, rather than total assets, to avold the danger of spurious correlatlon between the dependentvarlable and the slze varlable. Buslness rlsk ls measured by the varlabtllty of the stream of operatlng earnings. 1.e., earnlngs before lnterest and tax (EBIT). It ts estlmated as the standard devlation of the return to total assets, or ROA (the ratlo of EBIT to total assets) over the perlod 197347. Thts ls a standard measure of buslness rlsk proposed tn the llterature for lnvestor-owned firms (Brlgham and Gapenski); lt captures the varlablllty assoclated wlth the puie atd Lenun Riskand Equityin CooperativeslPoilbmanl 5 buslness dectslons of the firm before the effects of financtng dectstons, whtch are reflected ln lnterest payments, and government pollcy, whlch ls reflected ln ta:<es. Flnanclal dsk ls not dlrectly observable, and tt ts defined as the dlfference between total rlsk and buslness rlsk (Brlgham and Gapenskl). Total rlsk of a firm ts measured by the varlablltty of the stream of net earntngs to shareholders, and lt ls estlmated ln thls research as the standard derrlatton of the return to equtty before ta( (ROE) over the perlod 1973-87. Beforeta:( ROE ls used because not all cooperatlves ln the sample report profit after tax. Estimation The functlonal model (I) can first be spectfied as a llnear regresslon model descrtbtng the panel data ln the form: (ESlTAlu= c * PTBUSINESS-RISK' + 92FINAI{CE_RISK,* p3SIZE,,* uu. (2) The subscrtpt i ldentlfies the cooperatlve, | = I, . . . ,59; the subscrtpt t 1987. The two rtsk lndlcates the year ln the tlme series, t = 1973 varlables do not carry a tlme subscriptbecause they arebased on summary statlstlcs (standard devlatlons) for the perlod 1973-87. Tlme-serles cross-sectlonal data are usually prone to strong serial correlation. The panel data ln thls research are no exceptlon, productng a Durban-Watson stattstlc of O.3 for the OI5 regresston based on the model (2). To avotd the dlfhcultles assoctated wtth serlal correlatlon in OLS analysts of panel data, the model (2) ls summed over tlme and ls restated in the usual way (Hslao) ln terms of mean varlables over the period 1973-87 for each cooperatlve t: 188/TA\: c, * F'BUSINESS-RISI& * prSlZE, * u,. + F2FINAI{CE-RIS& (3} Here the varlables averaged over time for each coop are denoted by a superior bar. The rlsk varlables from the orlglnal model (2) remain unchanged. The model (3) does not attempt to dlfferentlate among cooperatlves on the basls of thelr tndustry classlficatlon. Thts model ln effect assumes that the lnterccpt c and the slope coeffictents Fr-ps are homogeneous across lndustrles. However, the commodity handled by an agrlcultural cooperative could be an tdenttfiable factor that affects rtsk ln addltlon to the standard devlatlon of ROA used as a meisure of buslness rlsk. A recent study (Lerman and Parllament, l99la) has demonstrated that the proportlon of debt (the complement of the proportlon of equlty) varles stgnlficantly across cooperatlves handltng dlfferent commodltles and performlng dtfferent functlons. To allow for the posslble effect of the commodltles handled by the cooperatlve, a categorlcal varlable ls lntroduced tn the model. The cooperatlves are classlfied tnto five categories based on commodtttes handled: datry marketlng cooperatlves, frutt and vegetable processors, bulk commodtty 1993 6 JounNelorAcmcurrun-uC@pEnnnoN Trble l.-Dlrtrlbudon of Coopcntlvcr by Connodtty Cetcgory Commodtty Category BuIk Datry Dtverslficd Fmlt & Vegetable lnputs: 9 Farm Inputs Farm lnputsr'Gratn l o 5 Graln Total Number of Cooperatlrres 6 lo 4 l5 24 59 marketers (sugar, rlce, and cotton), farm supply cooperatlves, and dtverslfied cooperattvbs. More than half the farm supply cooPeratlves ln the sample market graln ln addltlon to purchaslng farm tnputs for thelr members. The farm supply category ls thus a mlxed category, and, to slmpllff the classlficatlon, gftrln m:rrketlng cooperatlves that do not handle farm lnputsare also tncluiled h thts category. The combtnatlon of three groups of cooperatives lnto a stngle farm supply category lsJustlfied, as a statlstlcal analysls has shown that the subsamples of the mlxed farm lnpuUgraln cooperatlves, the graln marketlng cooperatives, and the cooperatlves that only sell farm lnputs are lndlsttngutshable by thetr equtty and rlsk characterlstlcs. The fifth tndustry category conslsts of the dtverslfied cooperatlves that are lnvolved wtth a very wlde vartety of commodttles, lncludlng farm tnputs and processed foods. The dlstrlbutlon of the cooperatlves by commodtty handled ls provlded ln table l. The model tested wlth the categorlcal commodlty varlable has the form: + p2FINANCE-RISK lEStTAlt= I-..]IIBUSINESS-RISIq " + fuslzE, + prcoMMoDITY+ ITCOMMODITY'BUSINESS-RISK + u,. (4) + p.COMMODITY'FINANCE-RJSK Thts regresston model, ln additlon to the three conttnuous variables of model (3), lncorporates the commodlty categortcal varlable and lnteractlon terms between the categortcal vartable and the two contlnuous rlsk varlables. Thts model ls known as "heterogenelty of slopes" model, and tt ts an extenslon of the analysts of covarlance model, whlch lncludes contlnuous and categorlcalvarlables but no lnteractlon terms (Freund and Llttell). The heterogenelty of slopes model (4) allows for posstble dlfferences ln the lntercepts and the coefffclents across lndustrles. Dlfferent lntercepts reflect dlfferences ln the average level of the ratlo of equlty to total assets across lndustrles, whtle the coefficlents of the lnteractlon terms allow for posslble dlfferences tn the rate of resPonse (tJre "slope") of the dependent varfable EQITA to rlsk tn dtfferent tndustrtes. Thus' the intercept c ln the model (4) represents the average level of the effect for one of the lndustrles Risk and Eq.ity in CooperativeslPailbmml atd l-cnwt 7 only (the so-called base or reference tndustry). Slmllar$ the slope coefficlents 9r, 9r, pe represent the response to the contlnuous varlables as estlmated for thebase lndustry. The othercoefficlents tn the model assoclated wlth the categorlcal varlable and the lnteractlon terms (9., Fs, Fo) modlfy the base-lndustry esttmates to produce lntercepts and response coefficlents for each tndustry. Thus, the coefficlent p. of the varlable COMMODITYallows for posstble devlatlon of the average level of the dependent varlable for each tndustry from that esttmated for the base lndustry. The esttmate of the lntercept for each commodtty category ls gtven by (cr * p.), where p. ls estlmated separately for each nonbase category. The tnteractlon terms between commodlty and the two rlsk mei$ures assume that the coefficlents 9r and p, of the contlnuous rtskvarlables may not be homogeneous acrlosscommodlty categorles. Thus, the estlmated coeffictent of the varlable BUSINESS-RISK for each commodtty category ts given by (9r + 9s), and the estlmated coefficlent of the varlable FTNAI{CE-RISK ls glven by (92 * F"), where the lnteractlon coefficlents pu and go are esttmated separately for each commodlty category. In other words, the model (4) assumes that both the average lwel and the rate of change of the proportlon of equlty wtth rtsk may dlffer among cooperatlves based on commodltles handled. Alternative Dependent Variable Model Although the ratlo of equtty to total assets ls an accepted measure of capltal structure of a firm, other measures are frequently used ln the literature. One such ratlo ls the ratto of equtty to the sum of all borrowed capttal and equlty. Here the denomlnator does not lnclude the firm's accounts payable, whtch ls a component of tts worklng capttal financlng. Thls ratlo may be parttcularly approprlate for cooperatlves, where a large proportlon of accounts payable are amounts owed to members for goods dellvered to the cooperatlve-a financlng component that probably has enttrely dtfferent rlsk characterlstlcs compared wlth other borrowed funds. Accounts payable to members, when reported as a sep:rrate ltem in the financlal statements of cooperatlves, averaged 16 percent of total assets for the sample used ln thts research over the perlod 1973-87. Thls, however, ls a lower bound, because ln many cases accounts payable to members is not reported separately from all accounts payable, whlch averaged 26 percent of total assets ln the sample. The actual level of accounts payable to members ls thus between 16 percent and 26 percent, whlch constltutes a signlficant portton of the cooperatlve-spectfic funds. Therefore, tn addttion tq the model (4), a model ls esttmated wlth the dependent varlable EQm -AP), where (TA - AP) excludes the "accounts payable" from the calculatlon of the cooperatlve's total assets. Pooling A structural factor specific to cooperatlves ls the dtstlnctlon between poolfng and nonpoollng marketlng cooperatlves (Cobla). Sporleder, Malick, and Tough have prevlously estimated pooling cooperatlves (defined as those wlth member marketlng agreements) to operate on smaller proportions of 8 Jounxel oF AcRrcuLTUnelCooptnenox 1993 equtty than nonpoollng cooperattves. They suggest thts result ts due to lower rlsk orposure of a poollng cooperatlve, where a portlon of management and markettng uncertalnt5l ls ellmlnated through the contractual commltment of the members to dellver thetr output to the cooperatlve. To the extent that poollng may alfect the cooperatlve's rtsk, the effect of pooltng ls anaJy?dfor the subsample of fifteen cooperatlve frutt and vegetable procressors, among whlch nlne are poollng. The number of poo[ng cooperatlves wlthln the other lndustrles ln the data base was too small to conduct a slmlliar analysts. The poo[ng status of a cooperattve ts based on tnformatton provlded ln the annual ffnanclal statements. Cooperatlves are defined tn thts study to be poollng lf the cost of goods sold does not lnclude payments for products dellvered by members: thls.ts the case when the cooperailves pool costs and revenues, leavlng the members wlth the restdual. The reported profit ln these poollng cooperatlves ls upward-blased, and, for purposes of analysls, lt ls adJusted according to a prevlously suggested methodologf, (Lerman and Parllament, l99lb). The analysts ts based on the model (4) wtth a categorlcal varlable added for the poollng status and the categortcal commodtty varlable omttted, as only the frutt and vegetable lndustry ls analyzed. The definltlons of the varlables used ln the varlous regresslons are summarlzed ln table 2. The estlmaUons are performed uslng the General Llnear Models procedure tn the SAS/PC package. Results The estlmated coefficlents of the linear regresslon model (3) are: a@1, = 0.3sa 2.24 BUSINBas-RISIQ s.6pFINANCE-RISI! o.os 5i6. (l5.0e1 (4.69, t-2.5O) (- 1.54) The respecttve t-staUstlcs are lndlcated tn parentheses under each esilmated coefficlent. For thls regresslon, the R-square ls 0.52 and the F value ls 2O.O2, whlch ls slgnlficant at the O.OOOIlevel. The esttmatlon resultS tndicate that theproportton ofequlty ln acooperatlve's capttal structure ls alfected by rtsk. As hypotheslzed, the esflmated coefficlent on the bustness rlskvarlable ls slgntficant and poslflve, tmplylng that the proportton of equlty ls dlrectly related to the vartablltty of net earnlngs before tax and lnterest; and the esttmated coeffictent on the financlal rtsk vartable ts stgntfieant and negailve, lmplylng that hlgher financlal rtsk is assoctated wtth a lower proporilon of equlty. The slze of cooperatlves, as measured by mean total sales, ls esilmated to lqve a negatlve effect on a cooperatlve's proporflon of equlty. The slze coefficlent, however, ts only estlmated to be slgntficantly dtffCrent from zero at the l5-percent level of slgnlficance. The estlmated coeffictents of the model wlth the alternailve dependent varlable, IWiiF,Eil, are: lEg(TA -AP)I = o.E{ + l.l4 BUSINESS-RISI! 1.12 FTNANCE_RISII O.O8 SIZE,, ( 1 3 . 6 1 )u . 3 4 t (-2.201 1-4.671 Risl and Eq,tity in CoopcrativeslPorlbrrctl atd Lennn Trblc 2.-De8nltton of VerLrblcr Urcd lD Anefyrb Varlable De6nlUon (E'E/rAII Mean rauo of equtty to total assets for cooperauve I over the pedod f973-87 Mean ratlo of equlty to total assets mlnus accounts payable for cooperatlve I over the perlod 197H7 Mean total salcs for coopcratlve I over the perlod 1973--87 (tn $ bllllons) Standard devlatton of cooperatlve t's rate of return to assets (ROA-net earnings bcfore lnterest and tax to total asscts) over the pertod 1973-87 Dllference between the standard devlatlon of cooperatlve l's rate of rcturn to equlty before tax (ROE) and the standard devlatlon of cooperatlve l's rate ofreturn to ass€ts over the p€rlod f973-87 Categortcal varlable that classtfies cooperatlvc I lnto one of five categorles based on commodlty handted: bulk commodlty, datry, dtverslfied, frutt and vegetable. or lnputs Vartable representlng lnteractlon between the mcasurc of buslness rlsk and the commodtty category vartable Varlable representtng lnteractlon between the mcasuFe of financlal rlsk and the commodlty category varlable Categorlcal varlable that classlfies cooperatlve t as a poollng or nonpoollng cooperatlve Error term lBgtff - AP)l stzEr BUSTNESS_R!Srt FII{,ANCE_RISI& COMMODITY COMMODITY'BUSINES9_RISK COMMODITY'FINA}ICE-RISK POOL, tlt The respectlve t-statlstlcs are agatn tndlcated tn parenthesis. For thts regresslon, the R-square ls O.34 and the Fvalue ls 9.46, whlch ts significant at the 0.OOOI level. Wlth accounts payable subtracted from total assets, buslness rlsk loses Its slgnlficant effect on equtty proportlons and slze lncreases lts effect. The estlmated coeffictent on the financlal rtsk varlable ls stgnlficant and negatlve, but the estlmated coefficlent on the bustness rtsk vartable ts not slgnlficantly dtfferent from zero. The slze coefnclent ls estlmated to be slgntficantty dlfferent from zero at the 5-percent level of stgnlficzrnce, tmplylng larger cooperatlves hold a lower proportton of equlty capltal. The coefficlents estlmated for the heterogenelty of slopes model (4), tncludlng the commodlty categortcal varlables along wlth the conilnuous rtsk and slz€ varlables, are gtven ln table 3. Agaln regresslons were run for tlle two dlfferent measures of the proportlon of equtty. The coefficlents Indlcated as llnes 14 ln table 3 are the esttmated coefficlents for the lntercept, rlsk, and slze varlables for the cooperatlves ln the lnputs category, usd as the base commodlty category tn the analysts. The coefficlent estlmates gtven ln ltnes 5-16 for the other four commodtty categorles must l0 JounNel or Acnrcurrunel CoopERArroN Trble S.-Ertlmrted CocfEcteatr of Rcgpcrlon Uodeb Dcpcndent Varlable: lndependent Varlable I lnterccpt 2 BuslnesE Rlsk 3 Flnanclal Rtsk 4 Slzc 7 dlverslfed o.46..r ( o.o5ol 2.72... ( 0.790) _ 1.7o... ( 0.293) -0.o3 o.64... ( 0.098) 2.39 ( r.561) _ l.go... ( 0.579) -0.67 { o.o59t ( o.uE) - o. l3.r 8 frult & vcgetable Bustncss Rlsk 'Commodlty ( o.oe{) -0.10 ( o.l5r) o.oo ( 0.331) -o.rl ( 0.2341 -o.t7 I o.r27l Handlcd 9 bulk clmmodltles lO datry dtverslfled 12 frult & vegetable Flnanclal RJsk t Commodlty Handtcd fg br.rtk commodltles 14 dalry 15 dlrcrstfed 16 frurt & vcgctable Rr: Mean Value of Dependent Varlable: Numbcr of Obscrnatlons: " E.tfutcd ' E.tlmtcd lF4t4TA - AP)t Paramcter Esttmetc -o.13. ( o.o77l -o.21 ( 0.167) -o.tl 6 datry Notc: IrfuDb.rr (E€/TA} Parameter Esttmatc ( o.o3o) Commodlty Handlcd 5 bulk commodltles ll 1993 - r.89 ( 2.s06) - 1.33 ( r.269) 4.93 ( 3.950) 5.20 (l l.oo4) - r.34 ( 1.026t - 1.99 ( 7.801) 6.23 12r.7351 -0.99 12.0281 l 06r. ( 0.451) o.24 ( 1.355) - l.2l ( 4.0871 l.16.r. ( 0.338) .73 .38 59 r .l 5 ( o.89r) 3.60 ( 2.6781 - 1.43 ( 8.074) 1.27. ( 0.6671 .53 .53 59 ln pJqrlh€ rrG rtrndrrd m. to bc d€nfhDoy bEt dlraEt trc6 IF rt tlrc t pcrc.trt .lgnlfdloc 16rl to bc rlgnffmtt dflaat floo 4rc rt rhc 5 pclccot lEnriam. to bc rl6pr0antt d|lllmt fro6 zrc rt lbc lO Fcrt .tgnlJL|ncc l6d. Risk and Eqrity in CooperativeslPorlbmml atd Lcnnn ll Conblncd Cocf,Eclente:Eqrdty to Total Aretc a3 Table 4e.-Ertlmrtcd a Fulcdon of Buslncer RlrL, Flnanchl f,1sk, and Slze Commodlty Inter€lpt lnputs BuIk Datry Dlrrcrslflcd Frutt & Vegetable 0,46... o.33^ 0.25 o.35 0.33^^ Bustness Rlsk Ftnanclal Rlsk 2,72.t' 1.39 7.65 7.91 t.37 _ l.7or.. -o.a[^^ - 1.46 -2.9r - 0.54^^ ^ -o.o3 -o.03 -o.03 -o.03 - o.o3 Table 4b.-Ectlmatcd Conblned Coef8clcDtss Equfty to Totd Agscts Lcrs Accounte Payable es e nrncdon of Buslness Rlsk, Flnencld RlsL, and Slzc Commodlty lntercept Bustness Rlsk Ftnanctal Rlsk Inputs Bulk Dairy Dtverslffed Fmlt & Vegctable o.64... 0.54 o.64 0.53 o.47 2.39 0.40 0.30 8.62 1.40 - 1.90 -0.75 - t.70 -3.33 -0.63^ Not.: '.' Ertlntcd rErufently ^^^ (lltfmt SlSnl0crntt ^^ SEnlfc.nlly dlfimt ^ SEilnc.ntly dtltcrsrt dttlmt from 'm (tm lhc €tlmt d fmm thc cctlmtcd fron thc crtlmrtcd -o.o7 -o.o7 -o.07 -0.o7 -o.o7 laEl. rt the I p.rccnt alglrrfenc. cfnddt for lnput! at thc I p.rccnt hrl of stgnlficana. at lhc 5 Frccrlt lcl of stgnlficance. o.t[d6t for lnprt ctnclot fs rnput! .t the lO Frnt l6d of slSntncrnc. be added to the correspondlng base-category estlmates ln llnes 1-3 to obtatn the final estlmates for each commodlty category. Tables 4a and 4b recast the estimated coeffictents from table 3 lnto the form of linear regresslon equatlons for the five commodtty categortes. For the regresslon ustng (Wm) as tle dependentvariable, the R-square lsO.72 and the Frralue 1s7.32, whtch ts slgntficant at the 0.OOOIlevel.The slgns of the estlmated coeffictents forthe riskvarlables.re as hlpothestzed, wtth both buslness and financtal rtsk factors slgntficant. Although the size coefficlent ls negatlve, lt ls not estlmated to be different from zero. The esttmated coefficlents on the commodlty classlficatlons and the lnteraction vartables lndtcate that stgntficant dlfferentld effects among the commodlty categorles exlst (see table 4). For the commodlty classlficatton regresslon model wtUl t@7(E - FIt as the dependent varlable, the R-square is O.53 and the F value is 3.26, whlch ts slgnificant at the O.OOI level. The estlmated coefficlents lndicate that the proportlon of equlty capttal ts only slgnlficantly affected by finan- 12 Jouruvrl oFAcnrcuLTrrMLCoopEnATroN Trble 6.-Erdortcd Equtty to fbudd Vqleteblc 1993 Coc6dentr of tbe pootbg Rcgfellon Uodel: Totel Arrctr u e Fuocdon of n-rutncr. RLL, RbL,Slrc, ud poolhg for the Enrtt end hd|rltrt Pararncter Esdmate Varlable lnterccpt 0.32... (o.0599) 1.37. (o.0669t - o.51.. {o.1798) -o.o2 (o.l68E) Bustness RJsk Flnanclal Rlsk Slzc -o.oo Nonpoollng (0.0479) Rt = .50 Mean Vdue of EQITA = .31 Nusrber ofObscrvaUons = l5 l'lote NuEbo ': ' PrUE[tcd Eatlndcd ln F 6rthG to bc rlgnl[cantly to bc algdncrntt r|t at|rdard 6rur. d{Icrrnt f]oD rcro rt tbc 5 pcrccnt rEntic.ncc dUfcrtnt fro6 rcrc rt thc tO pcrccnt.-lgnticrrE lciEt L:vrl. clal rlsk, and thts financtal rlsk effect for the frult and vegetablecategory ls slgnlficantly dtfferent from the other cooperaflves.The doeffictent of-thi slz: varlable that was estlmatedto havea slgnlficant effecton thts dependent varlable before the tntroducilon of the commodlty vartable lose3tts srgnlficancewhen the commodlty categorlcalvarlable ts tntroduced tnto the.model.{nparegUy the wlde varlablltty ln accollntspayableamongcooperatlves reduces the ercplanatorypower of the effect of bustness rt5t<ario slze-on the-proportlon of equtty to assetsexcludtngaccounts payable. The results of the regresslon tesilng the effect ofpooltng ontlie capttal structure of frult and vegetablecooperailvests reportedln table s.-The :tgns.-of the estlm_atedcoefficlents for the rlsk vartlbles are as prevtously hlpothestzed. wtth both buslness and financtal rlsk factors signtficanf. slze ls ag{n estlmated to not have a stgntficant effect on the pioporilon of equtty. Contrary to the results of Sporleder,Mallck, and Touih. ioottne 19not found to slgnlficantly affect a cooperailve'sproportton of equtty. Ii thls anarysls,the expltclt buslness and financtal rlsk factors have apparently eiptured the explanatorypowerof pooltngoperailons.The dtffeiCnce ln ffndlngs may be attrtbutable to a dlfferencern ue oract deffnltton of pooltng ttl th9 two studles and to the fact that the present analyals was conducted only for frr.tt and vegetablecooperaflves.whtle the rbults of the prevlous research were obtalned acrq*s pooltng cooperailves from all tndustrles. Risk and Equity in CooperativeslPorlianail atd l*nan l3 Concluslon Equl$ holdlngs among agrtcultural cooperatlves are found to be alfected by rlsk-related factors. The emplrlcal analysls tndlcates that the ratlo of equlty to total assets ls affected by measures of buslness rlsk and financlal rlsk, and, ln some cases, these rlsk factor effects are estlmated to dlffer based on the commodlty handled by the cooperatlve. The proporUon of equlty capltal ts found to be unrelated to cooperatlve slze, as measured by sales, ln most of the regres{rlons. The proportlon of equlty ts unrelated to whether the cooperatlve operates on a poollng basts. These results support other emplrlcal evldence (t ennan and Parllament l99la; Royer) lndlcattng that the equlty-based performance measures of U.S. agrtcultural cooperatlves (return on equlty, leverage) are very simllar to those of lnvestor-owned'firms. Therefore, in practlce, managers ln cooperatlves may follow the same equtty pollcy as murnagers ln lnvestor-owned firms, and we can erpect to observe the same relatlonshlp between equtty and rlsk that ls assumed for lnvestor-owned firms. Perhaps the common pressures of the competttlve buslness envlronment and the standard demands of the financlal communlty overwelgh the unlque features of cooperatlve equlty and account for thls stmtlarlty ln behavlor. References Brealey, R. A., and S. C. Myers. Pdnctples oJ Corporate Flnance.4th ed. New York; McGraw-Hlll, f991. Brlgham, E. F., and L. C. Gapenskl. Interme dtate Ftnanctal M anag ement. New York: Dryden Press, 1985. Cobla, D. W. "Spectal Toptcs for Marketlng Cooperatlves." lnCooperotTues tn Agflculture, ed' D. W. Cobla, pp. 195-220. Englewood Cllffs, N.J.: Prentlce-Hall, 1989. Cobta, D. W., and T. A. Brewer. "Equlty and Debt." ln Cooperattues in Agrlculture, ed. D. W. Cobla, pp. 243-66. Englewood Cltffs, N.J.: Prentlce-Hall. 1989. Cobla, D. W. ; J. S. Royer; and G. Ingalsbe. "Equlty Redemptlon. " In Cooperatlues in Agrtculture, ed. D. W. Cobta, pp.267-86. Englewood Cltffs, N.J.: Prentlce-Hall, 1989. Freund, R. J., and R. C. Llttell. SAS/or Linear Models: A Guide to the N{OUA and GLM Procedures. Gary, N.C.: SAS Instttute Inc., 1981. Hslao, C. A na lysl.s oJP onel D ata. Cambrldge : Cambrldge Untverslty Fress, 1989. Jermolowlcz, A., and T. Kennedy. Dtrectory oJF armer Cooperatlves. Washlngton, D.C.: USDA ACS Serv. Rep. 22, 1989. Kane, M. "Improved Ag Economy, Management Help Top 1O0 Co-ops Improve Returns to Members-Top IOO Cooperatlves: lg88 Ftnanclal Profile." Farmer Cooperattues. Oct. 1989, pp. f 4-18. Lerman, 2., and C. Parllament. "Slze and Industry Effects ln the Performance of Agrtcultural Cooperatlves." Agrtculturat Economics 6{ l99la): l5-29. . "Estlmattng the Accountlng Profitabtlity of Poollng Cooperatlves." Journal oJ Agrlbustness 9(l99lb):3-l l. l{ Jounxelor AcntcurrunuCoopen^trtox 1993 Royer,J. S. "AComparatlveFlnanclal RatloAnalyslsofU.S. FarmerCooperatlves Ustng Nonparametrtc Stattstlcs. " Journcl of AgrTculturalCaperatfon 6(l99ll:22-44. Schrader, L. "Equtty Capltal and Restructurlng of Cooperatlvesas InvestorOrlented Flrms. " J oumal oJ Agrtcultural Cooperation 4( I 989):4 I -53. Sporleder,T. L.; W. M. Mallck; and C. H. Tough. "Relatlonshlp of Poollng to Equlty Capltal and CurrentA^ssetsof LargeProducerMarketlng Cooperatlves." Journal oJ Agrlculturol Cooperatlon3( 1988):28-38. Staatz, J. Former Cooperattue Thary: Recent Deuelopments. Washlngton, D.C.: USDAACS Res.Rep.84, 1989.