Puttng Ther Money Where Ther Mouths Are: Consumer Wllngness to Pay for Mult-Ingredent, Processed Organc Food Products Marvn T. Batte a *, Neal H. Hooker a, Tmothy C. Haab a, and Jeremy Beaverson b a Department of Agrcultural, Envronmental, and Development Economcs, The Oho State Unversty, 2120 Fyffe Rd., Columbus, OH 43210-1067 b. Former graduate research assocate and Merchandser/trader, Archer Danels Mdland Company, 4666 Fares Parkway, Decatur, IL 62526. Ths s a pre-publcaton verson of a paper ultmately publshed n Food Polcy, Vol 32 (2) 2007:145-159. Abstract In response to dramatcally ncreasng adopton n consumer markets, the Natonal Organc Program (NOP) ntated novel labelng standards for food products n the U.S. n 2002. Ths program s a partcularly relevant standardzaton effort for mult-ngredent processed foods. Rather than a smple bnary message (organc or not), gradatons of organc content are now codfed. No exstng publshed study evaluates consumer wllngness to pay or motvaton to purchase n response to such a rch organc label. Ths artcle presents evdence of the mpact of the NOP through analyss of data collected n seven central Oho, USA grocery stores. Results suggest that consumers are wllng to pay premum prces for organc foods, even those wth less than 100 percent organc ngredents. The magntudes of WTP prema vared sgnfcantly among consumer groups, suggestng that targeted marketng may be effectve for organc merchandsers. Keywords: Mult-ngredent processed organc food; Natonal Organc Program; consumer ntercept survey; wllngness to pay; WTP
Puttng Ther Money Where Ther Mouths Are: Consumer Wllngness to Pay for Mult-Ingredent, Processed Organc Food Products Currently the U.S. organc ndustry s boomng wth annual ncreases n consumpton of 20 percent per year (Dmtr and Greene). Sales of organc foods ncreased from $5.5 bllon n 1998 to an estmated $13 bllon n 2003 (Food Marketng Insttute). One recent report suggests the market wll grow to over $32 bllon by 2009 (Packaged Facts). The rapd ndustry growth has led to questons about the regulaton of organc marketng. In 1999 more than 30 states had organc laws and more than 40 enttes provded thrd party certfcaton usng a varety of standards (Fetter and Caswell). Though dfferences between programs may have been subtle, t was very dffcult to consstently compare one organc product to another. Consumers were confronted wth a dverse array of organc standards at the state, retaler, or product level. In an effort to resolve ths confuson, the Natonal Organc Program (NOP), mplemented October 21, 2002, formalzed rules for organc producton, processng, certfcaton and labelng (Agrcultural Marketng Servce). The NOP ntated novel food labelng standards, partcularly relevant for multngredent processed foods. A smple bnary message (organc or not) s suffcent to dentfy the most commonly purchased organc tems produce. More nterestng, gradatons of organc content are now codfed and may (or may not) facltate marketng of mult-ngredent foods consdered by many to be the most lkely vehcle for further growth n the organc sector (Whole Foods, 2003; 2004; Organc Trade Assocaton, 2004). There are four levels of the clam covered by the NOP: 100% organc, 1
Organc (at least 95% organc), Made wth Organc Ingredents (at least 70%) and Some Organc Ingredents (less than 70%, the organc tems can be lsted ndvdually n the ngredents on the sde panel). The frst two categores can use the NOP seal (fgure 1, seal s green n color) on the front of the food package, and only the frst three categores can use the word organc on the front panel of the food package. The NOP seal may be used as a smple short-cut or perpheral cue (Petty and Cacoppo) of product qualty f consumers do not engage n sgnfcant analyss or argument processng, choosng to use the seal as suffcent evdence of an organc product. In ths case, the seal may cause the consumer to shut down product analyss and not carefully scrutnze the product's other label nformaton (such as that found on the Nutrton Facts panel, health clams or brand/product marketng communcaton messages). In addton, a less favorable opnon may be held about organc products whch do not exhbt the seal. The presence of these new categores of partal organc content mult-ngredent foods rases a number of mportant questons. How wll consumers perceve these product categores? Are they wllng to pay for hgher organc content, all else equal? Wll food companes use these categores as an nterm measure to address dffcult to source organc ngredents and/or to segment markets and offer multple versons of products thereby servng as a vehcle for prce dscrmnaton whch captures a larger proporton of consumer surplus? 2
Lterature Revew In the past 15 years consumer demand for nche products (ncludng organc, natural, and locally grown) has grown substantally (Dmtr and Greene). Although natural/specalty retal stores account for the largest porton of the market (47 percent of sales n 2003), 44 percent of organc food sales are made n manstream grocery stores (Organc Trade Assocaton, 2004). Ffty-seven percent of restaurants wth per-person dnners prced $25 or more and 29 percent of restaurants wth dnner costs n the $15 to $25 dollar range also offer organc optons (Organc Trade Assocaton, 2000). Whle some studes suggest that the motvaton to purchase organc and natural products derves from envronmental concerns, most conclude that the prmary motve relates to health concerns (Huang). Thompson provdes an excellent revew of the emergng lterature on consumer demand for organc foods, and recommends further studes, ncludng those whch better document atttudes, motves, and wllngness to pay (WTP) for a range of products, populatons, and market channels. Accountng for where foods are purchased s lkely to be mportant n understandng where potental growth n organc foods mght occur (Thompson, p. 1117). Although the authors are unaware of any publshed study that consders the specfc mpact of the NOP labelng regme or WTP for organc processed foods, there have been a number of studes of demand for organc characterstcs and other attrbutes n produce (see for example, Buzby, Ready and Skees; Dunlap and Beus; Thompson and Kdwell; Baker; and Wllams and Hammtt, 2000; 2001). Lourero and Hne suggest that commodtes wth "locally grown", GMO-free, and organc labels all can command 3
premum prces. Usng a contngent valuaton survey, they found that Colorado consumers were wllng to pay the largest premums for "Colorado grown" potatoes, followed by organcally grown and GMO-free. Suryanta also found that dentfcaton of local product (n ths case Hawa s foodstuffs) allowed capture of a premum prce for pneapples and macadama nuts. Wang and Sun found that Vermont consumers were wllng to pay more for organc apples and mlk produced locally and certfed by NOFA (Northeast Organc Farmng Assocaton). They also concluded that consumers most lkely to purchase these products were young, n households wth few members, and wth hgher household ncome. The standardzed labelng requrements of the NOP should beneft consumers by reducng confuson about the organc character of food products. Hutchns and Greenhalgh argue that to effectvely promote organc products t s necessary to develop a cohesve marketng strategy that depends on a better and fuller understandng of consumers, consders all partes n the food chan, and s ntated by leaders n the ndustry. As Fetter and Caswell observe, the success of the natonal standard n supportng the market for organc products wll ultmately rest on how well t matches the demands of consumers and other buyers (p. 72). Often, consumers value organc products not just because they perceve the products to be healther, but also because they perceve them to be more envronmentally frendly, and more supportve of small scale agrculture and local rural communtes (Wllams and Hammtt, 2000, 2001; Underhll and Fgueroa). 4
Govndasamy and Itala surveyed consumers at fve grocery retal stores n New Jersey n March 1997 to obtan estmates of WTP for organcally-grown fresh produce. Ther analyss showed that females wth hgher annual ncomes, younger ndvduals, and those who usually or always purchase organc produce were more lkely to pay a premum. They also conclude that the lkelhood of payng a premum goes down as the number of ndvduals n the household rses. Thompson and Kdwell, n a 1998 study of conventonal and organc produce purchases, concluded that famles wth chldren were more lkely to buy organc produce than those wthout chldren. Ths result was opposte to that found by Lourero and Hne and Wang and Sun who concluded that consumers wth chldren were less lkely to buy organc products. Interestngly, Wllams and Hammtt (2000, 2001) found few soco-demographc dfferences between organc and conventonal fresh produce shoppers, perhaps ndcatng that as organc offerngs become more pervasve so ndvdual market segments become less dstnct (Baker). Although there have been a number of studes of demand for organc commodtes, lttle has been done to understand the demand for mult-ngredent processed organc foods. There s no reason to presume that consumer motvaton to purchase such organc foods or WTP (f present) should match results found n studes of produce. Under the NOP, processed food products wth less than 100% organc ngredents may stll carry an organc dstncton through a varety of label mechansms. Hence, consumer nterpretaton and confdence n these labels are mportant. Ths paper provdes one of the frst assessments of a complex label message (scale of organc content) replacng a smple bnary message (organc or not). The research 5
presented here s tmely, conducted one year followng the mplementaton of the NOP. The research provdes nsght nto consumer demand for mult-ngredent processed organc foods, and tests the mpact of consumer awareness of the NOP on WTP for these products. Research Data and Methods A customer ntercept survey was conducted durng October and November, 2003. Sx stores of a U.S. natonal grocery chan (tradtonal grocery) were selected for the survey. Two stores were located n the nner cty of Columbus, Oho, USA, two stores were n suburban areas of Columbus, and two n small towns n predomnantly rural areas of central Oho. Customers were dentfed at random as they entered the store. In order to mnmze dfferences among stores arsng solely due to tme of survey (and systematc dfferences n shoppers that mght be assocated wth tme), all customer ntervews were conducted between the hours of 1:00 and 6:00 pm, Monday through Thursday. 1 Partcpants were asked to complete a short survey n the store that elcted nformaton about organc purchase behavor, knowledge of organc food labels, atttudes toward health and nutrton ssues, and household demographc nformaton. Approxmately one-thrd of the shoppers approached agreed to complete the ntervew and survey. One hundred nnety nne surveys were completed. Whle the experment conducted n tradtonal grocery stores was desgned to study choces made by the general populaton of Oho consumers, t s also of nterest to dentfy the characterstcs of shoppers that lead them to self-select nto the organc market. To provde a comparson to the tradtonal grocery, n March 2004, shoppers at a 6
natonal natural food store (specalty grocery) located n a suburban area of Columbus, Oho also were surveyed. Because the specalty store was reluctant to allow ntervews to be conducted n the store, shoppers were randomly ntercepted and asked to complete and return a take-home questonnare. The questonnare ncluded dentcal questons to those asked of the tradtonal grocery shoppers. Three hundred questonnares were dstrbuted and 102 were returned, a response rate essentally equvalent to the rate of partcpaton by tradtonal store shoppers. Even though the dfference n survey admnstraton creates dffcultes n comparng the two store types, the fact that response rates were approxmately the same across the two samples suggests the potental bas s neglgble. Table 1 ncludes descrptve statstcs for the sampled consumers dentfed by tradtonal and specalty grocery. There were substantal dfferences n the characterstcs of consumers n the two store formats. Specalty grocery shoppers were somewhat younger, less lkely to have chldren n the household, had more formal educaton, and hgher mean household ncomes. They also were much less lkely to be non-whte and were much more lkely to be vegetaran or vegan. It should be noted that the selecton of tradtonal grocery stores was purposeful - to ncrease the varablty of consumer characterstcs. For nstance, the nner cty, rural, and suburban tradtonal groceres dffered substantally by dstrbuton of race, ncome, educaton level, and other demographc measures. Wllngness to pay for organc food content: A prmary focus of ths study s to estmate consumers wllngness to pay for multngredent processed foods, and nvestgate the wllngness to trade-off mult-ngredent 7
foods contanng varyng levels of organc ngredents. Although a number of studes have estmated WTP for organc produce, no study has consdered mult-ngredent processed organc foods. Furthermore, no known studes have consdered wllngness to pay for mult-ngredent food products dfferentated by the four levels of organc content that are allowed under the NOP gudelnes. Certan propretary data on actual consumer purchases are beng collected (e.g., SPINS/AC Nelsen retal scanner data of natural, organc and sustanable products). However, these data cannot be matched to consumer attrbutes, hence questons of whch consumer attrbutes nfluence wllngness to purchase varous organc products cannot be examned. In addton, product strateges followed by food companes and retalers make t more dffcult to evaluate consumer choce wth scanner data. For nstance, most of the major food companes wth organc products offer them under a dfferent brand name, the product may vary substantally n product form or prmary ngredents from ther conventonal product, and the company may not offer products n each NOP category. Addtonally, retalers often place organc products n separate locatons from conventonal products, makng prce comparsons more dffcult. The contngent choce approach s a useful alternatve research tool for such an envronment whereby consumers can be asked to make choces among alternatve hypothetcal products, allowng for standardzaton of selected product characterstcs (taste, brand, etc.) but manpulatng key nformaton provded to the consumer (organc content). A payment card method was used to estmate consumers wllngness to pay for several food characterstcs ncludng level of organc content. Consumers were presented 8
wth the purchase of a hypothetcal adult orentated breakfast cereal product. Specfcally, they were asked: Assumng breakfast cereal s prced at $3.00 per box at your local grocery store, how much more would you be wllng to pay for each of the followng characterstcs? The prce premum ndcated s nterpreted as the wllngness to pay for that characterstc. Eght characterstcs were dentfed, and seven payment levels were offered, ncludng an opton to pay zero addtonal for the characterstc. The largest premum category was an open-ended range - more than $1.00 premum per box. A complete lstng of the food characterstcs (n the same sequence as presented n the survey), prce ranges, and the dstrbuton of consumer responses to ths queston are lsted n table 2. For ntal analyss, consumers are assumed to be wllng to pay the lower bound on the range of wllngness to pay ndcated. For example, f a consumer ndcates she s wllng to pay an addtonal $.25-$.49 for pestcde free foods, then we know she s wllng to pay at least $.25. The last two columns of table 2 report the mean and medan prema based on the lower bound assumpton. Both tradtonal and specalty grocery shoppers ndcated the hghest WTP for the same three characterstcs -- although n a dfferent order. Tradtonal grocery shoppers placed the hghest premum on pestcde free ngredents, followed by 100% organc and locally grown characterstcs. Specalty grocery shoppers placed the hghest premum on 100% organc ngredents, followed by pestcde free and locally grown characterstcs. It s also nstructve to note that specalty grocery shoppers were wllng to pay substantally larger premums for many food characterstcs than were tradtonal shoppers. Specalty 9
shoppers ndcated a mean WTP that was 100 percent larger for GMO-free foods, and more than 50 percent larger for 100% and 95% organc foods, locally grown foods, and pestcde free foods. However, specalty store shoppers dsplayed a smaller mean wllngness to pay for the less than 70% organc content. The larger WTP for hgh organc content foods combned wth the lower WTP for the lowest organc content foods may suggest that the specalty food shoppers embrace some threshold level of organc content beyond whch the product no longer warrants a premum prce. For both groups the results clearly ndcate a declnng WTP for foods as the percent of organc content dmnshes. Regresson analyss of WTP responses from the payment card data s used to provde a more complete understandng of these patterns. For any gven food product attrbute, consumers wll dffer n ther wllngness to pay a premum prce for that attrbute. Some wll be wllng to pay no addtonal prce. The determnstc and random factors that nfluence whether the ndvdual wll pay any premum for a gven food attrbute may be dfferent than those that nfluence the amount of premum that the consumer wll pay for that attrbute. For ths reason, we have mplemented a two-stage hurdle model to allow dfferent factors to nfluence the decson to pay a premum (.e. zero WTP) and the decson of how much of a premum to pay condtonal on payng a premum. Ths decson model s equvalent to the contnuous Cragg model (see Haab and McConnell, 2002 for a descrpton). The only dfference between the model estmated here and the Cragg model s the use of an nterval censored model n the second stage of estmaton to 10
account for the payment card nature of the data nstead of a contnuous truncated model as n the Cragg. Frst, we consder factors whch determne whether consumers are wllng to pay a premum for varous cereal characterstcs. Table 3 reports the results of a bnary Probt model on whether consumers are wllng to pay some premum (versus none) for each of the eght attrbute categores. Independent varables nclude demographc characterstcs: respondent age, ncome per person n the household, presence of chldren (age 18 or younger) n the household, an ndcator for whether the respondent has post hgh school educaton, a health ndex (scaled 0-100) ndcatng degree of concern for varous health attrbutes of food, an ndcator varable for race (whte=1, non-whte=0) and an ndcator varable for gender (female=1). Indcator varables for the specalty store sub-sample and for whether the respondent has prevously seen the NOP organc seal on a food product are also ncluded. It s expected that consumers n the specalty store and those aware of the NOP seal are more lkely to be wllng to pay somethng for foods contanng certan attrbutes (such as no pestcdes, no genetc modfcaton and hgher organc content). We have no expectatons on the relatonshps between demographcs and wllngness to pay for such attrbutes as we do not presume results of prevous WTP studes of organc produce drectly apply to mult-ngredent organc processed foods. Table 3 reports the results of the Probt models. Bold entres n the table ndcate statstcal sgnfcance of the parameter estmate at the 90% confdence level. Demographc varables have lttle nfluence on wllngness to pay a premum for cereal 11
attrbutes. Consumers wth chldren are less lkely to pay a premum for pestcde free cereal, and consumers wth hgher educaton levels are less lkely to pay a premum for less than 70% organc ngredents. It was expected that consumers who are more concerned about health ssues would be more lkely to pay premum prces for foods perceved to be healther. Results suggest that consumers wth a hgher health concern ndex were more lkely to pay premums for food wth lower levels of organc content (70-95% and less than 70%), but were no more lkely to pay premum prces for foods wth 95% or more organc ngredents (or pestcde-free, GM-free and locally-produced foods) than were consumers wth lower health concern ndces. As expected, shoppers at the specalty store are wllng to pay a postve premum for pestcde-free cereal, GM-free cereal and cereals contanng more than 70% organc ngredents. As the percentage of organc ngredents ncreases from less than 70%, to 70-95%, to 95% to 100%, the probablty a specalty store customer s wllng to pay a premum ncreases by 10%, 16%, 17% and 21% relatve to non-specalty store customers. These are the changes n the probablty of wllngness to pay a premum for a change n the specalty store ndcator from zero to one, holdng all other explanatory varables constant at ther respectve means. Awareness of the NOP seal has a postve and sgnfcant effect on the probablty of beng wllng to pay a premum across all categores except for enhanced flavor an ssue not addressed by the organc program. Interestngly, ths effect extends to locally grown (ngredents), agan not addressed n the NOP. In contrast to the specalty store customers, however, no clear pattern emerges n 12
the magntude of NOP awareness on lkelhood of payng a premum for varous levels of organc content. Consstent wth the two-stage Cragg model, condtonal on beng wllng to pay a premum for the varous attrbutes, consumers are asked to ndcate the amount of the premum they are wllng to pay. As dscussed above and reported n table 2, consumers wllng to pay a premum choose from a menu of premum ranges ndcatng the range n whch ther maxmum wllngness to pay for that attrbute falls. Because the respondent s choosng a range of wllngness to pay, each response provdes nformaton on the upper and lower bound of the premum, but not the exact amount. To model the exact wllngness to pay premum we use an nterval censored regresson model. Suppose wllngness to pay for ndvdual (WTP) s strctly postve, as s the case for those wllng to pay a premum, such that: WTP e X (1) where X s a vector of ndvdual specfc explanatory varables, s a conformng 2 vector of parameters to be estmated and s a mean zero constant varance error term that s ndependently and dentcally dstrbuted across ndvduals. Each respondent must choose the range of wllngness to pay from the pared premum sets: {[.01,.09],[.10,.24],[.25,.49],[.50,.74],[75,0.99],[1.00, ]}. For each response, we obtan an upper bound on the premum (U) and a lower bound (L). From equaton (1), the probablty that a respondent chooses the range [L,U] s: P X L WTP U P L e U P ln L X ln U X 13
Assumng the error term s normally dstrbuted wth mean zero and constant varance 2 ( ), the probablty of the true premum fallng n the range [L,U] s: P L WTP U ln U X ln L X (2) where s the standard normal cumulatve dstrbuton functon. Equaton (2) represents the contrbuton to the lkelhood functon of an nterval censored model for ndvdual. By assumng ndependent and dentcally dstrbuted error terms across respondents, the lkelhood functon for the nterval censored model s found by multplyng the ndvdual contrbutons to the lkelhood functon (2) across all ndvduals n the sample. Maxmum lkelhood estmaton s then used to obtan consstent estmates of the parameter vector and the error standard devaton. 2 From equaton (1), the resultng parameter estmates for are the percentage change n the wllngness to pay prema for a one unt change n the ndependent varables (often referred to as sem-elastctes). Table 4 presents the results of the nterval-censored data model estmated on the subsample of consumers ndcatng a non-zero wllngness to pay prema for each attrbute category. Agan, bold entres ndcate sgnfcance at the 90% confdence level. The ndependent varables ncluded n the model are the same as those dscussed for the Probt model earler n ths secton. Demographc varables appear to have more of an mpact on the amount of the premum consumers are wllng to pay than on whether they are wllng to pay a premum at all. Condtonal on beng wllng to pay a premum, older consumers are wllng to pay hgher premums across attrbute categores wth the 14
excepton of 95-99% organc and locally grown ngredents. Indvduals wth a hgher per person ncome n the household are wllng to pay larger premums for hgher organc content cereal (100% and 70-95%). Lkewse, consumers wth chldren are wllng to pay larger premums for 70-95% and 95-99% organc content cereals than are consumers wthout chldren. Whte consumers state lower premums for pestcde-free and locally grown ngredents than do non-whte consumers. Females are wllng to pay hgher premums across attrbutes, n partcular hgher organc ngredent content and pestcdeand GM-free ngredents, than are males. Agan, an nterestng but predctable pattern emerges for specalty food customers, who are wllng to pay hgher premums for pestcde-free, GM-free and hgher organc content ngredents than ther tradtonal store counterparts. In addton, the premum that specalty store customers are wllng to pay for organc ngredents ncreases wth organc food content. Surprsngly, consumers who are aware of the NOP seal report no sgnfcant dfferences n wllngness to pay prema than those consumers who are unaware of the seal. From the combned results of Tables 3 and 4, t appears that the NOP seal acts to educate some consumers that the ngredents n the product are worthy of a premum, but does not rase the sze of that premum condtonal on the consumer beng wllng to pay a premum for those types of ngredents. If consumer A s aware of the NOP seal and an otherwse dentcal consumer B s not, but both consumers are wllng to pay a premum for hgher organc content, the amount of the premum wll be the same for these two consumers. On the other hand, the earler Probt analyss ndcates that consumers aware 15
of the NOP seal are more lkely to be wllng to pay a premum than those that are not aware of the seal. The presence of the seal doesn t command an addtonal premum above that assocated wth a partcular level of organc content. As such t s unlkely that the seal s actng as a perpheral cue for consumers. Wllngness to pay estmates were calculated for the sample usng the models reported n Table 4. The mean, medan and extreme values for these estmates are reported n Table 5. The mean (medan) wllngness to pay for 100% organc cereal was $0.45 ($0.40) per box. Ths was followed n decreasng order by pestcde free, GM free, and locally-grown ngredents. Consumers were only wllng to pay $0.15 addtonal per box for a cereal wth less than 70% organc ngredents -- about one-thrd of the premum commanded by 100 percent organc cereals. Summary and Implcatons Ths study represents the frst research of ts knd followng the mplementaton of the Natonal Organc Program n October 2002 and explores consumer choce for a multngredent processed food wth varyng organc content as provded for by NOP label gudelnes. It ncludes data for both tradtonal grocery shoppers, wth consumers from cty-center, suburban, and rural stores, and shoppers of a specalty natural foods grocery store. Estmates suggest that consumers are wllng to pay premum prces for organc foods, even those wth less than 100 percent organc ngredents. The magntudes of the WTP prema vared sgnfcantly among consumer groups. Generally, specalty grocery consumers were wllng to pay substantally more than tradtonal grocery shoppers. 16
However, ths group was no more (or less) wllng to pay a premum for the lowest organc content level (less than 70% organc ngredents). Ths may suggest that specalty grocery shoppers (perhaps representatve of the most knowledgeable and dedcated organc consumers) dsplay a threshold amount of organc content below whch they wll not pay premum prces. Health concerns were expected to be mportant determnants of WTP for selected food attrbutes. Surprsngly, wth all else equal, the level of health concern dd not mpact the magntude of prema the consumer was wllng to pay (except for the 95% organc content product). The lterature assessng the mpact of chldren n the household on WTP for organc produce s mxed. For mult-ngredent foods, we found that the presence of chldren n the household had no mpact on the probablty of beng wllng to pay a premum for organc foods. However, condtoned on a wllngness to pay a premum prce, famles wth chldren were wllng to pay hgher prema for foods wth 70-95% and 95-99% organc ngredents than were consumers wthout chldren. Fnally, we were nterested n the mpact of the NOP on consumer choce. One aspect of the NOP s the new organc seal that can appear on foods wth 95 percent or greater organc content. Only 45 percent of consumers (38% of tradtonal and 60% of specalty shoppers) recalled havng seen the NOP organc seal on food products n the past. Awareness of the NOP was sgnfcant as a postve shfter of the probablty that a consumer would be wllng to pay a premum for foods wth organc ngredents. It s nstructve to note that NOP seal awareness ncreased the lkelhood of a nonzero WTP 17
not only for the two products that can bear the NOP seal, but also for the two lower organc content level products that can not bear the seal. However, awareness of the NOP seal was not sgnfcant n explanng the magntude of premum pad. Though lkely not servng as a smple cue, the seal may be used by some consumers as a (government-backed) sgnal of product qualty. Our results suggest that the NOP provson for mult-ngredent foods wth varous levels of organc content does have value to consumers: Consumers can choose from among an array of products wth varyng organc content and prce to select the one that s utlty maxmal. Furthermore, consumer awareness of the NOP organc seal apparently s an mportant determnant of consumer wllngness to pay premum prces. Consumers who were aware of the NOP organc seal were more lkely to pay a premum prce for cereal (rrespectve of level of organc content) than were consumers who not aware of the NOP seal. However, awareness of the NOP seal dd not nfluence the amount of premum that these consumers would pay. Specalty food store shoppers were sgnfcantly more lkely to be aware of the NOP organc seal that were conventonal grocery shoppng consumers (60% vs. 38%). However, an NOP awareness-specalty store nteracton term was not statstcally sgnfcant n any model, suggestng that the dfferences observed n specalty store shoppers are not due strctly to ncreased awareness of the NOP by these consumers. These results also suggest the merts of an ncrease range of producton and prcng strateges for producers of mult-ngredent organc foods. Producng foods wth 100 (or even 95) percent organc content may be substantally more costly than foods wth lesser 18
organc content due to the hgh cost of dffcult-to-source organc ngredents. Producers may fnd t more proftable to produce for the lesser organc content categores rather than payng substantally hgher prces for selected nputs. In concluson, t appears that the NOP has made a sgnfcant mpact on the organc market even though a majorty of consumers appear to have lttle knowledge of ts provsons. The four organc product categores provde optons both for consumers and producers. The evdence s that there s a demand for ntermedate levels of organc content. As consumer knowledge of organc producton methods and potental advantages and dsadvantages of these products ncreases over tme, we expect that consumer WTP for alternatve level of organc content wll change. Lkewse, as the organc supply chan becomes more fully developed, allowng greater ease of sourcng a broad range of a product's ngredents, the costs of these hgher organc content products wll lkely decrease relatve to lesser organc content alternatves. 19
References Agrcultural Marketng Servce. 2000. Natonal Organc Program: Fnal Rule wth Request for Comments (7 CFR Part 205). Agrcultural Marketng Servce, Natonal Organc Program, U.S. Department of Agrculture, Washngton, DC. Baker, G. 1999. Consumer Preferences for Food Safety Attrbutes n Fresh Apples: Market Segments, Consumer Characterstcs, and Marketng Opportuntes. Journal of Agrcultural and Resource Economcs, vol. 24 ss.1: pp. 80-97 Buzby, Jean, Rchard Ready, and Jerry Skees. 1995. Contngent Valuaton n Food Polcy Analyss: A Case Study of a Pestcde-Resdue Rsk Reducton. Journal of Agrcultural and Appled Economcs, vol. 27 ss.2: pp. 613-625 Dmtr, Carolyn and Catherne Greene. 2002. Recent Growth Patterns n the U.S. Organc Foods Industry. U.S. Department of Agrculture, Economc Research Servce. Agrculture Informaton Bulletn Number 777. 2002. Dunlap, Rley and Curts Beus. 1992. Understandng Publc Concern about Pestcdes: An Emprcal Examnaton. Journal of Consumer Affars, vol. 26 ss.2: pp. 418-438 Fetter, T. Robert, and Jule A Caswell. 2002. Varaton n Organc Standards Pror to the Natonal Organc Program. Amercan Journal of Alternatve Agrculture, vol. 17 ss. 2: pp. 55-74. Food Marketng Insttute (FMI). 2003. SuperMarket RESEARCH. Vol. 5 ss. 3. (Fall 2003). Avalable onlne: http://www.fm.org/newsletters/uploads/supermarketresearch/fall2003.pdf 20
Govndasamy, R., J. Itala. 1999. Predctng Wllngness-to-Pay a Premum for Organcally Grown Fresh Produce. Journal of Food Dstrbuton Research July (1999): pp. 44-53. Haab, Tmothy C., McConnell, Kenneth E. 2002. Valung Envronmental and Natural Resources. Northampton, MA. Edward Elger Publshng Lmted. (2002): pp. 190-218. Huang, C. 1996. Consumer Preference and Atttudes toward Organcally Grown Produce, European Revew of Agrcultural Economcs, vol. 23 (1996): pp. 331-342. Hutchns, R.K., and L.A. Greenhalgh. 1995. Organc Confuson: Sustanng Compettve Advantage. Nutrton and Food Scence, No. 6: pp. 11-14. Lourero, Mara L. and Susan Hne. 2002 A Comparson of Consumer Wllngness to Pay for A Local (Colorado-Grown), Organc, and GMO-free Product. Journal of Agrcultural & Appled Economcs, vol. 34 ss. 3: pp. 477-487. Natonal Organc Program. 2003. Background Informaton: What s n the NOP Regulatons? Accessed at http://www.ams.usda.gov/nop/factsheets/backgrounder.html on 8/1/03. Organc Trade Assocaton. 2000. Food Facts, Accessed at http://ota.com/foodfacts.htm on 8/1/03. Organc Trade Assocaton. 2004. "The OTA 2004 Manufacturer Survey Overvew." Accessed at http://www.ota.com/pcs/documents/2004surveyovervew.pdf on 6/12/2006. 21
Packaged Facts. 2004. "The U.S. Market for Organc Food and Beverages: The Manstreamng of a Trend". Accessed at http://www.packagedfacts.com/pub/977845.html on January 1, 2005. Petty, Rchard E. and John T. Cacoppo. 1986. Communcaton and Persuason: Central and Perpheral Routes to Atttude Change, New York, NY: Sprnger/Verlag. Suryanta, K. 1999. Products from Paradse: The Socal Constructon of Hawa Crops, Agrculture and Human Values vol. 17 (1999): pp. 181-189. Thompson, Gary D. 1998. Consumer Demand for Organc Foods: What we know and what we Need to Know. Amercan Journal of Agrcultural Economcs, vol. 80 ss. 5: pp.1113-1118. Thompson, G.D. and J. Kdwell. 1998. Explanng the Choce of Organc Produce: Cosmetc Defects, Prces, and Consumer Preferences. Amercan Journal of Agrculture Economcs vol. 80, ss. 2 (May 1998): pp. 277-87. Underhll, Shela E., and Enrque E. Fgueroa. 1996. Consumer Preferences for Non- Conventonally Grown Produce. Journal of Food Dstrbuton Research vol. 27: pp. 56-66. Wang, Q. and Sun, J. 2003. Consumer preference and demand for organc food: Evdence from a Vermont survey. Paper prepared for Amercan Agrcultural Economcs Assocaton Annual Meetng July (2003): pp. 1-12. Whole Foods Market Inc. 2003. One Year after USDA Organc Standards are Enacted More Amercans are Consumng Organc Food. Press Release, October 14 th. 3 pp. 22
Whole Foods Market Inc. 2004. Organc Food Contnue to Grow n Popularty Accordng to Whole Foods Market Survey. Press Release, November 17 th. 2 pp. 23
Wllams, Pamela R. D., and James K. Hammt. 2000. A Comparson of Organc and Conventonal Fresh Produce Buyers n the Boston Area. Rsk Analyss vol. 20 ss. 5: pp. 735-746 Wllams, Pamela R. D., and James K. Hammt. 2001. Perceved Rsks of Conventonal and Organc Produce: Pestcdes, Pathogens, and Natural Toxns. Rsk Analyss vol. 21, ss. 2: pp. 319-330 24
Table 1. Characterstcs of Sampled Customer Households. Characterstc Tradtonal Grocery Specalty Grocery Sample Sze 199 102 Age (years) 43.0 39.8 Percent female 69.7 79.0 Percent prmary food shopper 79.8 83.8 Percent vegetaran or vegan 4.1 26.0 Number n household 3.1 2.7 Percent of households wth chldren: 59.5 32.7 Educaton Percent Less than hgh school graduate 7.2 1.0 Hgh school graduate (or equvalency) 27.3 6.0 Some college, no degree 27.3 17.0 Assocate degree 8.3 5.0 Bachelor's degree 18.0 36.0 Graduate or Professonal degree 11.9 35.0 Race/Ethncty Percent Black or Afrcan Amercan 31.1 1.0 Amercan Indan or Alaska natve 1.5 1.0 Natve Hawaan or other Pacfc Islander 0.5 0.0 Hspanc / Latno 0.0 2.0 Whte 66.8 96.0 Mean Household Income $65,253 $74,304 Medan Household Income $42,500 $62,500 25
Table 2. Wllngness to Pay for Selected Breakfast Cereal Characterstcs. Cents per box Characterstc 24 49 74 None 1-9 10-25- 50-75- 99 > 100 Mean Premum* Medan Premum* Tradtonal Grocery Percent Pestcde Free 18.3 14.2 18.9 15.4 8.9 7.1 17.2 32.8 10.0 Enhanced Flavor 32.3 11.8 19.3 14.3 8.1 8.7 5.6 21.8 10.0 Genetcally Modfed Free 42.9 16.8 11.8 9.9 5.0 6.2 7.5 18.4 1.0 100% Organc Ingredents 28.4 16.0 11.8 13.0 10.7 9.5 10.7 27.7 10.0 At Least 95% Organc Ingredents 32.1 19.1 10.5 14.2 9.9 9.3 4.9 21.6 1.0 70-94.9% Organc Ingredents 39.6 17.0 10.7 14.5 8.2 6.9 3.1 17.3 1.0 Less than 70% Organc Ingredents 46.0 18.6 13.7 9.3 4.4 5.6 2.5 12.7 1.0 Locally Grown 23.8 17.7 18.3 11.6 11.0 9.2 8.5 25.8 10.0 Specalty Grocery Percent Pestcde Free 5.1 10.2 12.2 14.3 19.4 15.3 23.5 49.5 50.0 Enhanced Flavor 40.6 11.5 10.4 15.6 6.3 4.2 11.5 22.8 1.0 Genetcally Modfed Free 28.3 12.0 8.7 12.0 7.6 8.7 22.8 37.1 25.0 100% Organc Ingredents 9.3 6.2 11.3 16.5 10.3 19.6 26.8 52.0 50.0 At Least 95% Organc Ingredents 17.7 9.4 11.5 22.9 14.6 17.7 6.3 33.8 25.0 70-94.9% Organc Ingredents 26.0 13.5 20.8 19.8 14.6 4.2 1.0 18.6 10.0 Less than 70% Organc Ingredents 41.9 25.8 18.3 5.4 8.6 0 0 7.7 1.0 Locally Grown 14.4 13.4 11.3 12.4 16.5 9.3 22.7 42.2 25.0 * Each premum category s valued at ts lower bound and s measured n cents per box above $3.00 for a conventonal product. These are mnmum wllngness to pay measures.
Table 3: Probt Model to Explan Wllngness to Pay a Premum for Food Attrbutes. a Pestcde Enhanced GM 100% 95% 70-95% <70% Locally Free Flavor Free Organc Organc Organc Organc Grown Constant 0.146 0.280-0.207 0.434-0.041-0.188-0.333 0.051 Age 0.014 0.003-0.003 0.004 0.007 0.008 0.003 0.003 Income per Person n Household 0.003 0.007 0.003 0.001 0.003-0.001 0.006 0.002 Chldren Present n Household (Yes=1) -0.397-0.103 0.155-0.080-0.142-0.158 0.133 0.104 Post-Hgh School Educaton (Yes=1) 0.435-0.189 0.065 0.158 0.077-0.126-0.512-0.134 Health Index (100 s healthest, 0 least) 0.001 0.006 0.004 0.005 0.005 0.007 0.007 0.003 Race (Whte=1) -0.006-0.343-0.135-0.451-0.249-0.260-0.093 0.409 Gender (Female=1) -0.214 0.028 0.107-0.253-0.080 0.025 0.050-0.083 Specalty Store?(Yes=1) 0.521-0.021 0.435 0.825 0.549 0.468 0.257 0.286 Aware of NOP Seal? (Yes=1) 0.431 0.062 0.350 0.360 0.347 0.465 0.332 0.408 Observatons 240 238 232 242 237 236 234 240 a The dependent varable s one f WTP>0, and s zero otherwse. *Bold entres are sgnfcantly dfferent from zero at the 90% confdence level
Table 4: Interval (Group) Censored Model to Explan Determnants of Wllngness to Pay Prema Condtonal on Postve WTP. Pestcde Enhanced GM 100% 95% 70-95% <70% Locally Free Flavor Free Organc Organc Organc Organc Grown Constant 2.683 2.274 0.863 1.806 2.403 2.002 1.384 2.845 Age 0.017 0.013 0.023 0.012 0.001 0.012 0.014 0.008 Income per Person n Household 0.006-0.002 0.010 0.009 0.005 0.008 0.000 0.001 Chldren Present n Household (Yes=1) 0.172 0.197 0.028 0.224 0.446 0.299 0.366-0.066 Post-Hgh School Educaton (Yes=1) 0.084 0.146 0.275 0.155-0.199-0.319-0.038 0.223 Health Index (100 s healthest, 0 least) -0.006-0.003 0.002 0.003 0.007 0.005 0.002 0.003 Race (Whte=1) -0.500 0.007-0.376 0.017-0.226-0.187-0.006-0.726 Gender (Female=1) 0.489 0.623 0.999 0.473 0.372 0.293 0.538 0.355 Specalty Store?(Yes=1) 0.517 0.146 0.867 0.798 0.677 0.399-0.095 0.596 Aware of NOP Seal? (Yes=1) 0.181 0.089 0.355 0.133 0.095 0.101 0.109 0.225 Standard Devaton of Error Term 1.198 1.064 1.421 1.173 1.044 0.962 1.107 1.220 Observatons 208 157 145 189 174 155 130 192 *Bold entres are sgnfcantly dfferent from zero at the 90% confdence level
Table 5. Wllngness to Pay Estmates for Eght Food Attrbutes. WTP a -- Cents per Box of Cereal Attrbute Medan Mean Std. Dev. Mnmum Maxmum Pestcde Free 39.39 43.15 20.56 13.07 125.40 Enhanced Flavor 31.87 30.58 10.25 10.65 63.18 GM Free 31.54 38.72 29.06 3.99 147.30 100% organc 39.99 45.43 25.34 8.79 126.94 95% organc 30.09 32.53 13.97 9.15 103.85 70-95%+ organc 23.98 25.11 9.46 9.51 74.51 <70% organc 14.71 15.04 5.50 5.64 32.69 Locally Grown 33.48 35.70 14.77 11.81 96.83 a Wllngness to pay s calculated for sample observatons usng the equatons estmated n Table 4. 29
Fgure 1. Natonal Organc Program Seal 30
Endnotes 1 The reader s cautoned that ths method does not comprse a random sample, and thus the results can not be generalzed to the populaton of all consumers. Although consumer selecton s random, the store and contact tme are not randomly selected. A face-to-face ntervew of consumers was needed to accomplsh all goals of ths research project. Ths method ensures that the questons are completed by someone wth food shoppng responsbltes n the household and adds realsm to the food selecton experment featured durng the ntervew. Although we beleve that the sample dentfed s typcal of Oho food shoppers, results can be generalzed only to shoppers of these stores on the days and tmes sampled. 2 LIMDEP 8.0 provdes a standard routne for estmatng nterval censored data models wth the GROUPEDDATA command. 31