Consumers’ Willingness to Purchase Locally Produced ...ageconsearch.umn.edu/bitstream/31329/1/29010043.pdfConsumers’ Willingness to Purchase Locally Produced Agricultural Products:

Consumers’ Willingness to PurchaseLocally Produced AgriculturalProducts: An Analysis of anIndiana Survey

Mark D. Jekanowski, Daniel R. Williams II, and William A. Schick

Using a survey of over 320 consumers from across the state of Indiana, we estimate anordered probit model to determine the demographic and attitudinal factors which are most

important in predicting the likelihood of consumers to purchase products that are produced

within the state. Our results indicate that the willingness to purchase [ocafly produced

agricultural products increases with time of residency in the state, and we find a greater

tendency for female consumers to purchase such products. We also find that quality

perceptions play a critical role in these food purchase decisions. We underscore the

importance of maintaining minimum quality standards to maximize the effectiveness of state

level agricultural promotion programs.

The highly competitive nature of agricultural com-modity markets has since the early days of com-mercial agriculture challenged legislators to devisemethods of increasing or stabilizing farm incomes.While the bulk of agricultural poIicy is still imple-mented at the federal level, efforts at the state levelto protect and promote local agricultural interestshave existed for more than 60 years. State govern-ments have been involved in the advertising andpromotion of agricultural products since the 1930s(Halloran and Martin 1989).

Advertising by commodity groups and market-ing boards tends to focus on increasing consumerdemand for a single product, such as pork, milk orpotatoes. Ward, Chang and Thomson (1985) referto this as “generic advertising.” These campaignscould either be supported nationally, as are the“Pork-The Other White Meat” and the “milkmustache” campaigns, or by producer groupswithin a particular state, like “Washington Apples”and “Idaho Potatoes.” Funding usually comes di-

Mark P. Jekanowski is an agricultural economist with the EconomicResearch Service, USDA; Daniel R. Williams U is an economist withAgriculamd Marketing Service, USDA; and William A. Schick is aneconomist with the Dairy Institute of California. Comments from KevinKesecker of USDA/AMS, and two anonymous reviewers are gratefullyacknowledged. Partial funding for this project was provided by Indiana’sValue Added Grant Program.

rectly from the producers through a checkoff pro-gram.

Growing in popularity are state-funded pro-grams aimed at promoting, or at least identifying,all agricultural products produced within the state;adding a “family branding” aspect to generic pro-motions which typically focus on a single com-modity. These programs are expected to grow inpopularity as agricultural markets become increas-ingly global, and U.S. producers face greater com-petition from Latin and South American countries,especially in the fresh fruit and vegetable markets.Liberalized trade policies have already promptedsome producer groups to request mandated “coun-try of origin” labeling, and while the impetus hereis ostensibly related to food safety issues, the sidebenefit of increased demand owing to consumers’loyalty to U.S. products is most certainly well-known and appreciated. State promotion programscan be viewed as a type of “state of origin” label-ing initiated to protect local producers from inter-state competition by capitalizing on consumers’loyalty to their state of residence. By differentiat-ing products produced within the state, these pro-grams could complement national promotion ef-forts which tend to be aimed at building consumerawareness to particular agricultural commodities.

States tend to view agricultural promotion pro-

Agricultural and Resource Economics Review 29/8 (April 2000) 43-53Copyright 2000 Northeastern Agricultural and Resource Economics Association

44 April 2000 Agricultural and Resource Economics Review

grams as relatively inexpensive means by which tostimulate economic activity, especially in rural ar-eas. They are generally intended to appeal to con-sumers who have an interest in supporting localagriculture. The popularity of these programs hasbeen increasing, with at least 23 states activelyinvolved in promoting their own agricultural prod-ucts, and several others considering the introduc-tion of such a program (Williams 1995). Budgetcommitments vary by state, with some spending aslittle as $80,000 on agricultural promotion, andothers spending over $2.0 million (Williams 1995).

Examples of some popular state programs in-clude: “Ohio Proud,” “Jersey Fresh,” and “Virgin-ia’s Finest.” Programs such as these usually relyheavily on the use of a standardized logo or slogan,which can be displayed on point-of-purchase(POP) materials supplied to supermarkets, in salesflyers distributed by individual supermarkets, orvia television or radio advertisements. Three keyfunctions of state promotion programs are the fol-lowing: expanding consumer awareness of state-sourced products, motivating consumers to buystate-sourced products, and establishing new mar-kets or expanding existing markets domesticallyand/or internationally (Williams 1995).

Perhaps one of the most successful state brand-ing programs in recent years has been the “JerseyFresh” campaign, administered by the state of NewJersey, It was started in 1983 with a budget ofapproximately $350,000, and is credited with dou-bling consumer awareness of New Jersey agricul-ture in the first year of operation (Brown 1988).Adelaja et al. (1990) found that the “Jersey Fresh”tomato has a more inelastic demand with respect toprice, a more elastic income response, and fewersubstitutes relative to tomatoes produced in otherstates. Arizona’s recently implemented promotionprogram, “Arizona Grown,” has so far not shownthe same level of success. Patterson, et al. (1999)find little evidence that sales have increased orconsumer preferences have changed as a result ofthe campaign. This emphasizes the need for de-tailed knowledge of the factors that might contrib-ute to the success of promotion programs imple-mented at the state level.

Despite the growing interest in these programsamong state legislators, little research has beenconducted to determine the likelihood of success orthe factors that could influence a successful imple-mentation. This knowledge could be useful foridentifying and targeting those consumers mostwilling to build loyalties to commodities producedlocally, thereby ensuring the largest return possiblefrom the promotion budget. This is especially im-portant given that these promotions are often criti-

cized as an inefficient use of resources. For ex-ample, many researchers contend that state-levelpromotional efforts should be abandoned either infavor of multi-state efforts (e.g. Halloran and Mar-tin) or research to reduce costs and increase pro-duction efficiency (Wohlgenant 1993).

We do not focus here on the normative aspectsof the optimal distribution of government re-sources, rather we aim to identify factors thatmight be important in determining consumer de-mand for locally produced agricultural products.This type of information could be useful in pro-gram design since the success of any advertisingcampaign is tied to its ability to target those con-sumers who have the highest likelihood of pur-chase.

Conceptual Framework of AgriculturalPromotion Initiatives

As Waugh pointed out over 30 years ago, thegeneral purpose of commodity advertising is tochange the shape of the demand curve. Price ad-vertising for generic commodities, which is com-mon among chain stores and supermarkets, has theeffect of increasing demand elasticity by encour-aging greater consumption as soon as the pricedrops, Advertising, which attempts to differentiateproducts, has the opposite effect: it induces con-sumers to build loyalties to particular brands, de-creasing demand elasticity and possibly allowingthe seller to receive a price premium over the non-differentiated product. State sponsored promo-tional programs fall into this second category,

Building a brand image which effectively differ-entiates food products by their place of origin canbe a daunting task, since the market for fresh foodcommodities (especially fruits and vegetables) isoverwhelmingly price competitive. While somenational fresh food brands exist (e.g. Dole, GreenGiant, Chiquita), and some producer groups havesuccessfully differentiated particular commoditiesby their place of production (e.g., Florida OrangeJuice, California Raisins, Washington Apples),most fresh produce commodities are homogeneousand compete primarily on price and easily percep-tible quality differences. The wholesale marketsare predominantly generic, and quality differencesat the retail level tend to be attributed to the repu-tation of the grocery store. The challenge for statepromotion programs is to provide local producers avehicle for building consumer loyalty, so that pur-chase decisions may be based on attributes otherthan price and easily recognized quality differen-tials. This is especially important if the local pro-

Jekanowski, Williams, and Schick Consumers’ Willingness to Purchase 45

ducer faces a comparative disadvantage in produc-tion.

States that actively promote their own agricul-tural products do so in an attempt to develop alocal brand to compete against, what is in essence,a national generic (unbranded) product. The ge-neric brand appeals to the price shoppers, but thelocal brand will appeal to those shoppers who havea propensity to build brand loyalties, or who oth-erwise value the information provided in the ad-vertisement. An example could be the “JerseyFresh” tomato placed beside an unbranded tomatofrom Florida: if the promotion is successful thelocal brand will be more appealing to some con-sumers, either because of real or perceived differ-ences in quality. State promotion programs can in-fluence consumers’ perceptions of local agricul-tural products by providing information aboutsuperior quality or freshness, or they may simplyappeal to the parochial interests of consumerswishing to support local agricultural industries.The important assumption is that demand is notonly a function of prices and income, but also de-pends upon consumer preferences and perceptions,which could vary across consumers, and whichmight be influenced by advertising.

Not all consumers are equally susceptible to themessages conveyed by this form of advertising.The competing interests of price shoppers andbrand loyal consumers can be modeled conceptu-ally as a consumer duopoly model, similar to thatillustrated by Rao (1991). Consider a market withtwo brands of a particular product. Brand 1 is anationaJ, undifferentiated brand, while brand 2 isproduced locally and is labeled as such. Assume amarket of fixed size D, composed of segments Aand B. Segment A consists of price shoppers withno brand preferences, and it constitutes a fraction q

of the market, where O s q s 1. Segment B cor-responds to the remaining (1 – q) of the market.

Consumers in segment A choose brands basedon the following rule:

(1) pi <Pj +choosebrmdi, i= 1,2, j=3-i

i.e.—they are price shoppers. Consumers in seg-ment B choose products based both on price andthe relative preferences for each of the two brands,where ~k represents consumer k’s preference forthe local brand (brand 2). The choice rule for con-sumer’s in segment B is then:

(2) p2s PI + & -+ choose local brand

(3) p2 > p, + ~k + choose national generic brand

tl~therefore corresponds to a price premium com-

manded by the local brand. Consumers in segmentA are willing to pay no premium.

Heterogeneity is introduced into the consumerpopulation by allowing the proportion of consum-ers in segment B with 8 s x (a given premium) tobe denoted by F(x) where:

(4) F(x) = ~~j(8)d8 ~(x) >0, r < x < u

F(r) = O, F(u) = 1.

No consumer is willing to pay a premium greaterthan w and all consumers (in segment B) are will-ing to pay a premium of at least r (which could bezero) for the local brand. Therefore, if the localbrand (brand 2) charges a premium of x over thenational generic brand (brand 1), its share of con-sumers in segment B is given by (1 – F(x)): theproportion of consumers willing to pay a premiumgreater than or equal to x, The sales of the localbrand (denoted S~) and the national generic (de-noted S~) given a premium x are defined as:

(5) SL= D[(l - q)(l - F(x))+ q](x)],

(6) s~ = D[(l - g)F(x) + q(l - I(x))],

where Z(x) is an indicator variable given by:

(7) I(x) = 1 if xs O, Z(x)= Ootherwise.

If x <0, both consumer segments (A and B) pur-chase the local brand exclusively. If O< x c r, thelocal brand receives all of segment B but none ofsegment A, and if x > r, segment B is dividedamong the two brands according to F(x) above. Ifx > u, no consumer purchases the local brand.

When a state considers implementing an agri-cultural promotion program, it is assuming thatconsumer segment B comprises some significantproportion of the market, and that these consumersare willing to pay a premium (8 a x) high enoughto cover the expense of the program and any pos-sible comparative disadvantages in productionwhich might result in a higher price for the localbrand. The state promotion effort serves to ensuremaximum sales by identifying local products tothose consumers in segment B, and it might alsochange consumer preferences in favor of the localbrand-effectively increasing 8 in the above ex-ample. The following analysis focuses on deter-mining the attitudes, purchasing patterns, and de-mographics that best characterize consumers insegment B, i.e,,—those that are most likely to ex-hibit brand preferences for locally produced agri-cultural products. The results will aid state agen-cies in developing programs that most effectivelyreach their target market.


Model Specification

Consumer purchasing behavior is assumed to be afunction of several factors, including perceptionsof the quality and value of the product in question,prior shopping experiences, the degree to whichconsumers build loyalties to particular brands ofproducts, as well as the demographic compositionof the household. Unfortunately, obtaining detaileddata on actual consumer purchases of agriculturalcommodities is difficult, Most grocery stores usePLU (price look-up) codes to identify the price ofparticular produce items, but unlike UPC codes,these provide little information other than the typeof product, so purchases of local products are oftennot distinguishable from purchases of productsproduced elsewhere. A consumer survey is one ofthe few tools available to extract information aboutconsumer preferences for locally grown produce.

We believe that the likelihood of purchasing ag-ricultural products advertised as being producedin-state can appropriately be modeled as follows:

(8) Likelihood of Purchasing Locally ProducedAgricultural Commodities =

f(Perceptions of price and quality oflocally produced products, typically shopping

behavior, family income, demographiccharacteristics of the household).

Data for the variables in this model (described be-low) were collected as part of a consumer surveyconducted in Indiana. Several variables were usedto represent the right-hand-side factors. Specifi-cally, the model was estimated using the followingset of survey-response variables:

(9) Likelihood of Purchasing Locally ProducedAgricultural Commodities =

f(Perceptions of price of local produce,perception of quality of local produce, number of

visits over the past year to local farmersmarkets, degree of brand loyalty when shopping,

importance of product freshness, annualhousehold income, time as a resident of Indiana,

family size, education level of primary foodshopper, gender of primaiy food shopper, degree

of urbanization where respondent resides).

The survey questions are described in more detailbelow.

Data

The survey data used here comes from a largerstudy funded by the state of Indiana to investigatethe feasibility of a statewide identification, brand-

ing, and promotional program for Indiana-sourcedfood and agricultural products. This study includeda consumer telephone survey, part of which wasaimed at determining the factors which would mo-tivate consumers to purchase products identified aslocally produced. 1

The survey was conducted by IUD Marketing,in December 1994. Survey participants were cho-sen based on a random sample of residents from allcounties in Indiana, weighted by population, A to-tal of 498 people were surveyed, with 324 of therespondents providing enough information to beused in our empirical analysis. The questions ad-dressed consumer food buying behaviors, percep-tions of the quality and value of products producedin Indiana, the likelihood of purchasing locallyproduced food products at the grocery store, anddemographic characteristics of the household. Thefocus of the survey was both locally produced pro-cessed food products, such as bakery items, icecream, and wine; and unprocessed agriculturalcommodities such as tomatoes, melons, etc. Theresults are therefore not limited only to “produce,”but instead are applicable to any segment of thefood industry that hopes to capitalize on state loy-alty by identifying itself as being a “local” pro-ducer.z

One of the first questions of the survey was, “IfIndiana produced food products were clearly iden-tified and offered where you shop, how likelywould you be to purchase these products’?”Partici-pants were asked to rank their likelihood on a scalefrom 1 (not at all likely) to 5 (extremely likely).The response served as the dependent variable inthis analysis, but to make interpretation of the re-sults easier, we collapsed the responses into threecategories: 1 = “unlikely,” 2 = “neutral to some-what likely,” and 3 = “highly likely.”3

Responses to other survey questions served asthe independent variables. Two variables con-

] The complete survey was intended to identify current consumer at-titudes towwds a statewide promotionibranding program for food athome and away, and to collect opinions concerning various proposedstate slogans and logos. For this study we focused on the variables thatwe believed were most relevant to identifying consumer attitudes to-wmds Indiana agricultural products at the groce~ store, with the leastamount of redundancy (several of the questions were either open-ended,or were asked in such a way to be highly correlated with other questionsin the survey), A copy of and descriptive analysis of the complete surveyis available from tbe authors upon request.

2 For food processors, it is generally required that the agriculturalinputs be sonrced exclusively from with the state in order for tbeprocessed product to carry the state identification logo.

3 There was practically no difference in the pmameter estimates (signand significance) resulting from the use three categories in lieu of five,and most would agree that interpretation is easier with only three levelsof likelihood. Results based nn the five categories are available from theauthors upon request.


trolled for the perception of the quality and price offood products produced locally. Respondents wereasked to rank their quality perception of locallyproduced products (i.e., produced in Indiana),compared to other state’s products, on a 5 pointscale, A response of 5 meant the products wereperceived to be much higher in quality, 1 meantmuch lower in quality, with 3 meaning no qualitydifference. A similar question addressed price per-ceptions: a response of 5 meant the consumer be-lieved locally produced products tend to be muchhigher in price than those from other states, and 1meant they tend to be much lower in price. Again,a response of 3 meant the consumer believes nonoticeable price differential exists. We expect thathigher perceptions of quality, and lower percep-tions of price, will have a positive effect on like-lihood of purchase.

Three variables controlled for shopping behav-ior. Respondents were asked the approximate num-ber of times in the past year that they shopped at alocal farmers’ market, corresponding to the follow-ing four categories: zero, 1–5 times, 6–10 times,and greater than 10 times. This variable was codedto correspond to the midpoints of these responsecategories: O, 3, 7, and 11. We expect that thelikelihood of purchasing local products increaseswith increased frequency of visits to farmers’ mar-kets, since these people would tend to have agreater awareness of the products produced withinthe state, and shopping at farmers’ markets couldindicate a willingness to search for products pro-duced locally.

The respondent was also asked to rank his or hertypical degree of brand loyalty, where a responseof 1 meant the respondent is never brand loyal (i.e.,consumer is a price shopper), and 5 meant the re-spondent always tries to purchase particular brandsof products. We expect that a tendency towardsbrand loyalty will increase the likelihood of pur-chasing local products if they are so labeled.

For the final measure of shopping behavior, re-spondents were asked to rank the relative impor-tance they place on product freshness when shop-ping for groceries, with 5 meaning they are ex-tremely concerned about product freshness, and 1meaning product freshness is not a serious factor,We expect that people concerned about productfreshness will have a greater likelihood of purchas-ing locally produced agricultural products, espe-cially produce, given the perception that the timebetween harvest and delivery to the food store isminimized for products produced locally.

Income is practically always expected to play arole in consumption decisions, and here is no ex-ception. People with higher incomes have a higher

opportunity cost of time, and therefore tend to bemore susceptible to advertising, especially if it isinformative, Since labeling of locally producedproducts is a form of advertising, we expect thelikelihood of purchasing these products to in-crease with income. Respondents were asked tostate their annual household income in one of eightcategories, the lowest being under $10,000, thehighest over $75,000.

Several variables control for demographics, Thelength of time the respondent was a resident of thestate is expected to have a positive effect on thelikelihood of purchase, since a stronger identitywith the state is expected to be built over time,along with state pride, a greater interest in thestate’s economy, and possibly a desire to “helpneighbors.” Other variables include family size andthe level of education of the primary food shopperin the household (in one of five ranges, from somehigh school or less, to graduate work or degree),and the respondent’s age. A binary indicator con-trols for the gender of the respondent (1 = male, O= female). We have no strong expectations con-cerning the effect of these variables. Finally, re-spondents were asked the type of area which bestdescribes the location of their home, in increasingorder of urbanization, with 1 meaning “rural,” 2“small town,” and 3 “city.” We expect, if anything,for rural consumers to have a greater likelihood ofpurchasing local agricultural products since thesetend to have a greater awareness, and possibly ap-preciation for, the local agricultural economy.

The survey questions used in this analysis, alongwith descriptive statistics of the responses, are pre-sented in the appendix.

Model Estimation

We estimated the model described in equation 9,assuming it to be linear in the independent vari-ables. Due to the special nature of the dependentvariable (categorical), OLS cannot appropriatelybe applied to this model. Heteroscedasticity wouldlead to inefficient parameter estimates, and thenon-normal distribution of El would invalidateclassical tests of significance. In addition, OLScould result in predicted probabilities greater thanone or less than zero. An ordered probit or orderedlogit model will provide greater generality by as-suming each level of the dependent variable to bea censored apportionment of a continuous distribu-tion of purchase likelihoods. These models arenon-linear, and the area of the distribution associ-ated with each level of the dependent variable issimply a probability.


We choose the ordered probit specification.4Along with the variable coefficients, a thresholdparameter used to segment the distribution of like-lihoods into the three categories is estimated withinthe model. The probability associated with eachcategory is its respective area under a normalcurve, calculated as follows (Greene 1991):

(lo) Prob[y = 1]= @(-~’x),Prob[y = 2] = @(p,-~’x) - cP(~’x)Prob[y = 3] = 1- cI@ - (3’x)

where ~ is the threshold parameter mentionedabove, @ is the cumulative normal, and x is thevector of independent variables, which we will setequal to the mean. Maximum likelihood estimationwas conducted using LIMDEP econometric soft-ware (Greene 1996),

Model Results

The parameter estimates for the model estimatedabove are presented in table 1. No direct measureof goodness-of-fit, such as R*, is computable forthis type of model, However, a pseudo-R2 can becalculated based on the ratio of the unrestricted andrestricted log-likelihood values (Long 1997). Fol-lowing Ben-Akiva and Lerman (1985), and con-trolling for the number of regressors in the model,the pseudo-R2 is 0.11. Probit model significance isverified through a chi-squared test of the differencebetween the restricted and unrestricted log likeli-hood values. Here, with 11 degrees of freedom, thechi-square statistic is 29.07 and highly significant.

Table 2 presents the estimated probabilities as-sociated with each level of the dependent variable,calculated using the parameter estimates from ta-ble 1. The parameter estimates represent the mar-ginal effect of a change in an independent variableon the probability distribution of the dependentvariable, i.e., the probabilities in table 2. In otherwords, a positive parameter estimate implies thatart increase in that variable shifts the distributiontowards the right, resulting in an unambiguous in-crease in the area (probability) associated withhighest category (Y = 3), and an unambiguousdecrease in the probability for the lowest category(Y = 1). Depending on the threshold parameterestimate, the marginal effect on the middle cat-egory is ambiguous, However, in general, the signsof the parameter estimates are indicative of theeffect that each variable has on the likelihood ofpurchasing locally produced agricultural products;

Table 1, Parameter Estimates from OrderedProbit Model Predicting the Likelihood ofConsumers to Purchase Locally ProducedAgricultural Products

ParameterEstimate t-ratio

Intercept -0.6603 -0.730Price and Quality Perceptions:

Perception of Quality 0.3333 2,712**Perception of Price –0.0093 -0.068

Shopping Behavior:Number of Visits to Farmers

Markets 0.0172 0.961Degree of Brand Loyalty 0.1180 1.079Importance of Freshness 0,1190 1.015

Income:Household Income 0,0941 2.016**

Demographics:Time as a Resident 0.0255 1.981**Family Size 0.0110 0.404Level of Education -0.1431 –2.023**Gender (O = Female, 1 = Male) -0.2942 -1.763*Degree of Urbanization 0.1126 1.207

Threshold Parameter Estimate:

w, 1.8283 10.421**

n = 324, ~zl, = 29.07**

Single and double asterisks denote significance at the 10 ands~o leVelS, respectively.

positive coefficients imply a greater likelihood,and vise-versa.

The probabilities in table 2 indicate a high like-lihood of purchasing locally produced food prod-ucts when that option is available, Nearly 609Z0ofthe probability distribution is associated with thecategory representing a positive likelihood of pur-chase (y = 3). Most of the remaining distribution(just under 40%) is captured by the category rep-resenting neutral or somewhat likely (y = 2). Theprobability associated with a resident being un-likely to purchase locally produced food products(y = 1), given the variables in the model, is onlyabout 270 (table 2). This is evidence that consum-ers have a tendency to favor the local brand whenthat option is available.

Table 2. Estimated Probabilities fromOrdered Probit Model, of the Likelihood ofPurchasing Locally Produced AgriculturalProducts in a Grocery Store.

probability

4 The probit model assumes the errors to be nnrrnaily distributed,while the logit assumes a logistic distribution. Both assume K-( O,U). Inpractice, model choice tends to have little effect on the results, which wasalso the case here.

Highly likely to purchase local food products(Y = 3) 0,5879

Neutral or somewhat likely (Y = 2) 0.3919Unlikely to purchase local food products

(y = 1) 0.0202


The negative parameter estimate for the genderindicator variable (table 1) indicates that femalestend to be more likely than males to purchase foodproducts produced locally. Since this is a binaryindicator variable (where male = 1), it is informa-tive to calculate the predicted probabilities sepa-rately for both males and females. This can be doneby calculating the probabilities with the gendervariable equal to zero for females, and to one formales, instead of holding it at its mean. For fe-males, the probabilities associated with the threeincreasing levels of likelihood (y = 1, y = 2, y =3) are: 0.017,0.369, and 0.614, respectively; whilefor males they are 0.034, 0.468, and 0.499. It isevident that compared to males, females have botha higher probability of being likely to purchase,and a lower probability of being unlikely. The im-plication for states is that programs to promote orbuild awareness of local agricultural productscould have a larger effect when targeted towardsfemale consumers. This suggests that to maximizethe promotion effectiveness, advertising strategyshould focus on themes that appeal to female con-sumers, through media with large female audi-ences.

It is not surprising that the perception of thequality of local produce is positively related to thelikelihood of its purchase. This emphasizes the im-portance of maintaining minimum quality stan-dards, and suggests strong results from advertise-ments which promote the quality aspects of foodproducts produced locally. Building a brand imagebased on tangible quality differences will have astrong likelihood of success, However, if the qual-ity of local food products is comparatively low,due to negligence or mismanagement at any levelof the marketing chain, a state branding and iden-tification progrfi could have the opposite effect:the positive parameter estimate also indicates that alow perception of quality will decrease the likeli-hood of purchase.

The perception of the price of local produce,relative to that from other states, does not have asignificant effect on the likelihood of purchase.The mean of this variable was 2.81 (see appendix),indicating that on average consumers expect fooditems produced locally (i.e., in-state) to be slightlyless expensive than products from other areas. Weknow that this will not always be the case, sincealthough transportation costs might be lower forproducts produced locally, quite often these pro-ducers operate at a comparative disadvantage toother areas. Fortunately, the parameter estimate forthis variable indicates that even consumers whoexpect to pay more for local food products do nothave a significantly lower probability of purchase.

This is evidence that price does not weigh heavilyin the decision to choose local produce over thatfrom other states. Therefore, place of origin can bea source of product differentiation, and if promotedin a favorable light, might allow consumers to jus-tify a higher expense for products grown in-state.

None of the variables controlling for consumershopping behavior were significant at traditionallevels. We expected each to have a positive effecton purchase likelihood, but while the parameterestimates for each of these variables is positive,none is significantly different from zero. We do notbelieve that this implies that these factors are notimportant, but rather it illustrates the difficulty inmeasuring these presumably important factors. Forexample, asking the respondent how often he orshe purchases the same brand might be a poorproxy for their actual degree of brand loyalty. Amore accurate measure could be developed by ac-tually tracking their purchase history over time, butthis expensive option was not feasible. In otherstudies of consumer behavior, brand loyalty is of-ten considered to be a powerful force, and is thebasis behind much of advertising. We thereforebelieve that developing a consistent, easily recog-nizable brand or logo is an important part of anystate promotion activity.

The variable measuring the relative importanceof “freshness” was not significant, but the surveyrespondents indicated that “freshness” was quiteimportant in their purchase decisions (the meanresponse was 4.4 on a scale of 5.0, see appendix),The lack of significance in the estimated modelsuggests that consumers do not expect local prod-ucts to on average be any more fresh than productsproduced elsewhere. But since freshness is impor-tant to consumers according to the survey results,the implication for marketers is that emphasizingthe superior freshness of locally produced productswould likely be beneficial. The speed with whichlocal products can move from the farm to the pro-duce section of a grocery store is perhaps one oftheir most distinguished characteristics, and shouldbe emphasized accordingly.

The number of visits to farmers’ markets ismuch less subjective, and in our model is not re-lated to the likelihood of purchasing local products.This might indicate a lack of awareness of farmers’markets, or that farmers’ markets are an inconve-nient option. Despite the desire to purchase localproducts expressed in our survey, consumers ap-pear to be unwilling to incur the search and timecosts involved in purchasing some of their fooddirectly from the farm, or from an organized farm-ers’ market. By segregating and identifying prod-ucts which are produced locally, groce~ stores


provide a service to consumers that reduces oreliminates these search costs.

The willingness of consumers to purchase localfood products increases with the length of time theconsumer has resided in the state. This is the rela-tionship we expected, since pride in one’s state ofresidence likely builds over time, along with theawareness of the local agricultural economy, bothof which would be expected to increase the likeli-hood of purchasing products produced locally, Apossible implication for marketers is that advertise-ments emphasizing nostalgia, tradition, and the ag-ricultural heritage of the state might produce supe-rior results.

Household income is positive and significant.The opportunity cost of shopping time increaseswith income, thus making the consumer more sus-ceptible to branding and advertising, and less sen-sitive to price. A program design implication is thatpoint-of-purchase displays and other advertisingtechniques will have the greatest effectiveness iftargeted at consumers in relatively wealthy areas.The level of education has a negative effect, whichimplies that with education consumers become lesssusceptible to advertising and branding, and aremore likely to differentiate products based on tan-gible quality characteristics and price. This mightalso indicate that highly educated consumers are atsome level aware of the concept of comparativeadvantage, recognizing that not all areas can pro-duce high quality products with the same degree ofefficiency. These consumers are less likely to beinfluenced by a state-level branding program.

Neither family size, nor the type of communitythat the respondent resides in (rural, small town,urban) are significantly related to the likelihood ofpurchasing locally produced food items. This indi-cates little need to segment the market or targetconsumers based on these attributes. The fact thatthe type of community in which the respondentresides is not significant suggests that consumersliving in the city are generally as willing to pur-chase local agricultural products as those in ruralareas, which is important since urban areas com-prise a large proportion of the population base ofany state. A well-developed state promotion pro-gram would be expected to have broad consumerappeal across much of the population base.

Conclusions

We estimated a probit model to identify consumercharacteristics that might indicate willingness topurchase agricultural products produced locally.This information could be useful in designing asuccessful state-sponsored agricultural promotion

campaign. Our data was from a survey of over 320randomly selected consumers from across the stateof Indiana, Although we focused on Indiana, wefeel our results are widely applicable to manystates considering such a program.

The probability estimates from the probit modelindicate a strong willingness to purchase localproducts if that option is available. This is evidenceof a latent demand for such products, which in-creases the likelihood that identifying and promot-ing agricultural products at the state level will af-fect sales, directly benefiting the state’s producers.An important conclusion is that loyalty to one’sstate of residence may already be, or could be-come, an important factor in a consumer’s pur-chase decisions for food.

In our model, we found that household income,the quality perception of Indiana agricultural prod-ucts, and the length of time that a consumer hasresided in the state are all positively related to thelikelihood of purchasing agricultural products pro-duced in-state. The level of education has a nega-tive effect. We also found significant differencesbetween male and female respondents, with fe-males having a greater likelihood of purchasinglocally produced agricultural products. The percep-tion of the price of local produce relative to thatfrom other states, the propensity to shop at farm-ers’ markets, the length of time the respondentlived in the state, family size, and the type of com-munity the respondent lived in (rural, small town,or urban) were all insignificant.

We emphasize building a strong brand image topromote local agricultural products. Basing thisimage on quality characteristics and the relativefreshness of such products is likely to have a strongeffect. The quality of the products is especiallyimportant, since the perception of quality wasfound to have the strongest positive effect on thelikelihood of purchase. Allowing quality to fall be-low that of other states could actually lead to anegative effect from the promotion campaign—byhelping to identify products of lower quality.

Due in part to relatively low costs, the popularityof state sponsored promotional programs for agri-cultural products across the United States is ex-pected to continue to grow. These programs cancomplement larger national programs. In some na-tional programs, such as dairy, smaller regionalprograms can qualify for part of the nationalcheckoff. For seasonal items such as fruits andvegetables, state programs may be a low cost pro-motion vehicle that is implemented during thestate’s marketing window.

Our results show that consumers generally har-bor a strong desire to purchase food products pro-


duced in-state, Identifying and promoting these Long, J.S. 1997. Regression Models for Categorical and Lim-

products is expected to have a positive effect on ited Dependent Variables Series 7 of Advanced Quantita-

sales. The results presented here can help states to tive Techniques in the Social Sciences. Sage Publications,

target their campaigns to provide the most efficientLondon.

use of resources. Patterson, P. M., H. Olofsson, T.J. Richards, and S, Sass. 1999.

“An Empirical Analysis of State Agricultural Product Pro-

References

Adelaja, A. O,, R.G. Brumfield, and K. Lininger. 1996, “Product

Differentiation and State Promotion of Farm Produce: An

Analysis of the Jersey Fresh Tomato” Journal of Food

LXswibudon Research 21(3):73–85,

Ben-Akiva, M., and S.R, Lerman. 1985. Discrete Choice Analy-

sis: Theory and Application to Travel Demand MIT Press,

Cambridge, MA,

Brown, A.R. 1988, “ ‘Jersey Fresh’: A Fresh Idea in Farm

Products Marketing” In 1988 Yearbook of U.S. Agricul-

tural Marketing, U.S. Department of Agriculture, Wash-

ington DC, 256-260.

Halloran, J.M. and M.V. Martin. 1989. “Should States be in the

Agricultural Promotion Business’?” Agribusiness 5(1 ):65-75.

Greene, W.H. 1996. Econometric Software Incorporated. LIM-

DEP Version 7.0.

Greene, W.H, 1991. Econometric Analysis Second Edition,

Macmillan Publishing, New York,

motions: A Case Study on Arizona Grown” Agribusiness

15(2):179-196.

Rae, R.C. 1991. “Pricing and Promotion in Asymmetric Duopo-

Iies” Marketing Science 10(2):13 1-144.

Ward, R. W., J. Chang, and S. Thompson. 1985. “Commodity

Advertising: Theoretical Issues Relating to Generic and

Brand Promotions.” Agribusiness 1(4):269–276.

Waugh, F.V. 1959. “Needed Research on the Effectiveness of

Farm Products Promotions.” Journal of Farm Economics

41:366376.

Williams II, D.R. 1995. “Feasibility of a Statewide Branding

Program for Indiana-Sourced Food and Agricultural Prod-

ucts.” Unpublished Masters Thesis. Department of Agri-

cultural Economics, Purdue University.

Wohlgenant, M.K. 1993. “Distribution of Gains from Research

and Promotion in Multi-Stage Production Systems: The

Case of the U.S. Beef and Pork Industries” American Journal

of Agricultural and Resource Economics 75(4):642-651,

Appendix

Telephone survey questions used in the analysis, with the mean and standard deviation of the responses.All questions were asked of the primary food shopper. 324 useable responses were collected,

Dependent Variable:

If Indiana produced food items were clearly identified and offered where you shop, how likely wouldyou be to purchase Indiana produced food products? Would you say that you are:

Extremely likely . . . . . . . . . . . . . . . . . . . . . . . . 5 Mean (as asked): 3.57Very likely . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Standard Deviation: 0.93Neutral or somewhat likely . . . . . . . . . . ...3Not very likely.,....................,., 2 Mean (collapsed to three levels): 2.55Notat all likely . . . . . . . . . . . . . . . . . . . . . . . . 1 Standard Deviation: 0.56

Independent Variables:

In comparison to other state’s products, would you say Indiana produced food products are:

Much higher quality, . . . . . . . . . . . . . . . . . . . 5Highly quality . . . . . . . . . . . . . . . . . . . . . . . . . . 4About the same quality . . . . . . . . . . . . . . . . . 3 Mean: 3.57

Lower aualitv . . . . . . . . . . . . . . . . . . . . . . . ...2 Standard Deviation: 0.54

Much l~wer quality . . . . . . . . . . . . . . . . . . . . 1

In comparison to other state’s products, would you say Indiana produced food products are:

Much higher inprice . . . . . . . . . . . . . . . . ...5Higher in price . . . . . . . . . . . . . . . . . . . . . . . . . 4Priced about the same . . . . . . . . . . . . . . . . ..3 Mean: 2.81

Lower in price . . . . . . . . . . . . . . . . . . . . . . . . . 2 Standard Deviation: 0.50

Much lower in price . . . . . . . . . . . . . . . . . . . 1


Since January of this year, how many times have you purchased food at a Farmers’ Market?

More than 10 . . . . . . . . . . . . . . . . . . . . . . . . . 116-10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Mean: 5.191-5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Standard Deviation: 4.170 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0

In general, when purchasing food products which of the following best describes your behavior?

I always purchase the same brand . . . ...5I usuaily purchase the same brand . . . . . 4

Mean: 3.78. . . sometimes . . . . . . . . . . . . . . . . . . . . . . . ...3

rarely . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...2Standard Deviation: 0.62

. . .I never ~urchase the same brand . . . . . . . 1

How important is product freshness in your decision to purchase a food product?

Extremely important . . . . . . . . . . . . . . . . . . . 5Very important . . . . . . . . . . . . . . . . . . . . . . . ..4Somewhat important . . . . . . . . . . . . . . . . ...3

Mean: 4.40

Not very important . . . . . . . . . . . . . . . . . . . . . 2Standard Deviation: 0.57

Notat all important . . . . . . . . . . . . . . . . . . . . 1

Please stop me when I read the category that best describes your yearly total household income.Is it...

Under 10,000 . . . . . . . . . . . . . . . . . . . . . . . . . . . 110,001–25,000 . . . . . . . . . . . . . . . . . . . . . . . . . 225,001-30,000 . . . . . . . . . . . . . . . . . . . . . . . . . 330,001-45,000 . . . . . . . . . . . . . . . . . . . . . . ...4 Mean: 3.7545,001-55,000 . . . ..) . . . . . . . . . . . . . . . . . . . 5 Standard Deviation: 1.9355,001-65,000 . . . . . . . . . . . . . . . . . . . . . . . . . 665,001-75,000 . . . . . . . . . . . . . . . . . . . . . . . . . 7More than 75,000 . . . . . . . . . . . . . . . . . . . ...8

How long have you lived in Indiana?

Less than l year . . . . . . . . . . . . . . . . . . . . . . . . 1l–5years ..1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36-10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Mean: 17,31

11-20 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Standard Deviation: 5.66

More than 20years . . . . . . . . . . . . . . . . ...20

Including yourself and any children, how many people are currently living in your household?

Mean: 3.23Standard Deviation: 2.91

What is the last grade of school that you completed?

Some high school or less . . . . . . . . . . . . . . . 1High school graduate/equivalent . . . . . ...2Some college/technical degree . . . . . . . ...3

Mean: 2.72

Bachelors demee . . . . . . . . . . . . . . . . . . . . . . . 4 Standard Deviation: 1,21

Graduate wo&ldegree . . . . . . . . . . . . . . . . . . 5

Jekanowski, Williams, and Schick

What is the gender of the primary household food shopper?

Consumers’ Willingness to Purchase 53

Male: 75Female 249

Which of the following best describes the area in which your home is located?

Rural . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..l Mean: 1.94Small town . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Standard Deviation: 0.81City .,,..., . . . . . . . . . . . . . . . . . . . . . . . . . . ...3

Documents

Consumers’ Willingness to Purchase Locally Produced ...ageconsearch.umn.edu/bitstream/31329/1/29010043.pdfConsumers’ Willingness to Purchase Locally Produced Agricultural Products: