Local Wine Expenditure Determinants in the Northern Appalachian States

Transcription

1 Volume 46 Issue 2 Local Wne Expendture Determnants n the Northern Appalachan States Tmothy Woods a, Xuetng Deng b, La Noguera c, and Shang-Ho Yang d a Professor, Department of Agrcultural Economcs, Unversty of Kentucky, 402 Charles E. Barnhart Bldg., Lexngton, Kentucky, , USA. Phone: Emal: tm.woods@uky.edu b Ph.D. Student, Department of Agrcultural Economcs, Unversty of Kentucky, 333 Charles E. Barnhart Bldg., Lexngton, Kentucky, , USA. Emal: xuetng.deng@uky.edu c Assstant Professor, Agrcultural Economcs, 314D Flley Hall, Unversty of Nebraska-Lncoln Lncoln, Nebraska, , USA. Phone: Emal: la.noguera@unl.edu d Assstant Professor, Graduate Insttute of Bo-Industry Management, Natonal Chung Hsng Unversty, No.250, Guoguang Rd., South Dst., Tachung Cty, 40227, Tawan. Phone: Emal: bruce.yang@nchu.edu.tw Abstract Tennessee, Kentucky, Oho, and Pennsylvana have seen sgnfcant ncreases n the number of wneres n the past decade. Most of the wne dstrbuton has focused on premse sales, although a few of the larger wneres have started to explore other ways for market expanson. Ths study examnes wne expendture patterns for 1,609 wne consumers n ths four state regon. Expendture functons estmated for total wne expendture confrm expected factors that would postvely mpact wne purchases, such as wne knowledge and food preparaton. But t s also observed that greater wne expendtures are assocated wth greater nclnatons to buy local, suggestng opportuntes for local wneres to establsh a favorable pont of dfferentaton n ths market. Local wne expendtures are postvely assocated wth wne knowledge and educaton. Our results also suggest those that consume more wne spend more on local wne and have a strong preference for local products n general, suggestng there may be opportuntes for addtonal local food cross merchandsng partcularly n places where wne s already beng promoted and purchased n general. Keywords: wne expendture, local wne, wne knowledge, market segmentaton Correspondng author July 2015 Volume 46 Issue 2 30

2 Introducton Wne consumpton s ncreasng globally, and s expected to grow by 6.2% between 2010 and 2015, despte a long-term declne n consumpton n western European wne-producng countres (New York Daly News 2012). The rse n consumpton s largely drven by the Unted States, as t ranks number one n wne consumpton, accountng for 13% of global consumpton (Wne Insttute 2014). Amercans consume more wne than ever before, wth 19 consecutve years of volume growth (Wne Insttute 2013). Overall table wne consumpton n the Unted States ncreased from 213 mllon cases n 2000 to 323 mllon cases n 2014 (Wne Insttute 2015a). Consumers n the Unted States are not only consumng more wne, but also more expensve wne. The Unted States, as the world s largest retal wne market, spent more than $40 bllon on wne n 2010, accordng to Impact Databank (2010). The Internatonal Wne and Sprt Research predcts that the Unted States wll ncrease ts wne consumpton by 10% between 2011 and 2015 (New York Daly News 2012). Whle wne demand s ncreasng, so s the wne supply n the Unted States, manly for domestc consumpton. The U.S. wne ndustry was almost destroyed by war, Prohbton, and economc depresson for about half of the 20 th century. In the 1970s, the Calfornan wne ndustry started the natonal wne revval and several other states followed. In 1975, 34 states had wneres compared to 47 states n 1997 and 50 states n 2007 wth about 60% of the wneres located outsde of Calforna (Cannng and Perez 2008). U.S. wne exports have ncrease dramatcally snce the 1990s, wth a record of mllon gallons at a value of $872 mllon n 2007, almost a fve-fold ncrease (Cannng and Perez 2008). Wne producton and wnery numbers n the Northern Appalachan states of Kentucky, Oho, Tennessee and Pennsylvana have also ncreased sgnfcantly durng the past decade, although wne producton n the regon s stll qute small compared to Calforna. The market share for these new wne-producng states s very small. The Wne Insttute (2015a) estmates that n 2014 Calforna accounted for 60% of all wnes sold on the U.S. market by volume; other recent data suggest mported wnes accounted for 31%; and wnes from the other 49 states accounted for 8% (Hodgen 2011). Accordng to a wne statstcal report n 2009 from the Alcohol and Tobacco Tax and Trade Bureau (2010) Calforna produced 634 mllon gallons of wne (89% of the U.S. wne producton), Kentucky produced almost 2 mllon gallons of wne (0.28%), Oho produced 1.1 mllon gallons of wne (0.16%), Pennsylvana produced 0.8 mllon gallons of wne (0.12%), and Tennessee produced almost 0.3 mllon gallons of wne (0.04%). Accordng to Wnes & Vnes (2013), there were 3,532 wneres n Calforna, 166 wneres n Pennsylvana, 142 wneres n Oho, 66 wneres n Kentucky (Thornberry 2012) and 40 wneres n Tennessee n Wneres local to ths regon tend to focus on on-premse sales or nearby markets, mantanng a much dfferent market focus compared to larger wneres n large wne-producng regons (Woods et al. 2013). There are several challenges to local wneres n the Northern Appalachan regon n addton to ther small marketshare. Frst, local wneres use lmted marketng channels, relyng heavly on toursm or on premse sales. These tend to be partcularly small wneres that do not have access to other dstrbuton channels (Sun et al. 2014). competton s July 2015 Volume 46 Issue 2 31

3 ferce. Foregn producers are ncreasngly targetng the U.S. market as drnkng habts have shfted. Wne consumpton s decreasng n western European wne-producng countres such as France and Italy (Wne Insttute 2015b). Whle wne consumpton n the Unted States s ncreasng, to be compettve even n the regonal wne market, t s crucal for local wneres to understand consumers and develop effectve marketng strateges. We explore the factors drvng total wne expendture and compare them wth the factors drvng local wne expendture for consumers n Kentucky, Oho, Tennessee and Pennsylvana. We also explore the determnants of the probablty of tryng a local wne. To better understand wne consumers needs and buyng habts, t s necessary to have a comprehensve understandng of wne consumers characterstcs (demographcs, lfestyle, wne consumpton, knowledge and preferences). Wth ths nformaton, small wneres can use specfc marketng nstruments, lke target marketng, to promote ther products. An mportant concept n target marketng s recognzng whether those who are targeted show a strong affnty or brand loyalty to that partcular brand and understandng the values conveyed by that brand. Buldng brand loyalty s essental for local wne promoton and success n a crowded suppler market. Is there an opportunty for dfferentaton? Part of that brandng strategy s the dentfcaton of the wne produced as local to the local market. Market researchers apply dfferent econometrc methods to dscern segments wthn wne consumer markets. These methods nclude segmentaton accordng to geographc crtera, psychology, demographcs, purchasng behavor (propensty to purchase), occason for purchase or consumpton, benefts sought by consumers, etc. There s a vast body of lterature on wne market segmentaton. Costangro et al. (2007) argued that dfferent prces mean dfferent products, and segmented the wne market nto dfferent prce categores. Ther results confrmed that mplct prces for attrbutes dffer across prces categores and at least two dfferent wne classes exst: consumpton wnes and collectble wnes. Therefore, these classes dentfy dfferentated products that fulfll dfferent needs and should be consdered separately. After examnnng over 180 hedonc wne prce models, Oczkowsk and Doucoulagos (2015) dentfed a moderate prce-qualty correlaton suggestng the exstence of mperfect nformaton regardng wne qualty. They dentfed wne reputaton as one of the most mportant structural varables for prce-qualty studes and recommend wne producers to drect resources to mprove reputaton. The Wne Market Councl (2003) proposed fve major behavoral wne segments of the U.S. populaton by consumpton rate: super-core (consume wne daly), core (consume wne at least two or three tmes per month), margnal (consume wne at least two or three tmes per quarter), non-adopters (do not drnk wne but drnk other alcoholc beverages), and non-drnkers (do not drnk any alcoholc beverages). Lancaster and Stllman (2009) segmented wne consumers nto four categores based on generaton: Tradtonalsts (born between 1900 and 1945), Baby Boomers (born between 1946 and 1964), Generaton Xers (born between 1965 and 1977), and Mllennals (born between 1977 and 2000). Thach and Olsen (2006) conducted a demographc wne market segmentaton targetng mllennal wne drnkers. Ther results ndcated that there s a need for greater wne advertsng to ths group utlzng fun, socal, and relaxed settngs; more nnovatve packagng and labels; a focus on value wnes; as well as taste enhancements and envronmental characterstcs. July 2015 Volume 46 Issue 2 32

4 Johnson, Rngham and Jurd (1991) used conjont choce analyss to do behavoral segmentaton for the Australan wne market. They dentfed sx dstnct choce segments: dry wne enthusast, whte wne trendes, Moselle preferrers, prce-senstve whte drnker, red wne buffs, and popular red brand preferrers. They found that key profle areas were demographcs, values, lfestyle meda habts, brand behavor, and brand perceptons. Bruwer and L (2007) confrmed the exstence of fve lfestyle-related segments n the South Australa wne market. These segments are: conservatve, knowledgeable wne drnkers (19.2%), enjoymentorented, socal wne drnkers (16.2%), basc wne drnkers (23.5%), mature, tme-rch wne drnkers (18.2%), and young professonal wne drnkers (22.9%). They also recognzed the evolvng nature of ths market. Wthn the U.S. wne market, geographc segmentaton demonstrated that most wne consumers lve near major ctes, such as San Francsco, Los Angeles, Mam, Seattle, and Chcago (ACNelsen 2003). Aras-Bolzmann et al. (2003) treated country of orgn, qualty, varety and age as predctors of wne prces, usng data from the Wne Spectator magazne. Ther results confrmed that the North Amercan wne market recognzes dfferences n country of orgn, qualty and varety. Zhao (2008) compared the classfcaton systems and structure n the Calforna and French wne ndustres. The author found that smlar categores and wne attrbutes affect wne prce dfferently under dfferent classfcaton systems. Al and Nauges (2007) showed that n the short-term prcng depends on reputaton to a larger extent than n qualty by usng data on Bordeaux wnes. Ths lterature documents the heterogenety of preferences across wne consumers and subsequent opportuntes for targeted marketng. However, there are fewer studes on preferences for local wnes, partcularly across types of wne consumers. In ths study, we explore the answers to three research questons: 1) what are the determnants of total wne expendture? 2) what are the determnants of local wne expendture? and 3) what are the determnants of the probablty of purchasng a local wne? We use the results from a four state survey of 1,609 wne consumers n Pennsylvana, Oho, Kentucky and Tennessee. We use a market segmentaton model followng a Hartman consumer survey on natural foods consumers (The Hartman Group 2000). We classfy wne consumers nto three categores accordng to ther wne purchases: perphery (at least once per year), md-level (at least once per month), and core (at least once per week) (Woods et al. 2013). We dfferentate between total wne expendture and local wne expendture to dentfy dfferences among consumers choosng local wne to help local wneres develop effectve marketng strateges. There are several contrbutons from ths study. It contrbutes to the understandng of Northern Appalachan wne consumer characterstcs and expendture patterns, and subsequently provdes a framework for future market strateges for the development of wneres n general and local wneres n specfc based on segmentaton observatons. It s a reasonable expectaton that there may be some dfferences between the factors contrbutng to local wne purchase (defned here as produced wthn the state of the wne consumer) as opposed to wne purchases n general. July 2015 Volume 46 Issue 2 33

5 Data The data used n ths paper were collected usng a web-based consumer survey dstrbuted and managed by Zoomerang, an afflate of Market Tools, Inc. Each partcpant was double prescreened, to ensure they were at least 21 years old, and they were wne consumers. A total of 1,609 complete observatons were collected n September, Partcpants were recruted from Tennessee (403 observatons), Kentucky (402 observatons), Oho (401 observatons), and Pennsylvana (403 observatons). Survey partcpants were asked about ther wne consumpton and purchase habts n the past 12 months, ncludng ther expendtures on all types of wne, expendture on local wne, wne consumpton frequency, purchasng habts regardng dfferently prced wnes, past local wne experence, wne knowledge level, local purchase frequency for all products, lfestyle, as well as demographc nformaton. 1 Varables used as dependent and ndependent varables are defned n Tables 1 and 2 (see Appendx for Table 2). Table 1. Defntons and Sample Statstcs of Dependent Varables (N = 1,609) Varables Descrpton of Varables Mean Std. Dev. Mn. Max. Total_expend Local_expend Local_tred Categorcal varable from 1 to 6 f respondents ndcate ther average monthly expendture on ALL wne wthn the past 12 months ether on: 1. Less than $20; 2. $20-$39; 3. $40-$59; 4. $60-$79; 5. $80-$99; and 6. $100 or more. Categorcal varable from 1 to 6 f respondents ndcate ther average monthly expendture on State wne wthn the past 12 months ether on: 1. Less than $20; 2. $20-$39; 3. $40-$59; 4. $60-$79; 5. $80-$99; and 6. $100 or more. Bnary varable=1 f respondents have tred what they know to be a state local wne wthn the past 12 months From our sample of 1,609 consumers, 627 respondents (38%) ndcated that they tred a state local wne (defned as from a wnery wthn ther state) and 34% purchased local wne n the past 12 months, whle 45% vsted a local wnery n the past three years. There are observable dfferences n the absolute expendture levels, purchase frequency, and the frequency of wne purchases by qualty/cost category. Monthly wne expendture was self-reported and ranged from $10 to $110, wth a mean of $39 for the total sample and $34.62 for the local sample, wth only a small fracton (12.92%) reportng zero expendtures on local wne n the last year (no zeroes for the total sample). In terms of wne purchase frequency, 57.60% of consumers purchased wne at least once per month, and 12.11% purchased wne at least once per week n the total sample, whle 67.45% of consumers purchased wne at least once per month, and 16.74% purchased wne at least once per week n the local sample. Consumers n both samples buy more wne n the super category ($7-$14 per bottle) on average, 71.4% of total sample and 78.30% of the local sample consumers bought a bottle of wne prced $7-$14 often. Around 50% of the consumers beleved ther wne knowledge was average to above average n the total sample, compared to 62.67% n the local sample. More nformaton comparng expendture and consumpton characterstcs of our total and local consumers s found n Table 3 (see Appendx). 1 For more nformaton about the survey and data see Woods, Noguera and Yang (2013). July 2015 Volume 46 Issue 2 34

6 Methodology The methodology used to estmate the determnants of total and local wne expendture n ths study follows the random utlty theory, whch accounts for an optmzaton of consumer utlty for every choce consumers make on wne expendture. Therefore, an ordered logt model can be * ' * specfed as: y = xβ + u. We defne: y = j f γ j 1 < y < γ j n a gven M-alternatve ordered model where γ 0 =, and γ M =. The error term, u, s assumed to be ndependent and dentcally dstrbuted, and the ordered logt model has a logstc cumulatve dstrbuton z z functon: F ( z) = e /(1 + e ). Snce wne consumers were asked to choose ther total expendture n sx categores,.e. less than $20, $20-$39, $40-$59, $60-$79, $80-$99, and $100 or more, the M alternatve equals 6 and the ordered logt model can be framed as: * ' (1) y = xβ + u, (2) y = 1 f y * 0, * (3) y = 2 f 0 < y γ 1, * (4) y = 3 f γ 1 < y γ 2, * (5) y = 4 f γ 2 < y γ 3, * (6) y = 5 f γ 3 < y γ 4, * (7) y = 6 f y > γ, 4 To explan the optmal decson on wne expendture for each consumer, y * represents wne consumers who would be better off when they spend a certan amount of dollars for wne wthn a range at each expendture level. The explanatory varables, x, consst of wne consumpton frequency, geographc factors, varous lfestyle factors, wne knowledge, past experence wth local wne, general support for local food, common wne prce ponts, and demographc factors. The model specfcaton for total or local wne expendture s: 2 (8) Y = β0 + β1male + β 2 Age + β3whte + β 4Wne _ drnkers + β5income + β 6Income + β 7Educaton + 2 b8educaton + b9kds + b10urban + b11pa + b12ky + b13tn + b14resdency2 + b15resdency3 + β16buy _ local2 + β17buy _ local3 + β18food _ channel + β19prep _ freshfood2 + β 20Prep _ freshfood3 + β 21Wne _ knowledge2 + β 22Wne _ knowledge3 + β 23Grade _ Popular + β24grade _ Super + β25grade _ Ultra + β26grade _ Luxury + β27local _ Range + β28type _ Whte + β Typε Rεd + β Typε _ Frut + β Typε _ Champagnε + ε 29 _ where Y represents the category of wne expendture for total or local wne, β s are the estmated coeffcents, and ε s the error term. Dependent varables are defned n Table 1 and ndependent varables n Table 2 (see Appendx). The maxmum lkelhood method s used to estmate the ordered logt model. The estmated coeffcents and odds ratos are provded for nterpretaton. The odds rato s calculated by takng the exponent of the estmated coeffcent. A postve odds rato represents the odds of a specfc July 2015 Volume 46 Issue 2 35

7 wne expendture ncrease wth a hgher value of the explanatory varable. However, when the estmated coeffcent s negatve, the odds ratos would be between 0 and 1, the odds of a specfc wne expendture decreases for the explanatory varable. One of the assumptons for the ordered logt model s the proportonal odds assumpton, whch means that the estmated coeffcents among pars of outcome groups are the same. Therefore, based on the Ch-Square Score examnaton n the logstc procedure provded by SAS, a rejected null hypothess for the proportonal odds assumpton suggests that the ordered logt model s not vald and one should use a less restrctve model. Each ordered logt model wth total or local wne expendture s examned and the test outcome provded n Tables 4 and 5 (see Appendx). We assume that consumers optmze ther utlty when they decde to purchase a local wne. Thus, we also use random utlty theory to explan the consumer decson on purchasng local wne. The probablty of purchasng a local wne can be explaned by wne consumpton frequency, geographc factors, varous lfestyle factors, wne knowledge, past experence wth local wne, general support for local food, common wne prce ponts, and common demographc factors. In the determnants of the probablty of purchasng local wne, ths study follows the same ndependent varables as n equaton 8. A logstc model s used n determnng the probablty of purchasng a local wne. The logstc model can be specfed as: ' x β (9) ( ' e p = L xβ ) = ' x β 1+ e where β s an estmated parameter, x s a vector of regressors, and L (.) s the standard logstc dstrbuton functon. To explan the estmated parameters we use the margnal effect snce t accounts for the probablty regardng the ndependent varables. The calculaton of margnal effects s: L xβ (10) p y x L x L xβ β ' ( ) ' ' Pr[ = 1 ] ( ) ( )[1 = = = n ] β. j x x x n j j j The examnaton of the logstc model also provdes McFadden s Adjusted R 2, Correctly Predct, and Goodness-of-ft for the model n Table 6 (see Appendx). Results The general expendture functon for all wne among consumers n ths regon suggests several varables are mportant determnants to explan wne expendture varaton. The ordered logt regresson results are summarzed n Table 4 (see Appendx). State dummy varables were sgnfcant, suggestng some heterogenety n expendture across the four states. The propensty to buy more local food and prepare food at home were postve, suggestng these food purchase behavors are complementary to wne purchase. Wne knowledge s also postvely related to wne purchases. These results were expected. The frequency of wne purchase by cost category also proved to be a good ndcator of overall expendture. These varables are essentally frequency of purchase measures that one would expect to be postvely correlated to overall wne expendture, the frequent purchase of the hgher July 2015 Volume 46 Issue 2 36

8 cost luxury wnes provdng the largest mpact on expendture. Frequent purchase of whte or red wnes were also sgnfcant coeffcents. It s mportant to keep n mnd that ths total wne expendture regresson looks at expendture patterns for consumers specfcally n Tennessee, Kentucky, Oho, and Pennsylvana. These results do provde, however, a way to characterze wne consumers n the regon that could have promotonal mplcatons for regonal wne marketng n general. The Brant test suggested a volaton of the parallel coeffcents assumpton that mght be employed n a regular logt model. 2 The full multnomal logt model s presented wth specfc margnal effects reported for each total wne expendture class. The odds rato allows us to nterpret the coeffcents n terms of relatve lkelhood of a hgher value for the ndependent varable. A postve coeffcent estmate, such as Buy_local2 at wth an odds rato of 1.787, means the odds of spendng more on wne s tmes more lkely for those consumers ndcatng that they often or always purchase what they know to be locally produced foods. The odds rato, then, allows us to not only determne the postve or negatve effects, but the magntude of the effect. Smlar regressors were appled to expendtures on local wne, summarzed n Table 5 (see Appendx). Male wne consumers and those that reported larger numbers of wne consumers n the home were more lkely to have hgher local wne expendtures. Income also had a postve effect, although at a decreasng rate. Urban wne consumers were less lkely to have hgher local wne expendtures. Frequent wne purchasng n general, not surprsngly, s a strong determnant of local wne expendture as s nclnaton to buy local products. Wne consumers often and always purchasng local food were 4.1 tmes more lkely to spend more on local wne than those that never purchase what they know to be locally produced foods. Overall wne knowledge strongly mpacted expendture levels for local wne, suggestng wne connosseurs are more lkely to have gven local wnes a tral and not dsmssng them wthout experencng them. Wne consumers that purchased relatvely hgher prced wnes (partcularly the luxury category above $25/bottle) were also more lkely to have a hgher expendture on local wne. Local wne prces tend to be hgher due to ther smaller scale of producton and not partcularly targeted to the value prce shopper. It s not surprsng to see postve relatons to local wne expendture wth frequent purchases of hgher prced wne categores here. Frut wne consumpton turned out to be a strong determnng factor for local wne consumpton, nterestngly not sgnfcant for overall wne expendture. Consumers ndcatng that they often or always purchased frut wne were 2.38 tmes more lkely to have a hgher expendture on local wne compared to those that dd not. Frut wnes are a popular product among many small and local wneres, supplementng ther grape-based wnes as a means for product dfferentaton. The expendture functons for all wne versus just local wne suggest several mportant dfferences. Several varables were sgnfcant explanng local wne expendtures that were not sgnfcant for all wne. Gender (male) and ncome show up as postve factors for local wne 2 The results of the Brant test are avalable upon request. July 2015 Volume 46 Issue 2 37

9 expendture and urban shows up as a negatve factor none of these emergng as sgnfcant factors for overall wne expendtures. Fresh food preparaton, sgnfcant for overall wne expendture, was not a sgnfcant determnant for local wnes. The heterogenety observed for overall wne expendture across states also dd not present tself n the local wne results. The specfc relaton between local wne expendture and general wne consumpton s captured n core, md-level, and perphery varables ncluded n the local model. Core and md-level categores of general wne purchase are postvely nfluencng the local wne expendture (relatve to the perphery category), suggestng more general wne consumpton postvely affects local wne consumpton. The fnal model explored the actual lkelhood of purchasng local wne usng smlar determnants, wth local tral beng a smple response to tred/not tred a local wne wthn the past 12 months (Table 6, see Appendx). Ths bnary logt model would be expected to follow a somewhat smlar pattern observed n the expendture regresson, but provdng a more general perspectve of product awareness and lkelhood to consume, recognzng reported tral of a local wne s a less accurate consumpton measure than local wne expendture. Male, ethncally whte, and more frequent wne consumers were more lkely to have tred a local wne. Income and educaton were also postvely related, ncreasng at a decreasng rate (wth the negatve squared term). Urban consumers were less lkely to have tred a local wne. Tennessee consumers were less lkely than Oho (the base) consumers. Propensty to buy local products and knowledge of wne n general, as n the expendture functon, were also a strong determnants. Frequent consumpton of the mddle prced wne categores were sgnfcant. The local tral regresson also ponted to both frequent purchase of whte wnes and frut wnes as sgnfcant determnants of tral. The pseudo R 2 for ths regresson was whle correctly predctng 68.74% of the responses. Conclusons Wne consumpton per capta n the Unted States moved to ther hghest levels n 2013 at 2.82 gallons and a total natonal wne consumpton twce what t was n 1979 (Wne Insttute 2014), as noted earler. A steadly growng market has created demand for mports, large domestc producers, and small regonal wneres. Ths competton plays out n local areas where local wneres have pursued ther nche n the market. Heterogenety n wne consumer preferences creates the potental for segmentng and targetng wne consumers wth partcular tastes. The results of overall wne expendture suggests consderable varaton n who purchases wne and to what extent wthn the Northern Appalachan states of Tennessee, Kentucky, Oho, and Pennsylvana. Expendture functons estmated for total wne expendture confrm expected factors that would postvely mpact wne purchases, such as wne knowledge and fresh food preparaton. A postve connecton for local wnes wth the establshed wne consumer communty should result n a contnued growth for local wnes as general wne consumpton contnues to expand. It s also observed that greater wne expendtures are assocated wth greater nclnatons to buy local among general wne consumers, suggestng contnued opportuntes for local wneres to establsh a favorable pont of dfferentaton n ths market. These opportuntes could be pursued through jonng local foods merchandsng efforts of exstng grocery retalers, restaurants that featured local foods, or local food festvals and events. These results provde July 2015 Volume 46 Issue 2 38

10 some possble market growth drectons to wneres that have tradtonally lmted themselves to on-premse sales. Expendtures on local wne are a subset of total wne expendtures, but appear to have somewhat dfferent determnants. They are observed to be drven by gender (more by male consumers), overall frequency of consumpton, propensty to buy local, and overall wne knowledge. There appears to be some prce senstvty for local wnes. Wne consumers ndcatng frequent purchases of luxury prce category wnes had the hghest local wne expendture, but only 15% of the surveyed populaton ndcated purchasng wne n these prce ranges sometmes or often. The relaton between local wne expendture and a number of the ndependent varables clearly dffer across the sx expendture categores, as noted by the margnal effects. These results suggest some heterogenety of preferences for local wne among Northern Appalachan consumers. Some factors, such as ncome, number of wne consumers per household, wne knowledge, and frequent purchase of more expensve wnes are not surprsngly postvely assocate wth hgher local wne expendture. As local wne expendtures are postvely assocated wth general wne knowledge and overall educaton, these results reterate the opportuntes to promote and dfferentate local wne n the establshed wne consumer communty. Our results also suggest that local wne may become better accepted as perphery wne consumers expand wne consumpton and become more knowledgeable about wne n general. We fnd that those who consume more wne spend more on local wne and also have a strong preference for local products n general, suggestng there may be opportuntes for addtonal local food cross merchandsng partcularly n places where wne s already beng promoted and purchased n general Local wne tral results pont more drectly to evdence of dfferent consumer segments that could serve as target markets. Local wne has tended to have a better recepton or at least tral among hgher ncome, hgher educated consumers that already have a good knowledge of wne. Wne knowledge and consumpton n the regon s ncreasng but local wnes are stll an underdeveloped market. General educaton about wne seems to postvely mpact local tral and expendture. There s evdence regonally of more heterogeneous wne preferences among those consumers choosng more local wne especally preferences for frut wne. Whte grapes and frut for wne have been typcally easer to grow n the regon and subsequently easer to manufacture nto better wnes. July 2015 Volume 46 Issue 2 39

11 References ACNelsen ACNelsen Report on the Largest U.S. Wne Markets. Wne Busness Monthly, October In&dataId= [Accessed March 10, 2014]. Alcohol and Tobacco Tax and Trade Bureau Alcohol and Tobacco Tax and Trade Bureau: Statstcal Report Wne (Reportng Perod: January December 2009). Al, H.H., and C. Nauges The Prcng of Experence Goods: The Example of En Prmeur Wne. Amercan Journal of Agrcultural Economcs 89 (1): Aras-Bolzmann, L., O. Sak, A. Musalem, L. Lodsh, K.R. Baez, and L.J. De Sousa Wne Prcng: The nfluence of Country of Orgn, Varety, and Wne Magazne Ratngs. Internatonal Journal of Wne Busness Research 15 (2): Bruwer, J., and E. L "Wne-related lfestyle (WRL) market segmentaton: demographc and behavoural factors." Journal of Wne Research 18(1): Cannng, P. and A. Perez The Economc Geography of the U.S. Wne Industry. Amercan Assocaton of Wne Economsts Workng Paper No. 22, September Costangro, M., J.J. McCluskey, and R.C. Mttelhammer "Segmentng the wne market based on prce: hedonc regresson when dfferent prces mean dfferent products." Journal of Agrcultural Economcs 58: The Hartman Group Organc Lfestyle Shopper: Mappng the Journeys of Organc Consumers, October Hodgen, D. A U.S. Wne Industry U.S. Department of Commerce. Impact Databank The U.S. Wne Market: Impact Databank Revew and Forecast, 2010 Edton. Wne Spectator. Johnson, L.W., L. Rngham, and K. Jurd "Behavoural Segmentaton n the Australan Wne Market Usng Conjont Choce Analyss." Internatonal Marketng Revew 8. Lancaster, L. C, and D. Stllman When Generatons Collde: Tradtonalsts, Baby Boomers, Generaton Xers, Mllennals : Who They are, Why They Clash, How to Solve the Generatonal Puzzle at Work. New York: HarperBusness. New York Daly News U.S. Is Bggest Wne Consumer, Chna Jons Top Fve; People Payng More Money For Better Vno. [Accessed Aprl 2014]. July 2015 Volume 46 Issue 2 40

12 Nchols, R U.S. Wne Consumer Trends: Boomers Tastes Evolve, Mllennals Contnue to Drve Market Growth. &data d = [Accessed Aprl 2014]. Oczkowsk, E., and H. Doucoulagos Wne Prces and Qualty Ratngs: A Metaregresson Analyss. Amercan Journal of Agrcultural Economcs 97(1): Sun,L., M. I. Gomez, F. R. Chaddad, and R. B. Ross Dstrbuton Channel Choces of Wneres n Emergng Cool Clmate Regons. Agrcultural and Resource Economcs Revew 43(1): Thach, E.C., and J.E. Olsen "Market segment analyss to target young adult wne drnkers." Agrbusness 22: Thornberry, T Kentucky Wneres Grow n Number and Statue. Busness Lexngton. [Accessed Aprl 25, 2014]. Wne Insttute Wne Sales n U.S. Reach New Record: Record Calforna Wnegrape Crop to Meet Surgng Demand. /pressroom/ [Accessed Aprl 2014]. Wne Insttute Statstcs. [Accessed Aprl 2014]. Wne Insttute. 2015a. Wne Sales n the U.S. / [Accessed July 2015]. Wne Insttute. 2015b. World Wne Consumpton /2012_World_ Consumpton_by_Country_Calforna_Wne_Insttute.pdf [Accessed July 2015]. Wne Market Councl Wne Consumer Trackng Study Summary Wnes & Vnes North Amercan Wnery Total Passes 8, vnes.com/ template.cfm?secton=news&content= [Accessed Aprl, 2014]. Woods, T.A., L. Noguera and S.-H. Yang Lnkng Wne Consumers to the Consumpton of Local Wnes and Wnery Vsts n the Northern Appalachan States. Internatonal Food and Agrbusness Management Revew 16(4): Zhao, W Socal Categores, Classfcaton Systems, and Determnants of Wne prce n the Calforna and French Wne Industres. Socologcal Perspectves 51 (1): July 2015 Volume 46 Issue 2 41

13 Appendx Table 2. Defntons and Sample Statstcs of Independent Varables (N = 1,609) Varables Descrpton of Varables Mean Std. Dev. Mn. Max. Male Bnary varable=1 f respondent s male Age Contnuous varable; year of age Wne_drnkers Count varable for the number of wne drnkers at respondent s household Whte Bnary varable=1 f respondent s race s whte Income Contnuous varable; total yearly household ncome before tax ($1,000) Educaton Contnuous varable; year of educaton Kds Bnary varable=1 f respondent has kds under 18 at home Urban Bnary varable=1 f respondent s from urban (ncludng cty and suburb) OH Bnary varable=1 f respondent s from Oho. PA Bnary varable=1 f respondent s from Pennsylvana KY Bnary varable=1 f respondent s from Kentucky TN Bnary varable=1 f respondent s from Tennessee Resdency1 Bnary varable=1 f respondent has lved n the state for 1-4 years Resdency2 Bnary varable=1 f respondent has lved n the state for 5-9 years Resdency3 Bnary varable=1 f respondent has lved n the state for 10 or more years Core Bnary varable=1 f respondent has purchased wne for any occason wthn the past months at least once per week. Md_level Bnary varable=1 f respondent has purchased wne for any occason wthn the past months at least once per month. Perphery Bnary varable=1 f respondent has purchased wne for any occason wthn the past months at least once per year. Buy_local1 Bnary varable=1 f respondent never purchases what they know to be locally produced foods. Buy_local2 Bnary varable=1 f respondent sometmes purchases what they know to be locally produced foods July Volume 46 Issue 2 42

14 Table 2. Contnued Varables Descrpton of Varables Mean Std. Dev. Mn. Max. Buy_local3 Bnary varable=1 f respondent often and always purchases what they know to be locally produced foods Food_channel Bnary varable=1 f respondent watches the food channel or smlar programs Prep_freshfood1 Bnary varable=1 f respondent never prepares fresh food at home Prep_freshfood2 Bnary varable=1 f respondent prepares fresh food at home for 1-6 tmes per month Prep_freshfood3 Bnary varable=1 f respondent prepares fresh food at home for 7 tmes above per month Wne_knowledge1 Bnary varable=1 f respondent rates ther wne knowledge as a lttle and novce level Wne_knowledge2 Bnary varable=1 f respondent rates ther wne knowledge as an average level Wne_knowledge3 Bnary varable=1 f respondent rates ther wne knowledge as an above average and expert level. Grade_popular Bnary varable=1 f respondent purchases popular wne ($4-$7/bottle) at the frequency of sometmes and often. Grade_super Bnary varable=1 f respondent purchases super wne ($7-$14/bottle) at the frequency of sometmes and often. Grade_ultra Bnary varable=1 f respondent purchases ultra wne ($14-$25/bottle) at the frequency of sometmes and often. Grade_luxury Bnary varable=1 f respondent purchases luxury wne (above $25/bottle) at the frequency of sometmes and often. Local_range Contnuous varable; respondent defnes local wnery n terms of mle range from ther home. Type_whte Bnary varable=1 f respondent ndcates the whte wne purchasng frequency as often and usually/always. Type_red Bnary varable=1 f respondent ndcates the red wne purchasng frequency as often and usually/always. Type_frut Bnary varable=1 f respondent ndcates the frut wne purchasng frequency as often and usually/always. Type_champagne Bnary varable=1 f respondent ndcates the champagne/sparklng purchasng frequency as often and usually/always July Volume 46 Issue 2 43

15 Table 3. Consumpton Characterstcs Comparson n Total versus Local Samples. Consumpton Characterstcs Total Wne Expendture (1,609) Local Wne Expendture (627) Expendture <$ % 52.31% $20-$ % 25.52% $40-$ % 9.57% $60-$ % 4.31% $80-$ % 1.28% $100 or more 9.26% 7.02% Core 12.11% 16.74% Md_level 45.49% 50.71% Perphery 42.38% 32.53% Buy_local1 6.09% 3.34% Buy_local % 43.54% Buy_local % 53.12% Food_channel 75.69% 78.78% Prep_freshfood1 2.73% 1.75% Prep_freshfood % 20.26% Prep_freshfood % 77.99% Wne_knowledge % 37.33% Wne_knowledge % 43.54% Wne_knowledge % 19.13% Grade_popular 50.21% 50.87% Grade_super 71.40% 78.30% Grade_ultra 43.74% 52.95% Grade_luxury 15.96% 21.37% Local_range Type_whte 43.25% 51.03% Type_red 52.75% 59.16% Type_frut 33.37% 40.50% Type_champagne 14.84% 17.53% Note. Expendture level <$20 ncludes no zeroes for the total sample and 12.92% zeroes for the local sample. All consumers n the local sample have tred a local wne. Local_range s defned n mles. July 2015 Volume 46 Issue 2 44

16 Table 4. Total Wne Expendture Dependent Varable Total Wne Expendture Coeffcent O.R a Less than Margnal Effects $20-$39 $40-$59 $60-$79 $80-$99 $100 or more $20 Male e (0.106) (0.023) (0.001) (0.008) (0.006) (0.003) (0.005) Age e e (0.003) (0.000) (0.000) (0.0003) (0.0002) (0.0001) (0.0002) Wne_drnkers (0.069) (0.015) (0.000) (0.005) (0.004) (0.001) (0.003) Whte e (0.164) (0.037) (0.000) (0.013) (0.010) (0.004) (0.008) Income e e (0.003) (0.000) (0.000) (0.0003) (0.000) (0.0001) (0.0001) Income e e e e e e-06 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Educaton (0.378) (0.086) (0.002) (0.032) (0.023) (0.010) (0.019) Educaton (0.012) (0.002) (0.000) (0.001) (0.000) (0.000) (0.0006) Kds (0.119) (0.027) (0.000) (0.010) (0.007) (0.003) (0.006) Urban (0.101) (0.023) (0.000) (0.008) (0.006) (0.002) (0.005) PA (0.138) (0.031) (0.000) (0.011) (0.008) (0.003) (0.007) KY 0.549*** 1.732*** *** *** 0.035*** 0.016*** 0.032*** (0.141) (0.029) (0.005) (0.010) (0.009) (0.005) (0.009) TN 0.539*** 1.715*** *** *** 0.034*** 0.016*** 0.032*** (0.140) (0.029) (0.005) (0.010) (0.009) (0.005) (0.009) Resdency (0.257) (0.062) (0.009) (0.022) (0.013) (0.005) (0.010) Resdency (0.214) (0.048) (0.000) (0.017) (0.013) (0.006) (0.011) Buy_local ** 1.787** ** ** 0.035** 0.015** 0.030** (0.234) (0.053) (0.004) (0.019) (0.014) (0.006) (0.012) Buy_local *** 2.114*** *** *** 0.046*** 0.021*** 0.042*** (0.240) (0.052) (0.005) (0.018) (0.015) (0.007) (0.014) Food_channel (0.117) (0.027) (0.001) (0.010) (0.006) (0.003) (0.005) Prep_freshfood ** 2.269** ** ** 0.053** 0.025* 0.052* (0.393) (0.076) (0.018) (0.024) (0.026) (0.014) (0.030) Prep_freshfood * 2.047* * * 0.040** 0.017** 0.033** (0.386) (0.093) (0.017) (0.032) (0.020) (0.008) (0.016) Wne_knowledge *** 2.102*** *** *** 0.046*** 0.021*** 0.043*** (0.107) (0.022) (0.005) (0.008) (0.007) (0.004) (0.007) Wne_knowledge *** 2.987*** *** *** 0.066*** 0.073*** 0.039*** 0.084*** (0.165) (0.026) (0.015) (0.007) (0.012) (0.008) (0.018) July 2015 Volume 46 Issue 2 45

17 Table 4. Contnued Dependent Varable Total Wne Expendture Coeffcent O.R a Less than Margnal Effects $20-$39 $40-$59 $60-$79 $80-$99 $100 or more $20 Grade_popular (0.101) (0.023) (0.000) (0.008) (0.006) (0.002) (0.005) Grade_super 0.274** 1.315** ** ** 0.016** 0.007** 0.013** (0.112) (0.026) (0.002) (0.009) (0.006) (0.002) (0.005) Grade_ultra 0.532*** 1.702*** *** *** 0.032*** 0.015*** 0.029*** (0.113) (0.025) (0.003) (0.009) (0.007) (0.003) (0.006) Grade_luxury 0.882*** 2.415*** *** *** 0.061*** 0.058*** 0.029*** 0.061*** (0.141) (0.025) (0.010) (0.008) (0.010) (0.006) (0.013) Local_range e e e e-05 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Type_whte 0.427*** 1.532*** *** *** 0.026*** 0.012*** 0.023*** (0.099) (0.022) (0.002) (0.008) (0.006) (0.003) (0.005) Type_red 0.593*** 1.809*** *** *** 0.035*** 0.016*** 0.031*** (0.101) (0.023) (0.003) (0.008) (0.006) (0.003) (0.005) Type_frut e (0.153) (0.034) (0.001) (0.012) (0.009) (0.004) (0.008) Type_champagne e (0.164) (0.038) (0.001) (0.014) (0.009) (0.004) (0.008) Intercept (2.840) Intercept (2.840) Intercept (2.841) Intercept (2.841) Intercept (2.842) N. of observatons 1609 LR χ *** Proportonal odds test * b Note. Astersks ndcate levels of sgnfcance: * = 0.10, ** = 0.05, and *** = Wald Test was also performed n SAS for nference of each coeffcent, k * = k b k. a O.R. represents odds rato. b The result of the proportonal odds test suggests to use a less restrctve model, lke the multnomal logt model. However, we only present the outcomes of the ordered logt model for ease of nterpretaton, snce the outcomes of the multnomal logt model are very smlar. The outcomes of the multnomal logt model are avalable upon request. β : z b / s{ } July 2015 Volume 46 Issue 2 46

18 Table 5. Local Wne Expendture Dependent Varable Local Wne Expendture Coeffcent O.R a Less than Margnal Effects $20-$39 $40-$59 $60-$79 $80-$99 $100 or more $20 Male 0.318** 1.375** ** 0.022** 0.017* 0.003* * (0.153) (0.019) (0.011) (0.003) (0.001) (0.0005) (0.002) Age e e (0.005) (0.000) (0.000) (0.000) (0.000) (0.0005) (0.000) Wne_drnkers 0.178* 1.194* * 0.011* 0.004* 0.001* * (0.096) (0.011) (0.006) (0.002) (0.0009) (0.000) (0.001) Whte (0.248) (0.025) (0.015) (0.005) (0.002) (0.0003) (0.003) Income 0.012** 1.012** ** ** ** ** 3.3e-05* ** (0.005) (0.000) (0.000) (0.000) (0.000) (0.0006) (0.000) Income 2-6.5e-05** 0.999** 7.5e-06** -4.3e-06** -1.4e-06** -6.1e-07** -1.7e-07* -8.9e-07** ( ) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Educaton (0.573) (0.066) (0.038) (0.013) (0.005) (0.000) (0.007) Educaton e (0.019) (0.002) (0.001) (0.000) (0.0001) (0.001) (0.0002) Kds (0.168) (0.018) (0.010) (0.003) (0.001) (0.000) (0.002) Urban * 0.756* 0.033* * * * * (0.149) (0.018) (0.010) (0.003) (0.001) (0.0004) (0.002) PA (0.210) (0.025) (0.014) (0.005) (0.002) (0.0005) (0.003) KY (0.210) (0.024) (0.014) (0.004) (0.002) (0.0006) (0.003) TN e (0.210) (0.024) (0.013) (0.004) (0.001) (0.0005) (0.002) Resdency (0.388) (0.040) (0.023) (0.007) (0.003) (0.0009) (0.004) Resdency (0.318) (0.031) (0.018) (0.006) (0.002) (0.0007) (0.003) Buy_local *** 3.188*** *** 0.095*** 0.038*** 0.016*** 0.004** 0.025*** (0.248) (0.048) (0.023) (0.011) (0.005) (0.002) (0.008) Buy_local *** 2.053*** *** 0.048*** 0.016*** 0.007*** 0.002** 0.010*** (0.186) (0.022) (0.013) (0.004) (0.002) (0.0009) (0.003) Food_channel (0.539) (0.061) (0.035) (0.012) (0.005) (0.001) (0.007) Prep_freshfood *** 4.124*** ** 0.100** 0.036** 0.015** 0.004* 0.023** (0.540) (0.075) (0.039) (0.016) (0.007) (0.002) (0.011) Prep_freshfood (0.181) (0.022) (0.012) (0.004) (0.001) (0.0005) (0.002) Wne_knowledge (1.043) (0.170) (0.087) (0.036) (0.016) (0.005) (0.025) Wne_knowledge (1.036) (0.083) (0.049) (0.016) (0.006) (0.002) (0.009) July 2015 Volume 46 Issue 2 47

19 Table 5. Contnued Dependent Varable Total Wne Expendture Coeffcent O.R a Less than Margnal Effects $20-$39 $40-$59 $60-$79 $80-$99 $100 or more $20 Grade_popular (0.010) (0.002) (0.007) (0.149) (0.017) (0.009) (0.001) Grade_super (0.003) (0.0004) (0.002) (0.176) (0.019) (0.011) (0.001) Grade_ultra (0.003) (0.0004) (0.002) (0.170) (0.020) (0.011) (0.001) Grade_luxury 0.499*** 1.647*** ** 0.036** (0.004) 0.005** (0.0005) (0.002) (0.189) (0.027) (0.015) 0.013** (0.002) 0.001* 0.008** Local_range 3.1e e e-06 (0.005) 2.9e-07 (0.0009) (0.003) (0.001) (0.000) (0.000) 7.1e-07 (0.000) 8.5e e-07 Type_whte 0.264* 1.303* * 0.017* (0.000) 0.002* (0.000) (0.000) (0.145) (0.017) (0.009) 0.006* (0.001) * Type_red (0.003) (0.0004) (0.002) (0.153) (0.017) (0.010) (0.001) Type_frut 0.867*** 2.380*** *** 0.062*** (0.003) 0.009*** (0.0004) (0.002) (0.217) (0.030) (0.017) 0.022*** (0.003) 0.002** 0.014*** Type_champagne (0.006) (0.001) (0.004) (0.226) (0.024) (0.014) (0.001) Intercept (4.386) Intercept (4.388) Intercept (4.389) Intercept (4.390) Intercept (4.391) N. of observatons 1609 LR χ *** Proportonal odds test Note: Astersks ndcate levels of sgnfcance: * = 0.10, ** = 0.05, and *** = Wald Test was also performed n SAS for nference of each coeffcent. a O.R. represents odds rato. July 2015 Volume 46 Issue 2 48

20 Table 6. Probablty of Local Tral Dependent Varable Local_tred Coeffcent Margnal Effect Male 0.230* 0.047* (0.126) (0.025) Age (0.004) (0.0009) Wne_drnkers (0.079) (0.016) Whte 0.462** 0.090** (0.200) (0.037) Income 0.007* 0.001* (0.004) (0.0008) Income 2-4.7e-05** -9.6e-06** (0.000) (4.3e-06) Educaton 1.061** 0.215** (0.484) (0.097) Educaton ** ** (0.016) (0.003) Kds (0.140) (0.027) Urban *** *** (0.121) (0.024) PA (0.157) (0.031) KY (0.164) (0.032) TN *** *** (0.171) (0.031) Resdency (0.320) (0.063) Resdency (0.262) (0.051) Core 0.436** 0.040** (0.203) (0.043) Md_level (0.133) (0.027) Buy_local (0.287) (0.054) Buy_local *** 0.182*** (0.294) (0.060) Food_channel (0.140) (0.028) Prep_freshfood (0.401) (0.080) Prep_freshfood (0.393) (0.079) Wne_knowledge ** 0.112*** (0.130) (0.025) July 2015 Volume 46 Issue 2 49

21 Table 6. Contnued Dependent Varable Local_tred Coeffcent Margnal Effect Wne_knowledge *** 0.209*** (0.198) (0.042) Grade_popular (0.121) (0.024) Grade_super 0.339** 0.068** (0.137) (0.027) Grade_ultra (0.136) (0.028) Grade_luxury (0.177) (0.037) Local_range -9.4e e-05 (0.0008) (0.0001) Type_whte 0.398*** 0.082*** (0.117) (0.024) Type_red (0.121) (0.024) Type_frut 0.720*** 0.149*** (0.177) (0.036) Type_champagne (0.186) (0.036) Constant *** (3.655) Log Lkelhood Wald χ *** Pseudo R McFadden s Adjusted R N. of observatons 1,609 Correctly predct 68.74% Goodness-of-ft (χ 2 ) 1, Note. Astersks ndcate levels of sgnfcance: * = 0.10, ** = 0.05, and *** = July 2015 Volume 46 Issue 2 50