Agricultura 9: No 1-2 (Special issue): 49-59 (2012) Copyright 2012 by University of Maribor Premia for differentiated products at the retail level: can the market put a value on the mountain attribute? Cesar REVOREDO-GIHA1* and Philip LEAT1 1Food Marketing Research Team, Land Economy and Environment Research Group, Scottish Agricultural College (SAC), King's Buildings, West Mains Road, Edinburgh, EH9 3JG, UK "So much is missed in the word mountain food - there is culture but it is not a mountain culture, it is a Highland culture" "When you mentioned mountain food, I thought of goats and Heidi and Switzerland" "I wouldn't want to buy Venison from anywhere, like the South of England" Some comments about mountain food products from focus groups held in Edinburgh, Aberdeen and Fort William, August 2008 (Scotland, UK). ABSTRACT The purpose of this paper is, by comparing products with a mountain provenance with those from non-mountain areas, to explore whether the market puts a premium on the 'mountain attribute. First, we present a theoretical framework on attributes and cues that helps answering the question what is "mountain" representing in a products or in other term, is it an attribute or a cue. Second, based on a shelves survey collected as part of the EuroMARC, we analyse for several products (apples, sausages, water and cheese) and countries (Austria, France, Norway, Scotland and Slovenia) using a hedonic price regression approach whether a premium is paid for mountain food products in comparison with identified similar nonmountain food products. The results indicate that the answer is mixed and depends on the product and country. Thus, premia was found only in the case of cheese and for Austria, Norway and Slovenia. Key words: mountain quality food products, attributes and cues, hedonic regression INTRODUCTION The concept of mountain food product is a complex one because it evocates different images to consumers. This can be observed in the diversity of opinions reflected in the three comments, cited at the beginning of the paper, from focus groups held in Scotland in August 2008. The purpose of this paper is to explore whether the market puts a value to the mountain attribute at the retailer level or in other terms whether consumers are willing to pay an additional amount (i.e. a premium) for buying a mountain quality food product. This is studied using prices from products representative from several European ranges -Highlands, Alps, Scandinavia, Massif Central. The motivation for studying the current situation of mountain food products (or prices actually paid) instead of hypothetical ones expressing consumers intentions is due to the fact that there is always discrepancies between hypothetical and actual behaviour (MAFF 2000). Thus, whilst consumers may show high interest on mountain quality food products when responding a hypothetical survey this is not always reflected in their buying behaviour or in their willingness to pay the higher price that products of a higher quality may carry and therefore, in practice, one may not observe a premium for mountain quality food products. As mentioned, the concept of mountain food is a complex one and this has been transmitted to consumers in several ways. Thus, the mountain origin of the products has been displayed to consumers in several ways and including a number of pieces of information, such as through the word 'mountain' itself, the mention of a geographic name of a famous mountain range or region, but mainly via images of *Correspondence to: E-mail: cesar.revoredo@sac.ac.uk mountains without compliance with procedures of origin. In some cases, nutritional information or positive claims such as 'farm products', 'traditional products, 'natural', 'extra, 'typical, 'without preservatives' are mentioned. Within a more general framework, the interest on the marketing of mountain products is associated to find "market driven" ways for adding value to mountain food products as a prerequisite for the survival and the management of rural and cultural mountain diversity. This is motivated by the new orientation of the Common Agriculture Policy which looks to promote "market driven" type of production where European Union farmers will be expected to respond to market signals (Sylvander 1993, Ilbery 1998, Leat et al. 2000). Mountain areas, which represent at least half of the area of six European States, with the greatest proportions in Austria (73 per cent), Greece and Slovenia (78 per cent), and Slovakia (62 per cent) and more than 90 per cent of both Norway and Switzerland- represent an important challenge for Europe to achieve sustainable development, including quality of life and the continued production of high-quality food, deriving mainly from environmental and cultural factors (Nordregio 2004). The structure of the paper is as follows. First, we present a theoretical framework on attributes and cues that helps answering the question what is "mountain" representing in a products or in other term, is it an attribute or a cue. Second, based on information provided by a shelves survey collected as part of the EuroMARC project, we analyse for several products (apples, sausages, water and cheese) and countries (Austria, France, Norway, Scotland and Slovenia) using a hedonic price regression approach whether a premium is paid for mountain food products in comparison with identified similar non-mountain food products. Theoretical framework - attributes and cues In the 1960s, Kelvin Lancaster pioneered a new approach to consumer theory in which he broke away from the traditional idea that goods are the direct objects of utility, and that instead it is the properties or characteristics of the goods from which utility is derived (Lancaster 1966). Subsequent literature relating to the quality attributes of goods and services (e.g., Nelson 1970, Darby and Karni 1973, Andersen 1994) makes a distinction between 3 types of attributes (see also OECD 1997): • Search attributes - which can be ascertained prior to a product's purchase (e.g., the colour of a cheese, or the thickness of fat cover on a piece of meat). • Experience attributes - which cannot be determined prior to purchase but which can be ascertained during consumption (e.g., the creaminess and taste of a cheese, or the taste and tenderness of meat). • Credence attributes - which cannot be determined prior to purchase or during consumption (e.g., the level of welfare experienced by a lamb during its life, or in some cases whether a product's ingredients were actually produced in a mountain area). Caswell et al. (1998) consider the grouping of attributes into 'process' and 'product' attributes. Northen (2000), in developing the work of Caswell et al. (1998), distinguishes five types of product attribute, covering: food safety; nutrition; and sensory, functional and image attributes. Process attributes relate to features of the production process. Whilst consumers may purchase products in order to consume physical product attributes, they may also be concerned about process attributes - such as artisanal production methods or organic production - and therefore purchase a particular product in order to purchase these as well. Beyond the farm gate, features of the processing and marketing channel, such as length of meat maturation, may also constitute a process attribute. In some cases process attributes may influence the physical product, but in many instances this causal relationship - where it exists - may be weak. For example, it may be claimed that the extensive production environment of a beef animal in a mountain area may affect the final meat product, but it may be questionable as to whether this can be detected by consumers. In the case of organic production, the influence of this process attribute may well be detectable for some products and some consumers. Similarly, traditional production methods in a rural mountain setting may give rise to discernible taste, smell or appearance features. These two classifications of attributes into 'search, 'experience' and 'credence, as well as 'process' and 'product' attributes can be combined as shown in Table 1, where the focus is on an organic meat product from a mountain area. It should be recognised that some attributes may be of more than one type, e.g., the juiciness of a piece of meat might be apparent prior to purchase (a search attribute) but also confirmed during consumption (an experience attribute). Furthermore, there is clearly a linkage between some attributes, e.g., the fat content of a piece of meat or of a cheese may well influence its taste. The communication of quality attributes: the deployment of quality cues The question arises as to how quality attributes are communicated to consumers prior to purchase. Consumers' perceptions of quality prior to purchase are based on quality cues; stimuli which lead to the perception of certain quality attributes being present and which determine when, where and how a person responds (Kotler 1980). Quality cues may be categorised into intrinsic and extrinsic cues (Olson and Jacoby 1972, Olson 1977, Bello Acebron and Calvo Dopico 2000). Thus: • Intrinsic quality cues cannot be changed or manipulated without changing the physical characteristics of the product itself. • Extrinsic quality cues are related to the product but are not physically part of it. As noted by Oude Ophuis and Van Trijp (1995), extrinsic cues can be manipulated by marketing activity, without the need to change the product itself. Consequently, extrinsic cues need to be carefully developed and deployed if a product is to be sold to best effect. In the case of meat, the intrinsic quality cues will include Table 1: Categorisation of potential 'process' and 'product' quality attributes of organic meat production from a mountain area Process Attributes Food Safety Nutrition Product attributes Sensory Functional Image Animal welfare (C) Absence of Residues (C) Fat content (S, E, C) Appearance (S) Product life (S and E) Prestige Value (S, E, C) Biotechnology (C) Absence of artificial Hormones (C) Energy content (C) Taste (E) Preparation Convenience (S and E) Organic production (C) Absence of Additives (C) Vitamins and minerals (C) Texture (E) Consumption Convenience (E) Prestige value (S,E,C) Traceability (C) Absence of Toxins (C) Tenderness (E) Feed and Feeding system (C) Absence of Physical contaminants (E and C) Juiciness (S and E) Mountain Production Environment (C) Treatment(s) in processing (C) Freshness and Taste (S and E) Smell (S and E) Prestige value (S,E,C) Note: S = Search attribute, E = Experience attribute, C = Credence attribute. The classification of the attributes into search, experience and credence is that of the authors. (Source: Developed from Northen (2000)) physical definitive features of the product such as lamb of a particular origin, as well as visual cues such as colour, leanness or fat cover, degree of marbling, juiciness and the type of cut. Smell may also be an intrinsic cue. For cheese, the intrinsic quality cues may again include its provenance, along with the colour, smell, texture, etc. Many of these cues may not be perceived by consumers either because they are ignored or because information is not provided (Bello Acebron and Calvo Dopico 2000). Extrinsic quality cues may include the price of a product, its brand name, packaging, labelling and label information, point of sale information, other promotional activities, presentation in the sales outlet, the place of purchase (reputation/status of the outlet), and the influence of the salesperson (Steenkamp 1989). The communication of attributes via cues is represented in Figure 1. It indicates that product attributes are capable of being communicated by intrinsic cues. The attributes concerned will be of the 'search' type. It is important to note that, as Table 1 has indicated, a significant number of product attributes are of the 'experience' and 'credence' types. Andersen (1994) has argued that credence attributes cannot be communicated by intrinsic cues, and it may be that some particular experience attributes, such as tenderness and texture are not readily predicted from intrinsic cues. Thus extrinsic cues, along with intrinsic cues, are important in communicating product quality attributes. Intrinsic Cues E.g. for meat - colour, fat cover, juiciness, smell Extrinsic Cues E.g. price, packaging, name / brand, label, place of sale, sales person, promotional material Product attributes E.g sensory attributes (e.g. taste, texture), image attributes (e.g. provenance of product) Process attributes E.g.production environment (e.g. mountain area), production process (e.g. traditional methods) Main communication relationship Less likely (weaker) communication Figure 1 : The relationship between cues and attribute s (source: according to Northen, 2000) Process attributes are very largely credence in nature, so that the effective communication of process attributes -including the production environment, animal welfare and traditional production systems - is largely dependent on extrinsic cues. Mountain as an attribute and as cue Within this framework of concepts, the mountain attribute may embody both product and process attributes, which can be regarded as a: • Search attributes (where the provenance is clearly indicated by a verified source) • Experience attributes (where the product's attributes give rise to a different experience to that of the non-mountain product, such as a different taste) • Credence attributes (where the purchaser and consumer have to believe that the mountain provenance is real and that this conveys additional utility). The cues which convey the mountain attribute may in some instances be intrinsic, such as the smell and colour of mountain heather honey, but in many instances the mountain attribute and its various aspects may need to communicated by extrinsic cues in the form of labelling, packaging, a relatively high price, information from the sales person, etc. It should be noted that when the term 'mountain' is used in a label, the way that it is normally communicated to consumers, the label 'mountain' becomes a cue of a number of attributes associated with the specific mountain product, which can be product and process attributes. In this paper we examine whether price, through the existence of a price premium, is being effectively used and accepted as a cue for the mountain attribute. MATERIAL AND METHODS Input data The data used in this paper come from shelves surveys conducted in Austria, France, Norway, Romania, Scotland and Slovenia. The data from Romania was not used because it did not contain information about the prices of alternative non-mountain food products. The main purpose of the shelves surveys was to study how Mountain Quality Food Products are currently marketed, covering issues such as whether the products are marketed as mountain products, whether labels are used in the shop or whether the products are presented together, and information amount prices of mountain food products and of similar non-mountain food products, etc. As regards the way the shelves surveys were planned and conducted, it is important to note that they were not constructed following any sampling procedure, i.e., based on any known population. Strictly speaking, the sampling population was all the retailers that market mountain quality food products, however, the characteristics of this population are unknown. In this respect, the type of sampling used was random sampling with replacement, since each country was committed to collect 90 shelves. Table 2 presents a summary of all the information collected by the shelves surveys. In total information corresponding to 564 shelves was collected, which resulted in 1,765 products (i.e., a product in the analysis consists of each element comprising a shelf; therefore, if the same product is sold in two different shops, it counts as a two products). In addition, this information was collected from a total of 351 different outlets (i.e., shops). As regards of shelves, 59.6 per cent of them where collected in mountain areas and 40.4 per cent of the outlets were also from mountain areas. As regards the distribution by country, the two extremes were Norway, with a higher proportion of non-mountain shelves (43.8 per cent mountain /56.2 per cent non-mountain) and on the other extreme was Austria, where a substantial part of the shelves where from mountain areas (92 per cent mountain /8 per cent non-mountain). Even if controlling by repeated products the diversity of these was high. In order to make the analysis possible, the products from the survey were classified into 18 food product categories: mineral water, soft drink, cheese, other dairy, apples, pears, beef, fish, pig meat, sheep meat, poultry products, venison, moose, ham, sausage, other meat products, bread, honey and other food products. These products were further classified into 6 groups: beverages, dairy, fruits, meats, meat products and other products. The most popular product in the sample was cheese, with information was collected in 5 of the countries (except in Romania). It was followed by mineral water and sausages, which were collected by 4 countries. As regards the sampled outlets, these were classified into the following categories: cash and carry, discount shop, factory outlet, farmers shop, farmers market, foreign supermarket, hypermarket, mini-market, national supermarket, regional supermarket, specialty shop, vending machine and web shop. Most of the shelves collected came from national supermarkets (146 shelves or 26.8 per cent), specialty shops (97 shelves or 17.8 per cent), mini-markets (94 shelves or 17.3 per cent), and farmers markets (44 shelves or 8.1 per cent). As mentioned, the shelves surveys collected information about prices for mountain and similar non-mountain food products, which are the basis for the empirical work done in this paper. Table 3 present the information about the all the mountain food products for which an equivalent nonmountain food product price was present in the database. As shown, overall 22.7 per cent of the products had an equivalent non-mountain food product price recorded in the database. However, this percentage varied dramatically from one product to another and from one country to another. It should be noted that whilst this may reflect problems in the data collection, it can also be due to the fact that some of the products do not have equivalent non-mountain ones. Methodology As pointed out by Combris et al. (1997) the hedonic price method is a useful approach to study the price-quality relationship of a product. The method consists of a regression analysis of the price on the characteristics of the product. It has been used for both durable (e.g., automobiles) and non-durables (e.g., wine, cereals)1. 1 See Combris et al. (1997) for references about hedonic regressions analysis applied to the different type of products. Premia for differentiated products at the retail level Table 2: All the countries: Results of the shelves survey - results per country (counts) Austria France Norway Romania Scotland Slovenia Total Number of shelves By type of shelf Beverages Dairy Fruits Meats Meat products Other products 100 27 73 0 0 0 0 91 34 12 44 0 1 0 105 10 34 0 25 15 21 90 90 0 0 0 0 0 88 0 43 0 31 14 0 90 0 65 0 0 25 0 564 161 227 44 56 55 21 By type of outlet Cash and carry Discount shop Factory outlet Farmers shop Farmers market Foreign supermarket Hypermarket Mini-market National supermarket Regional supermarket Speciality shop Vending machine Webshop 11 4 0 0 13 0 0 0 22 18 26 0 6 0 10 0 1 7 0 28 9 29 1 6 0 0 0 23 0 0 3 0 0 6 61 0 11 1 0 1 0 0 0 0 0 0 89 0 0 0 0 0 0 0 3 19 1 0 0 10 18 0 37 0 0 0 0 11 13 20 11 0 0 16 2 17 0 0 12 37 14 33 44 11 28 114 146 21 97 1 6 According to mountain area In mountain areas Out of mountain areas 92 8 47 44 46 59 44 46 53 35 54 36 336 228 Number of products Beverages Mineral water Soft drink Dairy Cheese Other dairy Fruits Apples Pears Meats Beef Fish Pigmeat Sheepmeat Poultry products Venison Moose Meat products Ham Sausage Other meat products Other products Bread Honey Herbs and spices Other food products 410 94 91 3 316 293 23 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 230 95 95 0 23 23 0 74 66 8 0 0 0 0 0 0 0 0 36 14 22 0 2 0 1 0 1 283 34 32 2 95 59 36 0 0 0 68 4 0 10 44 1 3 6 39 0 39 0 47 17 2 20 8 246 246 246 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 232 0 0 0 84 76 8 0 0 0 109 34 11 13 15 8 28 0 39 0 15 24 0 0 0 0 0 364 0 0 0 271 155 116 0 0 0 0 0 0 0 0 0 0 0 93 2 35 56 0 0 0 0 0 1765 469 464 5 789 606 183 74 66 8 177 38 11 23 59 9 31 6 207 16 111 80 49 17 3 20 9 Total number of different outlets In mountain areas Out of mountain areas 68 64 4 77 33 44 35 11 24 90 44 46 37 19 18 44 28 16 351 199 152 The implicit price of a characteristic is defined as the derivative of the price with respect to the product attribute. Rosen (1974) has shown under which market conditions the implicit price can be interpreted as the value consumers place on an additional unit of the characteristic. If the estimated implicit price is not significantly different from zero, then the characteristic is not valued by consumers, or the characteristic is not considered important or relevant in connection with the product. Thus, the starting point is the estimation of the following equation: Table 3: All the countries: Distribution of cases for which the price of an equivalent "non-mountain" product was recorded in the database Austria France Norway Romania Scotland Slovenia Total Number of products 162 114 63 0 14 49 402 Total 410 230 283 246 232 364 1765 Beverages 45 55 0 0 0 0 100 Total 94 95 34 246 0 0 469 Mineral water 43 55 0 0 0 0 98 Total 91 95 32 246 0 0 464 Soft drink 2 0 0 0 0 0 2 Total 3 0 2 0 0 0 5 Dairy 117 5 31 0 14 40 207 Total 316 23 95 0 84 271 789 Cheese 107 5 12 0 14 20 158 Total 293 23 59 0 76 155 606 Other dairy 10 0 19 0 0 20 49 Total 23 0 36 0 8 116 183 Fruits 0 30 0 0 0 0 30 Total 0 74 0 0 0 0 74 Apples 0 24 0 0 0 0 24 Total 0 66 0 0 0 0 66 Pears 0 6 0 0 0 0 6 Total 0 8 0 0 0 0 8 Meats 0 0 3 0 0 0 3 Total 0 0 68 0 109 0 177 Beef 0 0 1 0 0 0 1 Total 0 0 4 0 34 0 38 Fish 0 0 0 0 0 0 0 Total 0 0 0 0 11 0 11 Pigmeat 0 0 0 0 0 0 0 Total 0 0 10 0 13 0 23 Sheepmeat 0 0 2 0 0 0 2 Total 0 0 44 0 15 0 59 Poultry products 0 0 0 0 0 0 0 Total 0 0 1 0 8 0 9 Venison 0 0 0 0 0 0 0 Total 0 0 3 0 28 0 31 Moose 0 0 0 0 0 0 0 Total 0 0 6 0 0 0 6 Meat products 0 23 9 0 0 9 41 Total 0 36 39 0 39 93 207 Ham 0 11 0 0 0 1 12 Total 0 14 0 0 0 2 16 Sausage 0 12 9 0 0 1 22 Total 0 22 39 0 15 35 111 Other meat products 0 0 0 0 0 7 7 Total 0 0 0 0 24 56 80 Other products 0 1 20 0 0 0 21 Total 0 2 47 0 0 0 49 Bread 0 0 16 0 0 0 16 Total 0 0 17 0 0 0 17 Honey 0 1 0 0 0 0 1 Total 0 1 2 0 0 0 3 Herbs and spices 0 0 0 0 0 0 0 Total 0 0 20 0 0 0 20 Other food products 0 0 4 0 0 0 4 Total 0 1 8 0 0 0 9 Y¡ = a0 + «iZji +a 2Z2i +... + a ^^ + u (1) Where Yj are the product prices, the Zj are the attributes and the a; are the parameters of the regression. The attributes considered in the analysis were introduced by means of dummy variables (i.e. dichotomous variables that take the value of 1 when a characteristic is present and 0 otherwise). The procedure used to introduce the dummies into the regression was the one in Oczkowski (1994), which avoids choosing a base category for the comparisons. For instance, one could consider in the case of the mineral water regressions, the category base 'still water from non-mountain origin sold in non-mountain areas by non-specialised stores' and all the parameters of the dummy variables in the regression would indicate deviations with respect to the base category. Thus, the parameter associated to a variable "mountain origin" would indicate whether 'still water from mountain origin sold in non-mountain areas by non-specialised stores' would receive a different price than the base category. Instead one may consider that all the parameters from the dummies indicate deviations with respect to the mean price but this requires reformulating the typical approach used when de aling with dummy variab les. The procedure used in this paper to introduce the dummyvaaiables into t:he regression -presented laere for comebateness sake- aanbe anpiainedbp meamof a simphe two dummy varieMemodehY = c^Dj-t a-D2 b e wherr is ihe first dummy variaHe rhat takrs tie raluy of 1 if saythr storaie ena monniain arerand U olberwiss; D2 ir thr aecrnd dummy vrriaMethat: taker lhr vafoe of 1 if tire store is -n r nam-mountain area rnd 0 otharwite. By conrtruatfon the two dummirs adduptoi ii.e. -U^e stars crn be m r mountain means outsfoe aa rtlaud thtrefore, onfy one snuuidbt cctcesibtcec:C in tise re-tsrsreon. Huwrnea, i= is possiUleto tmpeee aconeiraintin the regressionruch that the paramateas associatedtrthe dummies becomedeviations witcd eessect ds dlie mean ot tire d-tend-nr vaeiadis iwhich is measured by the intercept, i.e., = y ). Thus, using the corvDj va2D2=l,it is possible to estimate all the parameters from the model by running the following two regressions (2') and (2"): Y = an + 0 + «1 Y = ao D1 "I^D D2 - IpfPl (20 (2'0 The dummy variables in the analysis comprised four groups: first, attributes associated to whether the product was a mdunluin product, whfoh mcluded three dumnty vuriahles: (1) ^tar mountain ueodue- did noi have an ruaivalanr prodnct in At datanane, C2) the mountain vrodurt uar an aquivalert non-mountam ^j-^ductr in tire databaae and (3) no^ mountamfobdpeoduair/Saconn, attribu-er =rsociated to thr locationof fte stores,whichaunstrtedou tawa dummier:ili the rCrod war in a monrrtain aerr and li) the shiap waa natt in themauetain a^^tteri!, ai^raiti^tc^in assndatentothetype of store, which comprised three dummies: (1) small non-speriahseh rhop iLe^ dcseount svap, mini-market dending machine and web shop.), (2) specialised (shop factory outlet, farmers shop, farmers market, specialty shop and regional supermarket), (3) supermarkets and similar stores (i.e., cash and carry, foreign supermarket, hypermarket, national supermarket). Fourth, attributes associated with the product types (e.g., type of apples), which depended on the product and can be found on the regression notes. RESULTS AND DISCUSSION The results are presented in Table 4 and Table 5. Although the regressions could have been run for all the products in the database as far as the product price was recorded, the main idea of the paper was to compare the price of similar mountain and non-mountain products. Therefore, only those cases where a sizable number of non-mountain food products was present (at least 4 cases). In addition, the analysis was performed differentiating by products and countries. The statistical significance of the parameters associated to the variables x1 and x2 in the table indicate that the prices of the mountain food products (in the group without and with equivalent product) are different from the mean (above or below depending on the sign of the parameters). This was the case for sausage in France and cheese in Austria, France and Norway. In the case of Scotland and Slovenia the prices were notdifferent than the mean value. The parameters corresponding to x2 and x3 allow testing athe lhypothesis whether mountain food products carry a llpremium with respect to the non-mountain products. A epremium was found only on the case of cheese and only for Austria, France (though favouring non-mountain products), ttNorway and Slovenia. In Austria the parameter of x2 was not statistically different than zero but the non-mountain products was -1.125 €/Kg (i.e., 1.125 was the size of the premium). In the case of Norway, the premium was found to be more substantial and equal to 23.1 €/Kg and in Slovenia, itwas 2.5 €/Kg. As regards whether the location of the store had effect on prices (related to variables x4 and x5) it was only found positive in the case of Austria and Slovenia. In the case of Austria mountain areas carry a higher price in the case of water (in the case of cheese, the same is observed but it is not statistically significant). In the case of cheeses in Slovenia, the situation is just the opposite and it is store in non-mountain areas the ones that carry a premium. Variables x7 to x9 indicate that in some case specialised shops carry prices above average (this is for all the products not just mountain products). This is found for the case of cheese and water in Austria and only water in France. As for the remaining variables (product type) several lcharacteristics brought differences in prices but not in a lsystematic way. Overall the mixed results obtained from the empirical analysis may indicate that probably in not all the cases the mountain attribute can operate as a creator of value (i.e. a cc source of differentiation in the eyes of consumers or buyers) and this may differ by product and country. Table 6 is an attempt to organise the possible cases that may arise. Table 6 considers three degrees of differentiation: a G M O O 3 < O O c M M 0) C D O O T3 0) O V CU M 0) I T3 £ re en a) ra re M D re M « V n Q. re M £ <5 tfl tfl Sí ra Sí o c o ■c a) X a> 12 .ra O w a = a !« O O o O e & < o O > ft c CM M « QP C CM M « QP C CM o -h 00 o m o IS T3 =3 T3 -G 13 I .5 ¡ 13 > -73 la J2 ü o T3 C a '" u -s u g JS > ft J2 O _o ¡M tí fr » 8 -£3 00 X X ^ ¡^ ¡^ S „ II II II U o ^ o o M , ¡3 M M tí ■13 Ph C3 > uni> unoo^r [^00—I 00 in^ooooc^ r^ r^ r^ r^ r^ oo o^tNon^r un r^ r^ r^ r^ r^ r^ >—i oo 3 (3 o 13 tí u o ^ 2 13 M u S tí .2 H (U % £ C3 .tí O K £ * (U (U M & (U u .tí ■tí o u ^ ^ tí a a § 2 Q O 1—1 ii SÍ2 .tí H ffi O ■ i u