FACULTY OF ECONOMICS AND BUSINESS MSc ECONOMICS – MARKETS AND REGULATION
MARIA CRISTINA FETT FURTADO DE ANDRADE (11085533)
THE REGION-‐OF-‐ORIGIN EFFECT IN BRAZILIAN SPECIALTY COFFEE PRICES
AN EMPIRICAL ANALYSIS
SUPERVISOR: DR. SANDER ONDERSTAL AMSTERDAM, JULY 2017
Statement of Originality
This document was written by Maria Cristina Fett Furtado de Andrade, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
1 Introduction
Coffee was for a long time considered an almost homogeneous product (Teuber, 2010). Aligned with the trend of increasing interest for differentiated foods and beverages, a larger number of consumers are showing increasingly willingness to pay for differentiated coffees (Donnet et al 2007; CBI 2016) and roasters are looking to purchase high quality green coffee beans, which will potentially provide an exceptional cup to the final consumers (CBI 2016; Donnet et al 2008). Coffee is an experience good, i.e. its quality and unique characteristics can only be assessed after consumption. Moreover, coffee is also a complex differentiated good (Donnet et al, 2008), hence, not only the objective sensorial quality seems to determine prices, but also other reputation attributes of the considered coffee lot (Donnet & Weatherspoon 2006).
Studies (Donnet et al 2008, Teuber 2010, Teuber & Herrmann 2012, Wilson & Wilson 2014) show that one of the attributes that has a significant impact in specialty coffee prices is the quality reputation of the geographical origin of the product. Anecdotal evidence of the relevance of the origin as a differentiating tool is the fact that, in the market for specialty coffees, single-‐origin beans are almost the rule, being such origin information stressed by roasters in the packaging. This is in contrast with the mainstream commodity market, where a number of different origins (unknown to consumers) are blended together in order to achieve a homogeneous product in a large scale quantity (Donnet et al 2007, Teuber 2010, Lewin et al 2004).
Specialty coffees are defined by the Specialty Coffee Association of America (SCAA, 2009) as beans of outstanding quality (score of 80 points or higher in standardized sensorial grading) and with ‘unique flavor profiles, produced in special geographic microclimates’ (Rhineheart, 2009). Such definition of specialty coffees has been more and more accepted in the coffee industry (Donnet et al 2008). This definition stresses the impact of the conditions under which the coffee beans were produced, as configurations of the geographical origin influence directly the objective quality of the bean. Furthermore, the origin also has a direct impact on prices of specialty coffees through the region’s reputation, having such reputation been built upon the quality of the coffee provided by such region in the past (Shapiro, 1983), hence, it is a common reputation of producers from the region (Winfree and McCluskey 2005). It is important to notice that this definition makes reference exclusively to the quality of the cup, leaving production process credence attributes aside (such as ecologically or socially friendly certifications) (Donnet et al 2008).
The idea of geographical origin being used as a differentiating tool for specialty coffee reaches, besides the industrial organization scope, also a social scope. This movement towards specialty coffee consumption puts coffee growers, which are mostly small scale farmers, as the main responsible in producing the product attributes desired by consumers (Donnet et al 2007). When, indeed, the geographical origin is valued by consumers, it allows all the producers from the considered region to benefit from the shared reputation benefits. This different positioning provides such small farmers with some market power, thus providing coffee producers with the opportunity to earn a significant price premium for their product (Donnet et al 2008).
Although the specialty coffee market is well documented, empirical studies highlighting what determines the prices of specialty coffees are scarce, and there is a gap in the literature regarding the effect in prices of the reputation of different Brazilian coffee producing regions. Region-‐of-‐origin effect analysis has only been performed for three nations: Colombia (Teuber 2010), Ethiopia (Teuber 2010) and Honduras (Teuber 2008). Considering the importance of Brazil as a coffee producing country, having established its position as the main coffee exporter on the planet
since the beginning of the 20th century and being currently the second largest specialty coffee exporter in the world (BSCA, 2016), I intend to investigate the possible impacts of reputation of different coffee producing regions in the price of Brazilian specialty coffee. Thus, the main question I will try to answer is: What are the region-‐of-‐origin effects in the prices of Brazilian specialty coffee?
Considering the complexity of the characteristics of specialty coffee, it may be viewed as a bundle of attributes (Donnet et al 2008). Thus, to address my research question, I use the hedonic regression approach, following Teuber (2008). Teuber (2008) assumes that the observed price results from the implicit prices of each of the attributes that compose the product. The attributes considered are the information observed by consumers prior to purchase. To answer my research question, I retrieved data from the Cup of Excellence (CoE) competition in Brazil, the highest acknowledged competition in the coffee industry from 2012 up to 2016, in order empirically analyze the impact of regional reputation in the specialty coffee prices.
This thesis is divided as follows. Section 2, the literature review, will pursue to determine if region-‐of-‐origin can be considered a credible quality signal to consumers in an environment of imperfect information, as well as it will shed light on the main empirical findings derived from previous studies on price determinants of specialty coffee. Section 2 will also ground the choice of Brazil as the origin to be analyzed in this paper, and will pursue to characterize coffee production in the country. The goal of Section 2 is to consider empirical and theoretical elements that suggest that the investigation of region-‐of-‐origin effects in Brazilian specialty coffee prices is a relevant subject. Section 3 will present the methodology of the empirical model that was estimated in order to help answering my research question. Section 3 will include a detailed description on the Cup of Excellence competition, the source of the data used in this paper’s regression, as well as information regarding the variables considered and justification for the functional form chosen. Section 4 will then present the regressions results and in Section 5, conclusions will be drawn.
2 LITERATURE REVIEW
This section includes a review of the literature related to my research question. Subsection 2.1 will discuss the use of quality signals in an environment of asymmetric information as a mean to correct the market imperfection. Subsection 2.2 will then explore the empirical papers that have researched on the price determinants of specialty coffee, the methodology applied, the particularities of such studies and their main findings. Subsection 2.3 will add insights into the analysis by presenting characteristics of the Brazilian coffee market, as well as presenting evidences that indicate that different regions could have different reputations which would wield differentiated prices. Subsection 2.4 will present the main conclusions drawn from the literature review, aiming to ultimately justify the further empirical investigation on region-‐of-‐origin effects on Brazilian specialty coffee prices that will be performed in Section 3.
2.1 Reputation: quality signaling on imperfect information environment
The trend towards specialty food products has been increasing in the past decades (Carriquiry and Babcock 2004, Winfree & McCluskey 2005). Coffee consumption patterns seem to be followingsuch trend. The growth of the international demand for specialty coffee is estimated in the range of 5-‐10% each year since 1990s (CoE in Brazil and Honduras, Impact assessment 2015). Green coffee is a highly differentiated product (Donnet et al 2010), being differences in quality acknowledged by the industry. Green coffee quality classification consists of two steps. First, a physical analysis of a sample of the lot, consisting of the accountability of different defects commonly present in green coffee beans (SCAA, 2017; Teixeira et al 1995); samples with a high number of physical defects have less potential to provide a good cup (Teixeira et al 1995). After going through the physical analysis, the samples are cupped in order to evaluate the sensorial characteristics of the green coffee in hand and then graded on a scale that reaches up to 100 points. The flavor attributes to be evaluated according to the Specialty Coffee Association of America protocol include, but are not limited to: aroma, flavor, aftertaste, acidity, body, sweetness, defects and overall perception (SCAA, 2017). The SCAA considers any coffee granted 80 points or higher in the sensorial evaluation as being a specialty coffee. Even though there is an evident effort in the coffee industry to standardly categorize the quality through detailed sensorial classification and summarize it on a quality rating (Donnet et al 2010), coffee presents a number of complex characteristics (Donnet et al 2010). Coffees vary widely regarding its attributes, especially concerning flavor and aroma notes (Donnet et al 2008, Donnet et al 2010).
Reputation attributes are a relevant signal in an environment where asymmetric information is present (Shapiro 1983), which is the condition faced by consumers when dealing with experience goods. Experience goods are goods whose attributes can only be fully acknowledged after consumption (Carriquiry and Babcock, 2004). Most food products are experience goods (Winfree & McCluskey 2005), coffee included. If the attributes of a product could be completely observed prior to the purchase, consumers would be able to perfectly inspect and subsequently form their perception on the quality of the product in hand, leaving aside the impressions obtained through reputation (Shapiro 1983). On the other hand, when the attributes of a product are not perfectly observable prior to purchase, reputation serves as a signal. Consumers may use the quality of the output of the firm in the past as an approximation for current quality (Shapiro 1983, Winfree & McCluskey 2005), thus, solving the market imperfection. If the reputation of the origin is positive, the legal protection of such can serve as an interesting marketing tool (Winfree & McCluskey 2005), which helps justifying the increasingly adopted strategy of legal protection of coffee producing regions around the world (Teuber 2010, Teuber 2008).
Geographical indications of origin have been increasingly applied in the specialty food industry in order to signal quality to consumers (Winfree & McCluskey 2005) and this trend can also be identified in the green coffee industry (CBI, 2016). The term ‘specialty coffees’ and its definition were first proposed by Erna Knutsen in 1978, where she stressed the fact that specialty coffees would be those with differentiated note profiles originating of distinguished microclimates (Rhineheart 2009). The microclimates mentioned by Knutsen reference origin as an essential determinant of differentiated characteristics, thus, reinforcing the idea that origin is a credible reputation attribute for coffee. The use of geographical indications of origin as a credible quality signaling tool is justified by the existence of a collective reputation shared by local producers of the good (Winfree & McCluskey 2005) and the communication of the origin thus signals the existence of desired attributes to consumers.
2.2 Hedonic regressions and past studies of coffee reputation attributes
It seems justified that the impact of origin reputation in specialty coffee prices should be examined. Empirical studies performed on coffee show that, indeed, reputation attributes play a role in the definition of prices. Origin of production has been statistically confirmed to be an important signal to consumers (Donnet et al 2008, Teuber 2010, Teuber & Herrmann 2012, Wilson & Wilson 2014), but also the communication of other reputation attributes seems to have an impact in prices. Such empirical studies use the hedonic regression approach, where the product in hand is considered to be a bundle of its characteristics (Donnet et al 2008).
Hedonic pricing regressions assume that the observed price of a good is a function of the implicit prices of each attribute that composes the good (Teuber 2008). Therefore, coffee prices are assumed to be a function of the coffee attributes communicated to consumers prior to the purchase (Donnet et al 2008). Hedonic regressions have been widely applied in the analysis of price determinants of a number of products, specially wine (Donnet et al 2006). Hedonic regressions are a useful tool because they allow the understanding of what are the characteristics that are preferred by consumers, which is key to develop a sustainable supply chain, as coffee growers can then focus their production practices on obtaining the attributes valued by consumers (Donnet et al 2008).
The empirical hedonic studies on coffee consulted where performed over data retrieved from the Cup of Excellence competition, where top quality coffees are selected and auctioned. Origin of production was included as a reputation attribute, together with other variables that impact the objective quality of the green coffee bean, i.e. altitude and coffee tree variety and scarcity signals, i.e. lot size and coffee area size. The sensorial evaluation rating achieved in the competition was also considered in the author’s regressions (Donnet et al 2008, Teuber 2008, Teuber 2010 and Teuber & Herrmann 2012, Wilson & Wilson 2014), and it is considered to be a proxy of the taste experience of consumers upon inspection. Furthermore, the ranking in the competition is available to bidders, which communicates the competitive placement of that lot relative to other lots in the competition that year (Donnet et al 2008).
The data set has some particularities considering it is a competition and the considered studies (Donnet et al 2008, Teuber 2008, Teuber 2010, Teuber & Herrmann 2012, Wilson & Wilson 2014) find very significant effects in prices of both rating and ranking, which indicates that such attributes are important quality signals. It is relevant to point out that the considered studies (Donnet et al 2008, Teuber 2008, Teuber 2010, Teuber & Herrmann 2012, Wilson & Wilson 2014), as well as this thesis, perform their hedonic analysis on the procurement level. It is assumed that the demand in the procurement level is derived from the demand in the retail level (Teuber 2008). It is also crucial to stress the key role played by roasters in the education of consumers about specialty coffee attributes, in these fairly early stages of specialty coffee market development (Donnet et al, 2008). Coffee roasters preserve the origin identity of the coffee by including such information in the final product, which informs consumers about the different origins and educates them about attributes associated with specific origins (Donnet et al, 2008). In addition, considering the ease of communicating the rating and the ranking throughout the coffee supply chain (Donnet et al 2008), it shall not be ignored the possibility that the reputation effect of some variables could have been absorbed by rating and/or ranking, like it was hypothesized by Teuber (2008) and will be discussed further.
Therefore, considering that the importance of origin and other attributes as credible quality signals in the coffee industry has been explored by previous studies, the main insights originating from such papers will be analyzed in the following subsubsections. Subsubsection 2.2.1 will focus exclusively on the empirical findings regarding the impact of the reputation of the origin in specialty coffee prices. Subsubsection 2.2.2 will focus on other reputation attributes included in the discussed author’s regressions, in order to shed light on which are the attributes that can be statistically confirmed to signal quality to consumers.
2.2.1 Origin as a reputation attribute
The earliest attempt on identifying what determines prices of specialty coffees was made by Donnet et al (2008), where she used the hedonic regression model in order to estimate the implicit prices of the attributes of specialty green coffees auctioned in the Cup of Excellence (CoE) in an approach that seems to follow previous hedonic studies on price determinants of wine. Donnet et al (2008) considered two classes of attributes: one sensorial attribute – the rating in the cupping competition (objective quality); and four reputation attributes, namely the geographic origin, variety of the tree, ranking (competitive placement) in the competition and number of lots available (scarcity indicator). The authors also included control variables for years and for the commodity coffee price in the respective years. Twenty-‐one CoE competitions were considered, summing to 541 observations. The results from the hedonic regression indicate that quality ranking (hence, quality relative to other coffees in the competition that year) has the largest impact in prices, followed by the quality rating variable (which corresponds to the objective quality measurement). A significant inverse relationship was observed between the quantity available and the price, indicating that in the specialty coffee market consumers value exclusivity and scarcity. The coffee tree variety Pacamara had the only significant coefficient (10%), being statically more valued than Bourbon. About reputation effects of different countries of origin, the author considered Brazil, Colombia, Bolivia, El Salvador, Honduras and Nicaragua in the sample; having Brazil as the base category, four out of five country coefficients were found to be significant (but Bolivia, insignificant); all the nationalities considered were found to have a discount relative to Brazil when considering the reputation of the geographic origin.
Teuber (2010) focused on the impact of the reputation of the geographic origin, having performed not only a nation comparison (similar to Donnet et al 2008), but also approaching the origin effect in a regional level for Colombia and Ethiopia. Teuber (2010) includes, beyond the variables adopted by Donnet et al (2008), the coffee area (scarcity) and a dummy variable to control for coffees with certification schemes and she excludes the commodity coffee price indicators, sticking to year dummies; she also performs her analysis using data from the CoE competition. For the nation comparison, data from 2003 up to 2007 was considered, summing up to over 700 observations, and the coefficients estimated show the ranking to be the most impacting variable, followed by the rating; the quantity of lots available kept having a significant negative effect in prices, whilst the coffee area has a positive significant impact in prices. Organic certification schemes seem to have a positive impact in prices. No significant coffee tree variety effect was estimated. The country of reference considered by the author was Honduras and six out of seven nations have significant coefficients (except Costa Rica), being Honduras coffee discounted relative to all the
significant country coefficients. The highest premium achieved in comparison to Honduras is in Guatemala, followed by Bolivia and subsequently by Brazil.
Teuber (2010) also performed regressions for Ethiopia and Colombia regions, in order to measure the impact of the reputation of different regions of such countries in coffee prices; the author used data for 2005 and 2006 and did not control for varieties nor for certifications. The author included 111 observations for Colombian coffees retrieved from the CoE competition and 53 observations for Ethiopian coffee retrieved from the Ecafé Gold competition. For Colombia the ranking position has the highest impact on prices, having the rating also a significant impact. A significant impact of lot size in prices was observed, being such relationship negative. Five regions were considered; Huila was the reference category and all regions considered have highly significant coefficients, proposing an impact of regional reputation in prices; coffee producing regions in Colombia seem to have a discount relative to Huila, but the region of Meta. Considering Ethiopia, the rating has the highest effect in specialty coffee prices and the ranking placement doesn’t present significant coefficients. Scarcity has a significant positive impact in prices; two regions were considered, being the region of Yirgacheffe the base. The estimated coefficients suggest the existence of reputation effects in the prices of specialty coffees coming from different regions of Ethiopia, being the region of Sidamo discounted relative to Yirgacheffe.
Teuber (2008) further explored the effect of region-‐of-‐origin reputation in prices of specialty coffees, this time exclusively for the country of Honduras. The author stresses that Honduras had identified five coffee producing regions, having each a distinguished cup profile. Furthermore, one coffee producing region in Honduras, namely the region of Marcala, had legally protected its origin denomination in 2005, being the first legal protection of a coffee producing region in Honduras and in Central America (Teuber, 2008). The initiative of legally protecting the region of Marcala aimed to create awareness in the specialty coffee market about the quality of the coffee offered by Honduras, as such country’s product had been internationally recognized as being of average quality (Teuber 2008, CoE Brazil and Honduras, impact assessment 2012). For the regression, the author considered data from the Cup of Excellence in Honduras from 2004 up to and including 2007, summing 119 observations. The author added the altitude in the regression, controlled only for the first placement in the ranking and changed the scarcity variable coffee area for farm size, maintaining all other variables from the previous approach (Teuber 2010). The results indicate that the rating has the highest significant impact on the prices, having the lot size also a significant negative impact in green coffee prices; the ranking, as well as the altitude are significant coefficients. No statistically significant coefficients were computed for coffee tree varieties. The results, however, do not indicate a region of origin effect for Honduran coffee, in contrast with results found for Ethiopia and Colombia by the author (Teuber 2010). Teuber (2008) hypothesized that perhaps the variable rating could be absorbing the effect of the regions and estimated models excluding the score received by the lots in the competition. The regions remained having statistically insignificant coefficients and thus the hypothesis of a region-‐of-‐origin effect for Honduran coffee could not be confirmed.
Teuber & Herrmann (2012) later expanded her work by proposing an interaction term between the rating (current quality) and the origin (reputation), a ‘country-‐specific score effect’, arguing that the previous models presented were too simplistic to detect more complex effects. Moreover, the author included the nationality of the buyers (North American, European or Asian) in order to identify if the nationality of the buyers affect the valuation of different specialty coffee attributes. The investigation was performed in data from the Cup of Excellence competition from 2003 up to and including 2009, summing 1250 observations.
In the model Teuber & Herrmann (2012) excluded the farm size and the altitude from the regression, including an interaction term between country of origin and rating. The results obtained showed very significant results for the rating, the rankings (1st to 3rd) and for quantity available.
Statistically significant results also were obtained for organic certification and some varieties (Caturra and Bourbon). Regarding country of origin, the base chosen was Brazil and five out of seven countries presented very significant results. The significant coefficients computed indicate a discount of coffee origins relative to Brazil, with the exception of Guatemala, which seems to have a better reputation than Brazilian coffee. Coefficients for Colombia and Bolivia were statistically insignificant. Furthermore, the interaction term between origin and rating indicate that the rating achieved is a more important price determinant for origins with a lower reputation. The segmented regression proposed by Teuber & Herrmann (2012) indicated that indeed origins are perceived differently depending on the location of the buyer.
Finally, one last study could be found in the literature on hedonic price regressions for specialty coffee. Wilson & Wilson (2014) performed more sophisticated regressions on the Cup of Excellence data. The main contributions of Wilson & Wilson (2014) can be seen as the introduction of truncated regression models, as well as the addition of further interaction terms than the ones considered by Teuber and Herrmann (2012). Wilson & Wilson (2014) also divided the groups of consumers by continent of origin. The authors included the altitude of the farm where the coffee was grown, the coffee area of the farm and certification schemes in their regressions. Taking into account country of origin, the authors included Bolivia, Colombia, Costa Rica, El Salvador, Guatemala, Honduras and Nicaragua, being Brazil the base category. The results of the many regressions performed indicate that all the considered regions are priced at a discount relative to Brazilian coffee, and all the coefficients are highly statistically significant. Considering the increasing sophistication of the models implemented by Wilson & Wilson, the country-‐of-‐origin coefficients perform fairly similarly. The authors argument on the premiums received by Brazil as either being the result of the reputation of the Cup of Excellence competition in the country (as Brazil was the first nation to host such competition) or as being the result of an appeal of exclusiveness, given that Brazilian commodity coffee is seen as of lower quality (Wilson & Wilson 2014).
Considering the previously mentioned studies, empirical evidence seems to suggest that indeed, for coffee, origin matters. This is an interesting acknowledgement considering my intention to investigate the effect of reputations of coffee producing regions in the prices of specialty coffee. Furthermore, it is interesting to point out the unexpected empirical evidence that Brazilian coffees receive a premium relative to other coffee producing nations such as Colombia (Donnet et al 2008, Teuber 2010, Wilson & Wilson 2014). This conclusion is surprising considering that Brazilian commodity coffee is generally priced at a lower premium relative to Colombian coffees by the group price indicator of the International Coffee Organization (Figure 1, Appendix).
Therefore, after having identified the existence of origin reputation effect in specialty coffee prices, we shall briefly analyze the other attributes (sensorial and reputation) that will be included in my hedonic regression (following the approach of Teuber 2008). A summary of the main empirical findings of the effects of such attributes in specialty coffee prices will be presented next.
2.2.2 Other coffee reputation attributes effect on prices
Other attributes seem to help determining prices of specialty coffee, apart from origin. The other attributes that were included by the authors (Donnet et al 2008; Teuber 2008 and 2010,
Teuber & Herrmann 2012, Wilson and Wilson 2014) and that will also be present in my regression will be explored further in this subsubsection. All mentioned studies have found statistically significant effect of the sensorial variable score (rating) as well as of the ranking in prices. The rating indicates the sensorial experience achieved when consuming the considered lot (Donnet et al 2008), hence, higher ratings indicate a more pleasant cup experience while lower rates indicate a less pleasant cup experience. The ranking, on the other hand, indicates the competitive placement of the considered lot relative to other lots competing that year (Donnet et al 2008).
Although the rating and the ranking are correlated variables, i.e. the lots with higher ratings will also be on the top of the ranking, the ranking seems to communicate a more powerful message; it could be a bad year for the quality of the coffee of a certain country, but the appeal of being ‘the champion’ would still remain (Donnet et al 2008). Furthermore, both rating and raking are quality signals that are easily understood by consumers, thus, being valuable for the communication of quality though the coffee supply chain (Donnet et al 2008). It is important to point out that ranking could be capturing the effect in prices of other attributes, as the ranking is a reputation attribute that is of easy understanding by unexperienced consumers and actors in the supply chain that may not be familiar with the impact of more complex reputation attributes in the objective quality. This capture of effects was hypothesized by Teuber (2008) relative to rating, considering that no significant effect could be found for regions or varieties in the Honduran case, although the author was not able to confirm the hypothesis.
The size of the lots being auctioned also communicate a message to consumers, this time about scarcity and exclusiveness. All studies mentioned previously (Donnet et al 2008; Teuber 2008 and 2010, Teuber & Herrmann 2012, Wilson & Wilson 2014) identified statistically significant negative coefficients for lot size, indicating that the smaller the quantity available is, the more valued it is. The coffee area/farm size where the lot was produced could also be seen as an indication of scarcity (Teuber 2010). Teuber (2010) included farm size and Teuber (2008) and Wilson & Wilson (2014) included coffee area in their regressions. Teuber (2010) and Wilson and Wilson (2014) found a significant positive relationship between prices and the size of the coffee area within the farm; this counter intuitive result could be due to the fact that bigger farms usually have better resources to improve cultivation practices. Cultivation practices are amongst the factors that can impact the quality of green coffee beans (Söndahl et al 1995). No significant effect for farm size was found in the model by Teuber (2008). Furthermore, certification schemes seem to have a positive effect in prices. Teuber (2010) and Teuber & Herrmann (2012) found significant effect in prices for organic certification schemes, which seems to be aligned with the increasing interest of consumers regarding the conditions under which their food products were produced (CBI, 2016).
Altitude and variety of the coffee tree are also variables that are communicated to bidders prior to the Cup of Excellence auction and that potentially impact the prices through reputation effects. Both altitude and coffee variety impact the objective sensorial quality of the green coffee beans (Söndahl et al 1995), thus, reputation effects could be expected if such attributes are communicated to consumers, which is the case in the Cup of Excellence competition. We shall start with varieties, as genetic aspects play a key role in coffee quality and are especially relevant for this discussion. There are two main coffee species commercially produced, Arabica coffee and Robusta coffee, and the differences between them are well documented (Söndahl et al 1995). Robusta is generally recognized as being of inferior quality due to the higher amount of caffeine and poorer quality of the cup; chemical evaluations of the two coffee species have been conducted and chemicals distinctions have been found, being such believed to justify a part of the cup quality
differences (Söndahl et al 1995). The differences between both coffee species are so accentuated that the prices for both species are defined separately in the international commodity market, being Arabica coffees priced according to the New York exchange and being the prices of Robusta defined at the London exchange (Lewin et al 2004).
Specialty coffees mainly come from beans originating in Arabica trees, but the array of varieties that mutated naturally or were bred by humans is substantial, some of them being particularly attractive due to their disease resistance (Green Coffee Book 2008; Söndahl et al 1995). Hybrids between Arabica and Robusta, although attractive due to its higher resistance to diseases to which Arabica is susceptible, generally result in a poorer cup when compared to pure Arabica varieties (Green Coffee Book, 2008). Empirical studies do not give conclusive evidence about the reputation impact of coffee varieties. Donnet et al (2008) estimated an implicit price premium of 24% for the Pacamara variety relative to the base variety Bourbon (significant at 10%); no other variety had a significant coefficient (Catuaí, Caturra and Typica; base Bourbon). Teuber (2010) did not find any significant coefficient for variety (Catuai, Caturra, Pacamara, Typica; base variety Bourbon). Teuber (2008) also did not have results that suggested the direct effect of variety in prices, as none of the coefficients for the varieties were significant (Catuai, Caturra, IHC-‐90, Pacamara, Pacas; base variety Bourbon). These results indicate that the variety of the coffee tree have not developed a reputation that is recognized by consumers, thus, being an inefficient mechanism for coffee price differentiation.
Considering altitude, Arabica plants originated in the highlands of Ethiopia and similar growing conditions than those of that region tend to yield a mild and rich beverage (Söndahl et al 1995). For Arabica coffee, medium to high altitudes (1000-‐2100m) are preferred when considering tropical regions and medium altitudes (400-‐1200m) are preferred if the local of production is further from the equator line (9-‐24 north or south). Coffee beans that grow in higher altitudes tend be richer in sugar and other soluble solids if compared to beans that developed in lower altitudes. It seems fair to say that indeed the altitude of the plantation influences the objective quality of green coffee beans. Considering the reputation effect of altitude in prices, Teuber (2008) and Wilson & Wilson (2014) included altitude in the model, being altitude found to be statistically significant, having a positive impact in prices.
One final variable that could be considered a reputation attribute if the communication to consumers affected prices is the post-‐harvest process method chosen. The post-‐harvest process, according to Söndahl et al (1995), can have an impact on objective sensorial quality differences in green coffee beans. The three main methods to process the green coffee beans after harvest are the natural process (dry), the pulped natural process and the wet process. Comparing the post-‐harvest processing methods, coffees processed using the natural method tend to have more soluble solid content if compared to the coffees processed using the wet method, ranking the pulped natural process in between (Teixeira et al 1995); this abundance of soluble solid contents translates into richer body in the cup. The fermentation process through which green coffee beans are submitted in the wet method, on the other hand, accentuate floral and fruity notes of the bean. The papers referenced in this thesis have not explored reputation differences between post-‐harvest process methods; the data collected for this paper, however, allows me to calculate the coefficients for such. Considering the insights obtained from subsubsection 2.2.1, where it was shown that empirical studies suggest that reputation of different coffee producing origins indeed affect prices of specialty coffee, the focus will now be turned towards Brazil. Subsection 2.2.1 suggests that Brazilian coffee is priced at a premium with respect to other origins, but no study regarding regional reputation
differences and its impacts has been conducted so far. Therefore, the next sub section will focus on characterizing Brazil and its coffee producing regions, in order to obtain insights if indeed reputations effects could be expected from different regions.
2.3 Particularities of Coffee production in Brazil
The production and exports of coffee played a key role in the development of geopolitical structure and the economy of Brazil (IBGE, 2016). By the end of the 19th century, Brazil produced
three fifths of the total global coffee production (Martins, 2012), having maintained its position as the main producer on the planet ever since. Coffee is ranked amongst the main agricultural export products of the country, being the main product for the creation of work places in Brazilian agriculture (MAPA -‐ Ministério da Agricultura, 2017). Nowadays, the country is responsible for about one third of the global coffee production, having produced in the crop of 2015/2016 50,376 million 60 kilograms bags (of both Arabica and Robusta) out of a total of 151,438 million bags produced worldwide (ICO, 2017). For means of comparison, the second largest coffee producer on the planet is Vietnam, which is mainly a Robusta producer, having yielded in the crop of 2015/2016 28,737 thousand bags. The third main producer is Colombia, mostly an Arabica producer, having yielded in 2015/2016 14,009 thousand bags (ICO, 2017). Brazil is also a big consumer of coffee, having internally consumed 20,5 million bags of coffee in the crop year 2015/2016 (ICO, 2017).
Coffee production in Brazil is mainly located in the Southeast region of the country, having the production historically shifted from the Rio de Janeiro state to Minas Gerais, São Paulo and Espírito Santo (IBGE, 2016). One important characteristic of this dynamics is the fact that the vast majority of coffee exported by Brazil is in the form of green coffee beans, also due to the strong position of importing countries’ roasters and brands (CBI trade statistics 2016). Particularities of Brazilian coffee farm practices shall be elaborated. Teixeira et al (1995) states that coffee farms in Brazil vary from small (5-‐50ha) to very mechanized large plantations (200-‐3000ha), being the average size of a coffee plantation in Brazil greater than the average size of farms in other coffee producing countries. Labor costs are higher in Brazil relative to other coffee producing countries, which has incentivized the mechanization of the production, especially of the harvest, in order to reduce costs and keep the product competitive (Coe in Brazil and Honduras – Impact Assessment, 2015). Almost all coffee in Brazil is grown without shade (Söndahl et al 1995), which yields a larger crop, but requires a greater amount of attention regarding cultural practices in order to yield a high quality product.
Arabica coffees produced in Brazil are generally characterized by mild acidity, aromatic notes and strong body (Söndahl et al 1995), but the country has great potential in the specialty coffee market. Given the many different coffee producing regions, the vast number of microclimates existent in each of them, the number of different varieties cultivated and also due to the different post-‐harvest processing methods, Brazil can produce and array of differentiated coffee flavor notes (Söndahl et al 1995). Although given the vast potential of Brazil in the production of specialty coffee, the country has an international reputation of being a large scale average quality commodity coffee producer to overcome (Coe in Brazil and Honduras – Impact Assessment, 2015). Brazilian commodity coffee has been internationally used as the main blend ingredient in mainstream coffee brands, being priced at a discount relative to a number of other coffee producing countries. Also, the efficiency with which Brazilian coffee producers have been historically able to supply the international commodity coffee market has strengthen the view of the country as a commodity supplier, not a specialty coffee producer (Coe in Brazil and Honduras – Impact Assessment, 2015).
Nonetheless, Brazilian specialty coffee production has been growing steadily. It is estimated that Brazil has been increasing its production at an estimated average rate of 15% per year since 1999 and it is estimated that the specialty coffee production accounts for about 9% of national production (around 4 million bags) (Coe in Brazil and Honduras – Impact Assessment, 2015). Furthermore, efforts to promote Brazilian specialty coffee are evident, especially by initiatives promoted by the Brazilian Specialty Coffee Association (BSCA).
Therefore, considering the central position of Brazil in the international coffee market and the evidences of the increasing interest of coffee producers to enter the specialty coffee market, incentivized by the potential of achievement of a price premium through differentiation, we shall explore the different Brazilian coffee producing regions next, in order to understand their configuration and look for evidences of regional reputation. Subsubsection 2.3.1 will deal with the geographical configuration of coffee producing regions in Brazil while subsubsection 2.3.2 will explore in details the legally protected coffee producing regions in the country.
2.3.1 Coffee producing regions in Brazil
Brazil is a country of continental dimensions. Coffee is produced in a number of different Brazilian states, where geographic and climatic conditions, aside from cultural practices, vary greatly. As pointed out previously, this diversity in conditions can potentially yield an array of differentiated coffees, which is interesting for the specialty coffee sector. The Brazilian Specialty Coffee Association (BSCA) defined the current geographic definition of coffee producing regions in Brazil in 2014, having presented such configuration on the Annual Exposition of the Specialty Coffee Association of America in April 2014. The updated map entitled Brazilian Coffee Origins (2017) was obtained, and it is the one present in this paper. The BSCA does not define the coffee producing regions, as they obtain the municipal demarcation already established from a number of different sources, and formulate a unified map, in order to facilitate the understanding of the configuration of coffee production in the country (Marina Figueiredo, BSCA personal messaging 2017).
The map presents 25 coffee producing regions, located in ten Brazilian states and the Federal District, being the regions categorized by coffee tree species. Twenty of such regions, located in nine states and in the Federal District, produce Arabica coffee exclusively and two regions located in the Espírito Santo and Ceará states produce Arabica and Robusta, being such, therefore, the relevant regions for this research due to the fact that Arabica species yield finer cups in comparison with the Robusta species, being specialty coffees generally yielded from Arabica varieties. There are five legally protected regions that have their Geographic Indications (GIs) certificate, being all of them exclusively producers of Arabica coffee. The map Brazilian Coffee Origins (2017) can be consulted in the Appendix (Figure 2, Appendix).
The state of Minas Gerais is the leader in Arabica coffee production, having produced 30,427.9 thousand bags from a national total of 43,382.2 thousand bags of Arabica coffee produced in 2016 (CONAB, 2017). In the state of Minas Gerais are located six coffee producing regions, namely Sul de Minas, Chapada de Minas, Matas de Minas, Cerrados de Minas, Mantiqueira de Minas (GI) and Cerrado Mineiro (GI) (BSCA, 2017). The state of São Paulo accounts for the second largest production of Arabica, being responsible for the production of 6,031 thousand bags or Arabica coffee in the crop year of 2016 (CONAB, 2017). The six coffee producing regions in the state of São Paulo are the Região do Pinhal region (GI), the Mogiana region, the Alta Mogiana region (GI), the Média Mogiana region, the Marilia e Garça region and the Ourinhos e Avaré region (BSCA, 2017). Still in the