Why types of operations, trade associations, and production trends matter in the geographic branding of an emerging industry

ABSTRACT

Geographic branding has become an important strategy for economic actors to differentiate their products. While studies have examined geographic branding in well-established industries, less is known about the factors that lead to the adoption of geographic branding innovations in emerging industries. Results from a 2011 mail survey of Pennsylvania winemakers (N = 71, response rate: 50%) indicate that winery operation type (grape grower or wine-only), growth in wine and/or grape production, and trade association membership influence the extent to which geographic branding is adopted and in what forms (regional or state-based) it is used. Winemakers also identified challenges that they believe limit the economic growth of the industry. Findings shed light on the opportunities and limitations of geographical branding and its impacts on regional development.

KEYWORDS:

• Terroir
• branding
• geographic designation
• AVA
• AOC
• territorial branding

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 The Pennsylvania Winery Association designated regions are: Philadelphia Countryside, Lower Susquehanna, Pittsburgh Countryside, Upper Susquehanna, Lehigh Valley & Berks County, the Groundhog region, and Lake Erie. A map and trail descriptions are available at www.pennsylvaniawine.com.

2 In this case, the finite population correction can be calculated as: fpc = (N-n)/(N-1) = (142-71)/(142-1) = .71. When the sample is a large fraction of the population and sampling is done without replacement (as was the case here), applying the finite population correction reduces the sampling variance of statistics (Groves, et al., 2004, pp. 99–100).

3 The use of complete cases reduced our sample size to 59, which indicates overlap in missingness due to non-response on owner and winery age (n = 3), the importance of traditional and PA-grape only marketing (n = 5), and those who chose not to provide any response when asked to define ‘local’ (n = 9). Mean imputation was used to maintain these cases in the analysis and, while the magnitude of the odds ratios changed somewhat, the direction and significance levels did not.