a. Trends in the European wine industry

Wine is a cornerstone of European culture and commerce. Despite declining exports due to increased competition from the New World, Europe still produces around 68% of the world’s wine exports (Balogh & Jambor, Reference Balogh and Jámbor2017). Italy, France, and Spain lead global production, and the EU wine industry generates nearly 130 billion euros in annually (The Brussels Times, 2024, March 27).

In recent decades, alcohol consumption has declined in many regions. While European wine consumption has fallen, new markets like the U.S. and China have driven global growth (Castellini and Samoggia, Reference Castellini and Samoggia2018). Although the EU has seen a modest recent increase, consumption remains far below peak levels. In Italy, per capita wine consumption dropped from 93.5 L in 1977 to 38.01 L in 2014, and rose modestly to 42.1 L in 2023 (Barker, Reference Barker2024, April 25; Sellers and Alampi-Sottini, Reference Sellers and Alampi-Sottini2016).

The decline in European wine consumption has spurred innovation among wineries. Vergamini et al. (Reference Vergamini, Bartolini, Prosperi and Brunori2019) note that combining innovation with an understanding of societal and cultural factors can enhance firm performance. Obermayer et al. (Reference Obermayer, Kővári, Leinonen, Bak and Valeri2022) highlight production methods and marketing as key areas for innovation, with many small and medium enterprises leveraging social media and wine tourism (e.g., trails and festivals) to differentiate themselves. In smaller regions, wine tourism allows consumers to connect with the terroir, and geographic indication designations help these areas stand out in a competitive market (Vergamini et al., Reference Vergamini, Bartolini, Prosperi and Brunori2019). Given wine’s importance in European history, culture, and economy, wine tourism can be a valuable driver of economic growth (Vicente et al., Reference Vicente, Barroso and Jiménez2021).

b. Wine and tourism demand

Tourism demand research has expanded significantly over the past two decades. Cooper et al. (Reference Cooper, Fletcher, Gilbert and Wanhill1993) define tourism demand as travel to access tourist facilities and services. Tourism demand varies across regions and time periods. It has become a key element in many nations’ economic growth strategies. Research on this topic has grown substantially, Zhang et al. (Reference Zhang, Wang, Sun and Wei2020) note that fewer than 10 articles published in the early 2000s, rising to 40 by 2018.

Rising incomes have fueled increased spending on travel, transforming tourism into a discretionary, luxury experience. Turner and Witt (Reference Turner and Witt2001) emphasize that Gross Domestic Product (GDP) typically reflects the levels of economic development, which strongly affect demand across tourism segments. Supporting this, Peng et al. (Reference Peng, Song, Crouch and Witt2015) note that international tourism is often income elastic, particularly among visitors from developed regions, suggesting that foreign tourism functions as a luxury product.

Wine tourism, a subcategory of gastronomy tourism (UNWTO, 2016), is influenced by similar economic dynamics. Hall and Macionis (Reference Hall, Macionis, Butler, Hall and Jenkins1998) define wine tourism as visits to vineyards, wineries, or wine festivals where tourists can enjoy wine tasting and experience the attributes of a wine region. Wine tourism helps regions build brand awareness, educate consumers, gather marketing insights, and promote local wine styles (Lee et al., Reference Lee, Kladou, Usakli and Shi2022). Its benefits are felt not only at wineries but also in related sectors like lodging and restaurants (Storchmann, Reference Storchmann2010) and positively influence the competitiveness of wine-producing regions (Sun and Drakeman, Reference Sun and Drakeman2022).

Wine tourists are typically more affluent and spend more on premium experiences, such as winery tours, tastings, unique culinary experiences, and extended stays compared to general tourists (Gómez-Carmona et al., Reference Gómez-Carmona, Paramio, Cruces-Montes, Marín-Dueñas, Montero and Romero-Moreno2023; Mitchell and Hall, Reference Mitchell and Hall2006). These experiences support economic development by creating jobs in rural areas and attracting investment in hospitality infrastructure (Quadri-Felitti and Fiore, Reference Quadri-Felitti and Fiore2012; Vicente et al., Reference Vicente, Barroso and Jiménez2021). Beyond its economic impact, wine tourism helps preserve cultural heritage through storytelling and education about regional traditions (Frost et al., Reference Frost, Frost, Strickland and Maguire2020; Hall et al., Reference Hall, Johnson, Mitchell, Hall, Sharples, Cambourne and Macionis2009).

Despite its growing importance, research connecting tourism and wine consumption, especially with panel data, remains limited. Vicente et al. (Reference Vicente, Barroso and Jiménez2021) examine Spain and show that wine tourism positively contributed to local economic growth, but broader multicountry longitudinal analyses are scant. Given its potential as a driver of regional development and revenue, there is a need for a deeper understanding of wine tourism’s effects on local economies and consumption behavior.

c. Veblen goods

Veblen goods are high-priced items where demand may rise with price due to their symbolic value. Consumers are motivated by either invidious comparison (to stand out from lower groups) or pecuniary emulation (to fit in with higher groups), using luxury products to signal prestige (Bagwell and Bernheim, Reference Bagwell and Bernheim1996; O'Regan et al., Reference O'Regan, Choe and Yap2019). Veblen goods illustrate the complexity of consumer purchasing behavior. Their appeal lies in the price itself, which signals desirability, a characteristic that is common in luxury categories like fashion, art, and wine (Wright et al., Reference Wright, Yoon, Morrison and Šegota2023).

The Veblen effect is evident in the wine industry, where wine consumers are willing to increase consumption even as prices rise, treating wine as a luxury good rather than a commodity (Rousselière et al., Reference Rousselière, Petit, Coisnon, Musson and Rousselière2022). This dynamic supports wine tourism, as demand increases for high-end winery experiences (Heine et al., Reference Heine, Phan and Atwal2016). To address diverse market segments, producers offer wines ranging from affordable to ultra-premium tiers. Some producers achieve luxury status through small production runs, exclusive vineyard sourcing, renowned winemakers, and high ratings from critics, allowing them to command higher prices and prestige (Horowitz, Reference Horowitz2012). These luxury or “Veblen” wines enhance regional brand identity and attract tourists eager for premium experiences.

d. Hypothesis development

While previous research examines the effect of tourism on economic development using cross-sectional data (Pablo-Romero and Molina, Reference Pablo-Romero and Molina2013), studies specifically investigating the impact of tourism demand on wine consumption remain limited. Tourism has long been a source of local income, driving development in many destinations. The literature frequently highlights a positive relationship between tourism spending and regional economic growth (e.g., Contini et al., Reference Contini, Scarpellini and Polidori2009; Rachão et al., Reference Rachão, Breda, Fernandes and Joukes2019), with many viewing tourism as a remedy for economic stagnation and an alternative to traditional industries (Meyer and Meyer, Reference Meyer and Meyer2015). In rural areas, tourism can boost agricultural tourism, including the consumption of local products like wine. Saayman et al. (Reference Saayman, Saayman and Rhodes2001) note that the economic impact of tourism varies by the origin of tourists. Studies by Lee (Reference Lee2021) and Goh et al. (Reference Goh, Li and Zhang2015) show that domestic tourism-led growth strategies can positively affect regional economies. Since domestic tourists are likely to engage directly with local wine products through tastings, tours, and purchases, this study proposes a positive link between domestic tourism and wine consumption.

Although a linear relationship is expected at lower levels of tourism, the effect may diminish as spending increases. For instance, when a region’s wineries and tourism infrastructure operate near their limits, more spending may not increase wine-related engagement. This logic aligns with the concept of diminishing marginal returns, where tourism’s benefits taper off once spending exceeds a certain threshold (Zhang and Cheng, Reference Zhang and Cheng2019). Similarly, smaller local contexts often experience a tapering effect when infrastructure is reaches its limit (Chiu and Yeh, Reference Chiu and Yeh2017). In this setting, a nonlinear relationship may appear, especially if market saturation or overloaded infrastructure limits further gains.

International tourism is crucial for regional economies, particularly in high-value segments like luxury wine tourism. Foreign tourists seeking unique luxury experiences, such as vineyard tours, wine tastings, and wine pairing dinners, often make significant contributions to local economic development (Akinboade and Braimoh, Reference Akinboade and Braimoh2010). These visitors tend to spend more on local goods, including luxury items like wine (Hung et al., Reference Hung, Ren and Qiu2021). Given wine’s dual role as a local product and a luxury good, this study proposes that foreign tourism spending will positively impact wine consumption, as foreign tourists tend to favor high-end offerings.

Similar to domestic tourism, the relationship between foreign tourism spending and wine consumption may be nonlinear. At lower spending levels, tourists might prioritize other activities, like sightseeing or general attractions, over wine consumption. However, once spending reaches a certain threshold, tourists are more likely to engage in experiences like winery tours, tastings, and pairing dinners. Researchers have documented this threshold-driven shift, with more intense consumption emerging after higher financial and experiential investment (Chiang et al., Reference Chiang, Sung and Lei2017). Ridderstaat et al. (Reference Ridderstaat, Croes and Nijkamp2014) note that the economic impact of tourism may be nonlinear, with strong effects occurring when tourism spending reaches higher levels. This change in tourist behavior stems from greater investment in local development and high tourism expenditure.

Wine price is a key factor in consumers’ decision to purchase. According to the law of demand, higher prices generally reduce demand for a product. This inverse relationship between price and consumption is common with commodities, including mid-range and affordable wines (Jiang and Livingston, Reference Jiang and Livingston2015). However, in the luxury segment, wine can behave as a Veblen good, where high prices signal status and prestige, potentially increasing demand (Sjostrom et al., Reference Sjostrom, Corsi and Lockshin2016; Wright et al., Reference Wright, Yoon, Morrison and Šegota2023). While high prices typically deter consumers, luxury wines may attract buyers who view price as a marker of quality. Nevertheless, since mid-range and affordable wines make up most of the market, the traditional demand relationship likely holds, with higher prices leading to lower consumption.

As with the previous relationships, wine price and consumption may be nonlinear. While higher prices usually translate to reduced demand, premium wines may attract consumers who perceive price as a signal for prestige or exclusivity. This counterintuitive response reflects Veblen goods, where higher prices translate to increased rather than decreased appeal (Bagwell and Bernheim, Reference Bagwell and Bernheim1996). Similar studies on conspicuous consumption show that demand can increase when the price is publicly displayed and associated with a certain identity (Heffetz, Reference Heffetz2011). Therefore, the wine price index may follow a nonlinear pattern, dependent on consumers’ perception of wine as a commodity or a status symbol.

a. Data collection and variables used

The data for this study are sourced from various sources for 19Footnote ² European countries, as shown in Table 1 and detailed in Appendix A. Specifically, the wine consumption data are drawn from the University of Adelaide’s Wine Economics Research Centre, which aims to produce and promote high-quality research in the field of wine economics (Anderson and Pinilla, Reference Anderson and Pinilla2024). In this study, Belgium and Luxembourg are grouped together, to match the structure in the original dataset. Data on domestic tourism spending originate from the World Bank (World Bank, n.d.) and Eurostat (Eurostat, n.d-b.). International tourism spending data come from the World Bank (World Bank, n.d.) and Organization for Economic Co-operation and Development (OECD) (OECD, 2024, October 17). Eurostat is the source for the Harmonized Wine Price Index (Eurostat, n.d-a.). The full list of countries appears in Appendix A.

Table 1.

Variable definitions and sources

The authors log-transformed each variable to improve stability and interpretability, accounting for potential nonlinear relationships. The dataset includes 12Footnote ³ Eurozone countries and 7 non-Eurozone countries (see Appendix). Panel data are appropriate because they allow for tracking changes over time across multiple countries, offering greater precision and nuance than cross-sectional data (Hsiao, Reference Hsiao1985, Reference Hsiao2007).

b. The model

This study develops a model to examine the effects of domestic tourism spending, foreign tourism spending, and price on a country’s wine consumption. The model includes both linear and nonlinear terms to account for elasticity effects and test for the presence of the Veblen effect in wine tourism. These relationships are modeled with the following equation:

(1)\begin{equation}{\text{w}}{{\text{c}}_{it}}{\text{ = }}a{\text{ + }}{b_x}{X_{it}}{\text{ + }}{c_x}{X^{\text{2}}}_{it}{\text{ + }}{e_{it}}\end{equation}

where꞉

wc = Wine consumption

i = Cross-sectional component of the panel (i.e., country)

t = The time component of the panel

X = A vector of independent variables

X ² = A squared vector of variables identified by the U-test as nonlinear

a = Intercept

b_x, c_x = Coefficients

e = Residual error terms

The inclusion of squared terms allows the model to test for possible nonlinear relationships, similar to Veblen goods, where high prices do not reduce demand. To validate these relationships, this study applies the U-test for elasticity effects (Lind and Mehlum, Reference Lind and Mehlum2010). The U-test assesses whether spending increases have a consistent or varied effect on wine consumption, helping to identify turning points in demand. The U-test examines U-shaped and inverted U-shaped relationships by assessing values on both the low and high ends of the curve. A U-shape indicates a convex (nonlinear) relationship, while an inverted U-shape suggests a concave relationship. The framework for interpreting elasticity effects, based on Renshaw (Reference Renshaw2009), is shown in Table 2. In addition to the elasticity test, a threshold regression model is implemented to examine whether structural breaks arise in the relationship between foreign tourism spending and wine consumption. Further explanation is provided in the following section.

Table 2.

Elasticity effects

where

b_x is the coefficient of the variable

c_x is the squared version of the variable’s coefficient

c. Methods and procedures

The first step in preparing the panel data is to check for cross-sectional dependence, which identifies correlations across countries. Dependent cross-sections can lead to inaccurate inferences and spurious estimates (Pesaran, Reference Pesaran2015), often due to significant events affecting multiple countries. To address this, the analysis applies a Pesaran CD-test (Pesaran, Reference Pesaran2004, August).

To evaluate stationarity, this study applies three panel unit root tests: Fisher-type -Augmented Dickey–Fuller (ADF) and Phillips-Perron (PP) (Choi, Reference Choi2001; Maddala and Wu, Reference Maddala and Wu1999) and IPS (Im et al., Reference Im, Pesaran and Shin2003), at both the level and first difference. Stationarity enables long-term inferences and reduces the risk of spurious results. These tests help determine the integration of the variables. Additionally, the study applies Kao (McCoskey and Kao, Reference McCoskey and Kao1998) and Pedroni (Reference Pedroni1999) tests to assess long-term cointegration, determining whether a long-term equilibrium exists among the variables.

Model specification tests identify the appropriate estimation method. This involved the Hausman test, which determines whether fixed or random effects models are most appropriate. An F-test confirms the presence of country-level fixed effects. To address potential autocorrelation and heteroscedasticity, this study applies a Wooldridge test (Wooldridge, Reference Wooldridge2002). In panel data, autocorrelation occurs when error terms are correlated over time, while heteroscedasticity refers to nonconstant variance across observations (Bollerslev, Reference Bollerslev1986). Given the potential for these issues, the analysis applies a fixed effects regression with Driscoll–Kraay standard errors to generate robust estimates (Driscoll and Kraay, Reference Driscoll and Kraay1998).

Finally, the analysis conducts a threshold regression to examine nonlinear relationships (Hansen, Reference Hansen1999). A grid search method identifies the optimal threshold value. Afterward, the model is estimated with interaction terms to examine effects above and below the threshold.

a. Cross-sectional dependence

Cross-sectional dependence is a common issue in panel regression analysis, as it indicates interference between units (in this case, countries). As shown in Table 3, all variables except wine consumption exhibit cross-sectional dependence, indicating significant correlations across countries for domestic tourism spending, foreign tourism spending, and the harmonized wine price index. This suggests that European countries share economic and cultural factors influencing tourism and pricing behaviors. To address this, the study uses panel analysis with robust standard errors to improve the reliability and accuracy of estimates.

Table 3.

Cross-sectional dependence tests

Note: Statistical significance is indicated by asterisks:

* at the 10% level, ** at the 5% level, and *** at the 1% level.

b. Unit root tests

Unit root tests, including ADF, PP, and Im–Pesaran–Shin, assess stationarity across all variables (Table 4). These tests are chosen for their ability to handle unbalanced panels. To account for between-subject variability, the variables are adjusted to the cross-sectional mean. Results indicate that each variable is stationary at first difference, with domestic tourism spending showing stationarity at both level and first difference. Stationarity is essential for ensuring long-term reliability of interpretations, suggesting that the variables are well suited for further analysis.

Table 4.

Unit root tests

Note: Statistical significance is indicated by asterisks:

* at the 10% level, ** at the 5% level, and *** at the 1% level.

c. Cointegration tests

The current study uses the Kao (using ADF and Unadjusted Dickey–Fuller tests) and Pedroni (using Modified PP and ADF tests) methods to assess potential long-run relationships among the variables. Table 5 presents the results, including overall and between-variable assessments. The significance of the tests supports the presence of cointegration, indicating stable long-term relationships between tourism spending, wine consumption, and price changes. These results suggest that the analysis can proceed using the level form of the variables rather than their first differences.

Table 5.

Cointegration tests

Note: Statistical significance is indicated by asterisks:

* at the 10% level, ** at the 5% level, and *** at the 1% level.

d. Elasticity effects

Elasticity effects are evaluated to identify potential nonlinear relationships typical of Veblen goods. Hausman specification tests assess whether squared terms improved the model. The results suggest that foreign tourism spending exhibits a nonlinear relationship, while domestic tourism spending and the harmonized price index do not. Specifically, the insignificant results for domestic tourism spending (t = 0.61; p > .10) and wine price (t = 1.21; p > .10) indicate no elasticity effects for these variables. However, the significant result for foreign tourism spending (t = 6.83; p < .10) indicates a nonlinear relationship in which foreign tourism spending initially decreases wine consumption but later increases it at higher levels. Thus, the regression model includes foreign tourism spending as a squared term. Hypotheses H_1b and H_3b are not supported.

e. Linear panel regression and robust estimation

The panel regression analysis (Table 6) sheds light on the relationships between domestic tourism spending, foreign tourism spending, the harmonized wine price index, and wine consumption. The insignificant positive coefficient for domestic tourism spending (β = 0.046; p > .10) does not support H_1a, indicating that domestic tourists do not influence wine consumption in their home markets. This finding suggests that wine may already be integrated into domestic culture, making tourism-related consumption less pronounced. As mentioned previously, no nonlinear elasticity effects appear for this relationship (t = .052; p > .10), indicating a linear impact of domestic tourism spending on wine consumption. Thus, H_1b is not supported.

Table 6.

Fixed effects panel regression of wine consumption with Driscoll–Kraay standard errors

Note: Statistical significance is indicated by asterisks:

* at the 10% level, ** at the 5% level, and *** at the 1% level.

Foreign tourism spending shows a more complex relationship with wine consumption. The significant negative linear coefficient (β = –0.209; p < .10) and significant positive convex elasticity effect (β = 0.044; p < .01) suggest that foreign tourist spending initially reduces wine consumption, possibly due to supply constraints or lack of familiarity with local wines. However, beyond a certain threshold, wine consumption begins to rise, likely due to increased engagement with local products, indicating a shift toward wine tourism. These findings support H_2a and H_2b.

The harmonized wine price index shows a significant negative relationship with wine consumption (β = –0.328; p < .01), suggesting some level of price sensitivity in overall wine consumption. Thus, hypothesis H_3a is supported.

f. Nonlinear threshold regression

To determine the appropriate model, the authors conduct both an F-test and Hausman specification test. The F-test (F = 8.88, p = 0.0031) reveals significant country-level effects. In addition, the Hausman test (χ ² (3) = 20.02, p = 0.0002) confirms that the fixed effects model is preferred over random effects. These results support the use of the fixed effects specification throughout the analysis.

To account for the cross-sectional dependence in the initial Cross sectional dependence tests, Driscoll–Kraay standard errors, which are robust to cross-sectional dependence, autocorrelation, and heteroscedasticity, are incorporated into the analyses. The model confirms the presence of nonlinear relationships between foreign tourism spending and wine consumption. Specifically, the linear effect of foreign tourism spending is negative (β = − 0.210, p = 0.071), while the squared version is positive and significant (β = 0.044, p = 0.007), supporting a U-shaped relationship between foreign tourism spending and wine consumption.

Taking a more detailed look at the nonlinearity, this study conducts a threshold regression that designates foreign tourism spending’s squared term as the threshold variable. A grid search reveals that the optimum threshold is 5.55, which minimized residual sum of squares. The results suggest that foreign tourism spending has a significant negative effect below the threshold (β = −0.114, p = 0.021), while the same effect becomes weaker and insignificant above the threshold (β = −0.050, p = 0.242). This result (Figure 1) illustrates the dynamic that initial increases in foreign tourism may displace local wine consumption behaviors. However, beyond a critical point, consumption increases, which may be due to further development of the destination and its infrastructure. In all, these findings suggest that foreign tourism spending has a nonlinear effect on wine consumption, which is context-dependent.

Figure 1.

Wine consumption (in kiloliters) as a function of foreign tourism spending (in billions USD). The dashed vertical line indicates the estimated threshold.