Complementary or Congruent? The Effect of Hosting Tesla Charging Stations on Hotels’ Revenue

Abstract

Electric vehicles (EVs) are a promising sustainable technology for transitioning to a low carbon economy. However, there is a lack of research on how tourist destinations such as hotels and attractions could benefit from EVs. Using evidence from monthly revenue data of 2,774 hotels in Texas of United States (US) between 2015 and 2018, this paper quantifies the economic benefits of hotels hosting Tesla’s charging facilities and finds that nearby attractions amplify the benefits. Further, the study investigates the heterogeneity of the benefits across different hotel segments and their dynamics. The findings reveal that upscale hotels benefit more than luxury as well as mid-price and cheaper hotels from hosting Tesla charging facilities. After Tesla introduced the Model 3, these benefits increased for upscale hotels but decreased for luxury hotels. These findings have important implications for the hospitality and tourism industries to better adapt to the emerging EV transition.

Keywords

electric vehicles charging infrastructure hotels brand alliance tourist attraction Tesla

Introduction

Electric vehicles (EVs) are one of the major sustainable technologies that are helpful for the transition to a low carbon economy. As EVs are penetrating into the mainstream, tourism, and hospitality industries have been advocating the adoption and use of EVs in business practice. For example, Orlando in Central Florida of the United States (US) has launched Drive Electric Orlando (DEO) and Green Destination Orlando (GDO) initiatives to promote destination sustainability (Fjelstul 2014). New Jersey Board of Public Utilities (2021) launched the EV tourism program in September 2021 to encourage setting up EV chargers at tourism destinations such as unique attractions and hotels. More recently, the car rental company Hertz is going to purchase 100,000 Tesla cars for its customers to rent (CNN Business 2021), which will also have a huge impact on drive tourism (Duval 2020).

Despite these market advancements, existing tourism research on EVs is limited to the perception of EVs being inadequate for tourism travel (Fitt 2021), largely due to consumers’ driving range anxiety (Dong, Liu, and Lin 2014; Lim, Mak, and Rong 2015). In response to this adoption barrier, destinations can invest in constructing charging facilities in their sites to meet consumer demand. However, if charging facilities are not an influential factor in consumers’ choice of destinations (Gerdt, Wagner, and Schewe 2019; Wang et al. 2020), the investment will not be effective in meeting demand or increasing revenue. Yet there is a lack of research in the tourism literature on how destinations can better adapt to the emerging trend toward EV transition (Fitt 2021; Hopkins 2020).

According to Newmeyer et al. (2018), when a property hosts an EV charging facility, it forms a co-location brand alliance with the EV charging brand. The prior literature identifies the importance of consumer perception of brand fit when firms develop a brand alliance or adopt a co-branding strategy (e.g. Decker and Baade 2016; van der Lans, Van Den Bergh, and Dieleman 2014). However, there are opposite views regarding whether the focal brand should choose a more similar or a more dissimilar brand as its co-branding partner. Some researchers find that congruency is positively associated with the evaluation of band alliance (Decker and Baade 2016; Simonin and Ruth 1998), while others advocate the complementary effect between partnering brands (Koschmann and Bowman 2018; Park, Jun, and Shocker 1996).

Motived by these opposite views, we set out to answer the central research question of this study—do destinations such as hotels benefit from hosting EV charging facilities? To answer this question, this study uses empirical evidence from Tesla’s destination charging (DC) stations in hotels, and the monthly revenue data of hotels in Texas, US. Using the Tesla DC station data for 116 hotels and monthly revenue data for 2,774 hotels in Texas between 2015 and 2018, we quantify the extent to which the hotels benefited from hosting Tesla DC stations in their properties.

There are two major challenges to conducting this econometric analysis. The first is ensuring a rigorous design for causality inference that should mitigate confounding issues such as self-selection issues and extraneous factors that may affect hotels’ installing decisions. Self-selection issues in this case may occur because the choice of hotels may not be randomly distributed but driven by hotels’ characteristics such as hotel size, revenue, and location. Extraneous factors from the firm, market, and the environment may confound the causal effect. The second challenge is the heterogeneity treatment effect that may arise from different hotels. Given that the hotel market has significant segmentation in relation to brand and customer, the causal effect of hosting the EV charging facility on hotel revenue should be considered heterogeneous across different segments of hotels.

Thus, this study employs a difference in differences (DID) empirical strategy after matching the hotels that have installed a Tesla DC station, that is, hosting hotels (treatment group) against hotels that have not installed a Tesla DC station, that is, non-hosting hotels (control group) based on propensity score matching (PSM) (Rosenbaum and Rubin 1983). The DID model reveals that the monthly revenue in the hosting hotels increased by 25.90% after installing a Tesla DC station in their properties, while the hotels in the control group suffered a 4.70% decrease in their monthly revenue. Given that hotels and Tesla are in completely different industry sectors, hotels’ decision to host Tesla DC stations brings together complementary functional (utilitarian) attributes in brand alliance (Newmeyer, Venkatesh, and Chatterjee 2014), and thus leads to positive financial consequences for hotels.

We also investigate how the popular tourist destinations in the same city amplify or attenuate hotels’ benefits from hosting Tesla DCs. By collecting the location information of the top 150 attractions in Texas, our analysis finds that the hotels located in the city with more top attractions are likely to receive higher revenue after hosting Tesla DCs, which suggests that tourists to these popular destinations are attracted to stay in hotels with Tesla DC and thus contribute to the increase of hotels’ revenue. We also conduct additional comparisons against the moderating effects using Texas’s top 50 and top 100 attractions. The comparisons show a “superstar” effect that the DC-hosting hotels can attract significantly higher benefits if they are located in the cities with more top 50 attractions than those in the cities with top 100 or top 150 attractive destinations.

Furthermore, we examine the heterogeneous effect among different segments of hotels by testing the moderating role of hotels’ price segments on the economic benefits of hosting Tesla DC stations. Our analysis reveals an inverted U-shaped effect on the revenue of hotels hosting Tesla DC stations in three price segments. That is, upscale hotels benefit most from installing Tesla DC stations, followed by luxury hotels and then mid-price and cheaper hotels. Further, we find that when Tesla extended its product line into the mass market, the revenue of hosting hotels decreased. However, the effect was heterogeneous across different price segments of hotels. Our three-way interaction analysis revealed that after the Tesla Model 3 was made available in the market, upscale hosting hotels had significantly higher revenue than mid-price and cheaper hotels. However, luxury hosting hotels had significantly lower revenue than mid-price and cheaper hotels after the Tesla Model 3 was released to the market. These findings reveal important implications for hotels’ alliance strategies when adapting to emerging EV transition.

The remainder of this paper is organized as follows. Section 2 provides a brief overview of the Tesla Destination Charging Program. Section 3 presents the empirical analysis, and Section 4 explains the empirical methods. Section 5 presents the empirical analysis results and robustness checks. Section 6 discusses the theoretical contributions and practical implications of this study, as well as future research directions.

An Overview of the Tesla Destination Charging Program

As a leading EV maker, Tesla is a unique company that not only designs and produces its EV models (e.g. Model S, Model X, and Model 3), but also develops and runs its own charging network to offer exclusive charging services to Tesla vehicle owners based on “a unique charging port and connector that works for all their charging options” (Alternative Fuels Data Center of the US Department of Energy 2021). More specifically, Tesla has established two types of charging service networks, one provides superchargers and the other provides DC-based slow charging (Lambert 2020). The Tesla Supercharger is a direct-current-based fast-charging service that can recharge up to 200 mi of driving range in 15 minutes (Tesla 2021). Tesla owns and runs its Supercharger network through strategically located Supercharger stations, primarily located along well-traveled highways in the US and other countries (Tesla 2016).

In comparison, the Tesla DC stations offer a slow charging service using 208 V of alternating current charging equipment that can add up to 44 mi of driving range per hour. More importantly, Tesla DC service is typically provided through partnerships with hospitality providers such as hotels, restaurants, and resorts. When Tesla expanded its DC program, it offered hosting properties free or discounted DC chargers and often paid for the installation of these stations, while hosting properties participating in the DC program had to cover the electricity costs when their customers chose to charge their Tesla cars in these properties (Lambert 2017a, 2020).

According to Alternative Fuels Data Center of the US Department of Energy (2018), there were 2,611 stations in Tesla’s DC network in the US in September 2018, among which approximately 70.2% of DC stations were hosted by hotels, followed by 9.3% in parking lots, 5.7% in restaurants, and so on. Thus, our study focuses on the effect of hosting Tesla DC stations on hotel revenue.

Major businesses claim that it demonstrates their social goodwill and responsibility when hosting this charging infrastructure. For example, Radisson Hotels joined the Tesla DC network as part of the hotel chain’s Think Planet Initiative, which aims to “reduce the carbon footprint of our hotels and improve our contribution to a low carbon society” (Radisson Hotels 2021). Starting in October 2015, Hilton partnered with Tesla and General Electric to install EV charging stations in 100 hotels in the US (Lambert 2017b). By November 2018, 350 Hilton hotels worldwide, mostly in the US, had joined the Tesla DC program (TopHotelNews 2018). Thus, it seems that such businesses believe it will benefit them to decide to join the charging network and form a brand alliance (Newmeyer, Venkatesh, and Chatterjee 2014) with a charging network operator (e.g. Tesla).

Literature Review and Hypotheses Development

There are three streams of research relevant to this topic. The first research stream addresses consumer EV adoption. As EVs are becoming one of the most promising sustainable technologies for achieving low carbon and green transport targets (Huang and Qian 2021), customers’ adoption of EVs continues to be influenced by technical, psychological, and environmental factors (Rezvani, Jansson, and Bodin 2015). It is well recognized that EVs have a shorter driving range and slower recharging/refueling speed than conventional gasoline cars (Kempton 2016; U.S. Department of Energy 2016). Further, the actual driving range of EVs is highly uncertain, depending on many environmental factors such as road grade, traffic congestion, driving pattern, and weather conditions (Hao et al. 2020; Wager, Whale, and Braunl 2016). Thus, installing public charging stations is considered the most important solution for extending the driving range of EVs (Qian, Grisolía, and Soopramanien 2019). Studies have revealed that customers do not simply treat the EV as a product but also pay attention to driving range and charging service factors, which jointly form a product-service system for EVs (Cherubini, Iasevoli, and Michelini 2015).

Second, our research is also related to the general literature of sustainable tourism (Buckley 2012; Hall 2019; Ruhanen, Moyle, and Moyle 2019). As Welford and Ytterhus (2004) point out, sustainable tourism needs to enhance the management of destinations through the networking and collaboration between different service providers in interrelated sectors such as hotels and transport. More recently, as sustainable tourism has been focusing on the more relevant, big-picture, and hard-hitting topics, such as climate change (Ruhanen, Moyle, and Moyle 2019), the use of EVs is considered the most impactful technological advancement on the interface between tourism and transport (Duval 2020). It can also help the destinations to develop a positive brand image and demonstrate corporate social responsibility, such as the Drive Electric Orlando (DEO) and Green Destination Orlando (GDO) initiatives (Fjelstul 2014). In the tourism literature, however, there is limited research to quantify the impact of EVs on tourism, largely due to the questionable assumptions that EVs are inadequate for tourism travel and thus they are unlikely to result in substantial changes to tourism (Fitt 2021).

The third relevant research stream examines brand fit in the tourism and marketing field. Brand perception is recognized as a critical factor affecting customers’ choice of a focal service, such as hotel selection (Casidy, Wymer, and O'Cass 2018). Relevantly, the brand fit is the value judgment made by the customer on the degree of alignment of two associated brands, through which customers develop quality expectations (Norman 2016). Specifically, Casidy, Wymer, and O'Cass (2018) argue that consumers’ perception of their relationship with hotel brands influences brand performance through factors such as share of wallet and revisit intention. Yan and Shen (2021) show that brand quality and satisfaction significantly affect the brand fit in tourism.

Further, the literature identifies the importance of consumer perception of brand fit when firms develop a brand alliance or adopt a co-branding strategy (e.g. Decker and Baade 2016; van der Lans, Van Den Bergh, and Dieleman 2014). In a co-branding process, partner selection is a key decision for the focal brand, for example, properties hosting charging stations (Newmeyer, Venkatesh, and Chatterjee 2014; van der Lans, Van Den Bergh, and Dieleman 2014). However, the existing literature has not reached any consistent conclusion about whether the focal brand should choose a more similar or a more dissimilar brand as its co-branding partner (van der Lans, Van Den Bergh, and Dieleman 2014). One consideration is that organizational dissimilarity in firm size, industry scope, and country-of-origin image between co-branding partners can negatively affect brand fit perception (Decker and Baade 2016), and brand image consistency between the two partner brands is positively associated with the evaluation of brand alliance (Simonin and Ruth 1998). However, researchers have also argued that a complementary effect can be better achieved if two partner brands have moderate dissimilarity (Denizci Guillet and Tasci 2010; Park, Jun, and Shocker 1996), and congruent associations (e.g. the perceived functional associations of both brands) can weaken the positive effect that each brand contributes to the ingredient brand alliance performance (Koschmann and Bowman 2018).

Inspired by the prior literature on brand alliance, the debate on the congruent versus complementary association between brands, and the collaboration between different service providers in tourism, we develop the following research hypotheses as summarized in Figure 1.

Figure 1.

The number of Tesla DC Stations in Texas from 2015 to 2018.

Main Effect on Financial Performance

Earlier literature evaluating brand alliances has identified the importance of product category fit (Baumgarth 2004; Park, Milberg, and Lawson 1991; Simonin and Ruth 1998). Specifically, when consumers evaluate product category fit in a brand alliance, they assess whether the two products or services can complement or substitute each other (Völckner and Sattler 2006); in other words, whether the two products or services possess the same product or service attributes, or can fulfill the same functions (Park, Milberg, and Lawson 1991). Johan Lanseng and Erling Olsen (2012) further point out that product category fit or complementarity is more important in functional than emotional brand alliance. In the functional brand alliance, each brand has relatively distinctive functional attributes. Thus product category fit can be achieved as long as the capabilities of the partner brands are compatible with each other (Johan Lanseng and Erling Olsen 2012).

Along with the line of reasoning, hotels and Tesla EV charging service in our study context are originally from different product/service categories, but both of them provide functional services (i.e. accommodation and EV charging, respectively) that are likely to evoke associations in consumers’ minds regarding solving consumption problems (Johan Lanseng and Erling Olsen 2012). Importantly, these two types of functional services are complementary with regards to capabilities and competence dimension of brand personality (van der Lans, Van Den Bergh, and Dieleman 2014), as hotel guests who arrive by driving Tesla vehicles are likely in urgent need of recharging their EVs before they leave for the next or return journey. That is, the EV charging capability is a value-added service that hotels can add into their service portfolio for their guests, in addition to the demonstration of hotels’ social responsibility of advocating environmental sustainability. Therefore, we hypothesize that:

H1: Hosting Tesla DC charging facilities at hotels has positive influence on the hotels’ revenue.

Moderating Effect of Hotels’ Segments

In addition to the functional associations, brand alliance literature also emphasizes the perceived emotional association between partner brands (Koschmann and Bowman 2018). Importantly, the congruency on brand images between partner brands is positively associated with the evaluation of brand alliance (Simonin and Ruth 1998). Besides, out of the five dimensions of brand personality (Aaker 1997), van der Lans, Van Den Bergh, and Dieleman (2014) find that the high level of coherence or similarity in an extrinsic dimension of sophistication, which is represented by attributes such as glamorous, upper-class, and charming, can result in more favorable brand alliance evaluation.

In the context of our study, Tesla has a very prominent brand image. Mangram (2012) argues that Tesla’s brand image or position embodies the characteristics of being high tech, attractive, reliable, and environmentally friendly. Long et al. (2019) identify five factors of images that Tesla-familiar consumers associate with the brand, including skeptical, symbolically appealing, pro-social innovation, economic and expensive images. Importantly, Tesla targeted middle to upper-middle-class consumers when beginning its mass production of Model S in 2012 (Mangram 2012). Thus, if hotels choose to join in Tesla’s charging service network by hosting DC facilities, it is unlikely that different segments of hotels would have the same levels of brand alliance consequence; instead, hotels that share a similar brand image or position with Tesla would have higher benefits from hosting Tesla SC facilities. When using the hotel price as the proxy of hotel segmentation and thus brand position in the market, we develop the following hypotheses:

H2a: Upscale hotels benefit more from hosting Tesla DC charging facilities than mid-priced and cheaper hotels;

H2b: Luxury hotels benefit more from hosting Tesla DC charging facilities than mid-priced and cheaper hotels.

The Influence of Tesla’s Product Line Extension

The prior literature has not paid sufficient attention to the dynamic evaluation of brand alliance when one partner brand in the brand alliance extends its product line and thus changes the target market. Taking Tesla as the example, it has gone through different stages of product strategy (Mangram 2012): (1) entering a niche market segment of the high-end sports car market to introduce the Tesla brand; (2) entering luxury vehicle sedan market by introducing Model S and Model X; and (3) entering the mainstream vehicle market by introducing mass-market vehicles such as Model 3.

Drawing on the congruency literature of brand image and position in brand alliances (Simonin and Ruth 1998; van der Lans, Van Den Bergh, and Dieleman 2014), we argue that when Tesla extended its product line from luxury vehicle market to the mass market, consumers in the luxury hotels would have a weaker emotional association with the Tesla brand. In comparison, Tesla’s such production line stretching to the mass market would strengthen the emotional association between consumers of upscale hotels and the Tesla brand, given that the Model 3 is classified as an upscale sedan by the US Department of Energy.¹ Therefore, we hypothesize that:

H3a: The benefits of upscale hotels from hosting Tesla DC facilities increase after Tesla introduced Model 3.

H3b: The benefits of luxury hotels from hosting Tesla DC facilities decrease after Tesla introduced the Model 3.

Moderating Effects of Popular Tourist Attractions

In recent years, popular tourism destinations, such as Orlando in Florida, have advocated EVs to achieve destination sustainability at the city level (Fjelstul 2014). This can be explained by the general finding in the tourism literature that tourists’ destination image, which is described as impressions of a place or perceptions of an area, has a positive effect on not only tourist destination satisfaction but also destination choice of both tourists and their social network members (Pan, Rasouli, and Timmermans 2021). Furthermore, the use of EVs is considered the most impactful technological advancement on the interface between transport and tourism (Duval 2020), and is likely to lead to substantial changes to tourism mobility, destination development, and the creation of accompanying facilities (Fitt 2021).

In the context of this study, Texas has a number of popular attractions located in different cities. The prior literature has suggested that locating EV charging facilities at or near tourist attractions could improve the tourist experience of charging and increase visitation (Fitt 2021; Wang 2011). Therefore, we argue that the popular attractions can strengthen the attractiveness of Tesla DC facilities hosted by hotels in the same city. Specifically, if there are more popular attractions in one city, the hotels in the same city are likely to achieve higher financial revenue after hosting Tesla DC facilities compared with the similar DC-hosting hotels in the cities with fewer tourist attractions. Therefore, we propose that:

H4: The benefits of hotels from hosting Tesla DC facilities are higher in cities with more popular tourist attractions.

Data

This study collected data from several different sources: (1) for information on the location of EV charging stations, we sourced data from the AFDC of the US Department of Energy; (2) for data on monthly hotel revenues, we sourced data from the Texas Comptroller of Public Accounts; (3) for the additional hotel information (e.g. hotel characteristics), we sourced data from Smith Travel Research (STR); and (4) for the popular attraction information, we sourced the attraction data of Texas from TripAdvisor website.

Tesla Destination Charging Station Data

Data on the Tesla DC stations were extracted from the dataset of the Alternative Fueling Station Locator in the AFDC of the US Department of Energy. This dataset is regularly updated and verified by the National Renewable Energy Laboratory and includes information about refueling stations of a range of alternative fuels for automobiles, including biodiesel, compressed natural gas, electric, ethanol, hydrogen, liquefied natural gas, and propane. Regardless of the type of alternative fuels, the AFDC data contains rich information, such as each station’s name, street address, zip code, city and state, type of access (private or public), hours of operation, accepted payment methods, ownership type, facility type (i.e. the type of facility at which the station is located), and the date that the station opened. For the EV stations of interest in this study, the dataset also provides information about the numbers of Level 1, Level 2, and direct-current fast-charging EV supply equipment, as well as about the connector types and the name of the EV charging network.

For this study, we identified and extracted Tesla DC stations based on them being located in the state of Texas (using the keyword “Texas” in the database search) and based on the type of EV charging network (i.e. using the keyword “Tesla Destination” in the database search).² As shown in Figure 2, there were only two Tesla DC stations at the beginning of 2015, but the number of stations had increased to 37 by 1 May 2015. There was a continuous expansion of Tesla DC stations in Texas between January 2016 and April 2017. Tesla established 42 new stations in September 2017 to reach 175 DC stations in Texas, and the network size remained stable until December 2018. The AFDC dataset indicates that among these 175 stations, 129 were located in hotel facilities, 10 in restaurants, and all other types of facilities (e.g. parking lots, shopping centers, and airports) had fewer than five DC station locations. We further validated each hotel based on the hotel census data we collected from the STR.

Figure 2.

Summary of research hypotheses.

Hotel Data

We follow Zervas, Proserpio, and Byers (2017) to collect hotel revenue data from the Texas Comptroller of Public Accounts and hotel census data from the STR. Specifically, as the dependent variable of our analysis, the monthly room rental income was obtained from the Texas Comptroller of Public Accounts (2019). In addition, the dataset also includes hotel information such as hotel name, hotel address, and a number of rooms available for rent. However, this dataset for monthly hotel revenue provides information for any property that rents a room or space to guests at the cost of $15 or more each day, which thus captures not only hotels, but also bed and breakfasts, private condominiums, apartments, and houses, and Airbnb properties (Texas Comptroller of Public Accounts 2019). To identify the real hotels and exclude non-hotel properties, we again follow Zervas, Proserpio, and Byers (2017) to cross-reference the hotel revenue data with the US hotel census data from STR. STR is a leading global company in benchmarking and analytics for the hospitality industry and tracks more than 65,000 hotels with nearly nine million guest rooms across 180 countries (Wheeler and Simonelli 2019). The STR US hotel census data provide rich data fields, including the hotel name, address, and opening date, whether the hotel is currently closed, price segment, number of floors, size of meeting spaces, whether there is a restaurant, whether there is a convention space, and operation type (i.e. independent, franchise, or chain management). In addition, the STR US census database defines the price segment differently in Metropolitan Statistical Area (MSA) markets and in rural or non-metropolitan areas (Carvell, Canina, and Sturman 2016). In MSA markets, the STR defines five price segments of hotels based on the average room rate: luxury (top 15%), upscale (next 15%), mid-price (middle 30%), economy (next 20%), and budget (lowest 20%). In rural and non-metropolitan markets, the STR merges the luxury and upscale segments into upscale (top 30%) and thus forms four price segment categories. Although the STR hotel census provides over 5,000 hotels in Texas, we found only 2,774 hotels reporting tax in 2015–2018, and thus we focus on these hotels into our empirical analysis.

Popular Attractions Data

The data of popular tourist attractions in Texas were extracted from the website of TripAdvisor.³ Specifically, we collected the name, location, and city information of 150 most popular attractions sorted by traveler favorites. Out of these 150 most popular attractions, the city of Waco is ranked highest with 10 attractions, followed by Austin and Fort Worth with 7 attractions each. When checking the number of popular attractions in each of the 51 cities with DC-hosting hotels, we find that 24 cities have zero popular attraction and 9 cities have 1 attraction in each.

Empirical Methods

We test our research hypotheses by using PSM (Rosenbaum and Rubin 1983) followed by the DID approach. In our sample, 116 of 2,774 hotels joined Tesla’s DC charging network to host DC stations between 2015 and 2018. Among these, 10 hotels installed both Tesla and non-Tesla chargers, which leads to 106 hotels with Tesla DC chargers only as the treatment group in our study. There might be a self-selection issue because the choice of hotels may not be randomly distributed, but driven by their expected benefits, such as building customer loyalty as Tesla promotes for the DC program.⁴ To correct for the self-selection issue, we employ a PSM approach (Rosenbaum and Rubin 1983) to match each hotel that installed a Tesla DC charger (i.e. from the treatment group) with another hotel without a Tesla DC charger (i.e. from the control group) based on nearest neighbor matching considering the temporal and spatial variability. Our PSM approach is conducted by running a Probit model on observable characteristics of hotels to estimate their likelihood of being selected into the treatment group (Marquis and Qiao 2020).

Table 1 presents the PSM results, which confirm the good quality of the matching. That is, in the matched sample, the decision of whether to install a Tesla DC station can be considered a random choice for different hotels to the extent that we can rule out self-selection occurred for these hosting hotels. Specifically, the balance check after matching shows insignificant differences for all characteristics between the treatment and control groups; the mean percentage bias is 4.7% in the matched sample, lower than the suggested threshold of 5% (Marquis and Qiao 2020; Rosenbaum and Rubin 1983); and the pseudo-R² of the PSM Probit model is reduced to 0.016 in the post-matched sample. Figure 3 illustrates the matching performance, suggesting the randomized pattern that the treatment and control groups are more similar after PSM.

Table 1.

PSM Probit Regression Results, Balance Check, and Percentage Bias Reduction.

	Coefficient	Balance Check after PSM		% Bias
Variable	(s.e.)	Difference	p-value	Reduction	Post-Matching
Number of rooms	0.000 (0.000)	−0.03	.97	98.9	−0.6
Location of the hotel (with reference to town)
Airport	−0.098 (0.215)	1.10	.27	−1105.4	13.4
Urban	−0.099 (0.203)	−0.16	.87	91.7	−2.5
Suburb	−0.076 (0.142)	−0.58	.56	45.9	−7.9
Interstate	−0.050 (0.187)	0.47	.64	44.4	5.8
Resort	0.535^† (0.296)	0.27	.79	85.6	4.7
Price level of the hotel (with reference to luxury hotels)
Budget	−2.063*** (0.344)	0.00	1.00	100.0	0.0
Economy	−1.386*** (0.236)	0.00	1.00	100.0	0.0
Mid-price	−0.791*** (0.185)	0.00	1.00	100.0	0.0
Upscale	−0.349* (0.174)	−0.28	.78	92.2	−4.5
Number of floors	−0.001 (0.017)	−0.47	.64	86.5	−8.6
Restaurant	−0.002 (0.138)	−0.42	.68	88.1	−6.5
Convention	0.011 (0.309)	0.00	1.00	100.0	0.0
Total meeting space	0.000 (0.000)	−0.34	.74	85.7	−7.2
Operation type of the hotel (with reference to Independent Hotels)
Chain hotels	0.079 (0.238)	0.19	.85	81.6	2.8
Franchised hotels	0.212	−0.78	.44	66.5	−10.0
	(0.176)			Pre-matching	Post-matching
No of observations	2,764	Mean bias (%)		38.4	4.7
Pseudo R-square	0.210	Pseudo R–square		0.210	0.016

†

p < .1; *p < 0.05; ***p < .001; constant not reported to save space.

Figure 3.

Graphical illustration of PSM matching performance.

We then apply the DID approach to identify the treatment effect of hosting Tesla DC stations by comparing the change in hotel revenue before and after hosting, with a baseline of differences in the revenue of hotels in the control group that did not host a Tesla DC station in the same period. The unit of analysis in our study is a hotel in one month within our observation horizon between January 2015 and December 2018. Our base model with two-way fixed effects is specified as follows:

\begin{matrix} y_{i t} = α_{0} + α_{1} {PostDC}_{i t} + α_{2} H o t e l_D C_{i} {* PostDC}_{i t} \\ + γ_{1} X_{i t} + μ_{t} + θ_{i} + ε_{i t}, \end{matrix}

(1)

where $y_{i t}$ is the dependent variable that represents the financial performance of hotel i in month t. In our empirical analysis, $y_{i t}$ is operationalized as the logarithm of a hotel’s monthly room rental revenue. To avoid the logarithm transformation issue with zero revenue cases, we adopt the safe-log transformation with ln( $y_{i t}$ + 1). In equation (1), $α_{0}$ is the constant; ${PostDC}_{i t}$ is a dummy variable with the value of 1 if the DC station has been available since month t in hotel i and its matched hotel, and 0 otherwise; $H o t e l_D C_{i}$ is another dummy variable, with the value of 1 for hotels that installed Tesla DC equipment in our observation horizon (i.e. the treatment group), and 0 otherwise (i.e. the control group). Thus, $H o t e l_D C_{i} {* PostDC}_{i t}$ is the DID term and its coefficient $α_{2}$ captures the effect of installing Tesla DC stations on the dependent variable. $X_{i t}$ consists of a range of controlled characteristics at the hotel, region, and market levels. Hotel-level factors include the number of rooms in the hotel, hotel age by the end of 2018, the type of hotel location (e.g. urban, suburban, airport, interstate, resort, or towns), the number of floors, dummy variable of hotel restaurant, dummy of conference space, and types of hotel operations; the regional-level factor is the number of hotels in the same zip code; the market-level factors are the price segment of the hotel, the cumulative sales of Tesla EVs in the US and the number of Tesla vehicle models available in the market. We further include a time fixed effect $_{t}$ to control for unobserved factors affecting all hotels at time t. $θ_{i}$ is also included as the hotel fixed effect to control for time-invariant unobserved factors. $ε_{i t}$ is the error term.

While the coefficient $α_{1}$ in equation (1) measures the total effect of installing a Tesla DC station on a hotel’s revenue, we are interested in whether this effect differs across different tiers of hotels and the dynamic effect before and after the new Tesla product (i.e. Model 3) was launched. Thus, the following specification extends the base model by introducing an interaction between the DID term defined above ( ${DID}_{i t} = H o t e l_D C_{i} {* PostDC}_{it}$ ) and the price segmentation variable ( ${PriSeg}_{i}$ ), so that we can examine whether and how the DID effect is moderated by the price segment of hotels.

\begin{matrix} y_{i t} = α_{0} + α_{1} {PostDC}_{i t} + α_{2} {DID}_{i t} + α_{3} {DID}_{i t} * {PriSeg}_{i} \\ + γ_{1} X_{i t} + μ_{t} + θ_{i} + ε_{i t} . \end{matrix}

(2)

Specifically, we include three price segments of hotels: luxury, upscale, and mid-price and cheaper (a combination of the STR’s mid-price, economy, and budget segments). Every hotel falls into one of these three price segments as specified in the STR database. Therefore, with reference to the hotels in the mid-price and cheaper price segment, the coefficient vector $α_{3}$ captures the moderating effect of whether upscale or luxury hotels benefit more by offering Tesla DC charging service.

In addition, Tesla launched the new product Model 3 in our observation horizon, which stretched its product line downwards from producing the full-size luxury sedan Model S to a compact model. Specifically, the US Department of Energy classifies the Tesla Model S as a luxury sedan and the Model 3 as an upscale sedan, and according to the US Energy Protection Agency, the size classes of Model S and Model 3 are “large car” and “compact car,” respectively.⁵ Thus, it is also important to investigate whether this product line stretching changed the economic benefits to hotels of having a Tesla DC station. Hence, equation (3) is specified to extend equation (1) by adding the term $Model 3_{t}$ , which is a dummy variable to indicate whether the Model 3 was available (i.e. since July 2017) or not (i.e. before July 2017), and an interaction between $Model 3_{t}$ and the DID term.

\begin{matrix} y_{i t} = β_{0} + β_{1} {PostDC}_{i t} + β_{2} {DID}_{i t} \\ + β_{3} Model 3_{t} + β_{4} {DID}_{i t} * Model 3_{t} \\ + γ_{1} X_{i t} + μ_{t} + θ_{i} + ε_{i t} . \end{matrix}

(3)

where $β_{4}$ measures the change in financial performance related to having a Tesla DC station when the Tesla Model 3 was available in the market, compared with before the Model 3 was available.

Further, we examine whether the tier of hotel affects the benefit to the hotel of installing a Tesla DC station before and after the launch of Tesla’s Model 3. Hence, the specification is as follows:

\begin{matrix} y_{i t} = β_{0} + β_{1} {PostDC}_{i t} + β_{2} {DID}_{i t} \\ + β_{3} Model 3_{t} + β_{4} {DID}_{i t} * Model 3_{t} \\ + β_{5} {DID}_{i t} * {PriSeg}_{i} + β_{6} {DID}_{i t} * Model 3_{t} * {PriSeg}_{i} \\ + γ_{1} X_{i} + μ_{t} + θ_{i} + ε_{i t}, \end{matrix}

(4)

where $β_{6}$ measures the relative benefits of upscale or luxury hotels, with reference to those of mid-priced and cheaper hotels, in the period after July 2017 when the Tesla Model 3 was launched in the market.

Moreover, we examine how the number of the popular attractions in the city moderates the financial benefits of the hotels in the same city after they hosted the Tesla DC infrastructure. By defining the moderating variable of the number of the popular attractions in the city of hotel i as ${PopAtt}_{i}$ , the base model is equation (1) is extended to

\begin{matrix} y_{i t} = α_{0} + α_{1} {PostDC}_{i t} + α_{2} {DID}_{i t} \\ + α_{3} {PopAtt}_{i} + α_{4} {DID}_{i t} * {PopAtt}_{i} \\ + γ_{1} X_{i t} + μ_{t} + θ_{i} + ε_{i t}, \end{matrix}

(5)

where $α_{4}$ measures the relative benefits of DC-hosting hotels in the cities with more popular attractions, compared to the DC-hosting hotels in cities with fewer attractions.

Empirical Results

Base Model Results

We first examine the effect of installing Tesla DC stations on the financial performance of hotels. Table 2 presents the effects of installing Tesla DC stations in hotels on the monthly revenue of these hotels. The results demonstrate a significant improvement in monthly revenue (β = 0.279, p < .001) in these hotels after becoming Tesla’s charging partners by installing Tesla DC infrastructure in their properties, and thus H1 is supported.

Table 2.

Estimation Results of the Base Model.

	DV: Log of Monthly Revenue
Variables	Coefficient	Robust s.e.
Constant	6.303***	1.489
Post-DC installation	−0.048	0.037
Treated hotels × post-DC installation	0.279***	0.056
Log number of hotels in the same zip code	−1.279***	0.376
Number of rooms	0.001	0.002
Hotel age by the end of 2018	0.133**	0.044
Hotel location: airport^a	6.794***	1.598
Hotel location: urban^a	2.481***	0.614
Hotel location: suburb^a	1.640^†	0.983
Hotel location: interstate^a	0.651	0.729
Hotel location: resort^a	0.996	1.000
Upscale hotels^b	2.389**	0.729
Luxury hotels^b	1.427	2.595
Number of floors	−0.731^†	0.425
Restaurant	−0.350	0.468
Convention	−32.511	26.211
Total meeting space	0.365	0.296
Chain hotels^c	14.132**	5.001
Franchised hotels^c	6.321***	1.613
Cumulative sales of Tesla cars in US (in 1,000)	−0.009	0.006
The number of available Tesla models	1.079	0.751
Hotel fixed effect	Yes
Year-month fixed effect	Yes
No of observations	9,689
R-square	0.629
Adjusted R-square	0.619
Log-likelihood	−13,626.64
Akaike’s information criterion	27,779.29
Bayesian information criterion	29,667.30

The reference category is hotel location of town.

The reference category of hotel price segment is mid-price and cheaper hotels.

The reference category is independent hotels.

†

p < .1; *p < .05; ** p < .01; *** p < .001.

As illustrated in Figure 4, the base model results reveal the following: (1) the average monthly revenue of all hotels in the pre-DC period was approximately 532.74 (=e^6.280 – 1), without considering hotel characteristics; (2) the average monthly revenue of the hotels in the control group was 507.66 (=e^{(6.280 – 0.048)} – 1) after their matched hotels installed a Tesla DC station; and (3) the average monthly revenue of the treatment hotels in the post-DC period increased to 671.04 (= e^{(6.280 – 0.048} +^0.279) – 1). Therefore, compared with the control group, the monthly revenue of the treatment hotels increased by 32.18% after installing the Tesla DC station on their property.

Figure 4.

An illustration of the DID Effect in the base model.

For the significant control variables, we find that the number of hotels in the same zip code area has a negative effect on hotels’ monthly revenue, which implies that local competition reduces hotel income. In addition, hotels in airports, urban, and suburban areas have a higher income than hotels in towns. Further, upscale hotels have a significantly higher income than mid-priced and cheaper hotels. Hotels with more floors are less likely to have a higher income. Chain and franchised hotels have a significantly higher income than independent hotels. The cumulative sales of Tesla cars and the number of available Tesla models have no significant impact on the hotel income.

Heterogeneous Effects by Price Segments of Hotels

In addition to the overall effect of hosting a Tesla DC station in hotels examined in the base model, we further investigate the heterogeneity of such benefits across different tiers of these hotels. That is, we examine how the economic benefits of installing a Tesla DC station are moderated by the price segments of the hosting hotel.

Figure 5 illustrates the numbers of hotels and hosting hotels in different price segments in Texas. Of the 2,658 non-hosting hotels in the sample, there were 162 luxury hotels, 356 upscale hotels, 542 mid-price hotels, 517 economy hotels, and 1,081 budget hotels. In comparison, the data reveal that the Tesla DC stations are disproportionately hosted by more expensive hotels. Of the 106 hosting hotels, 36 are luxury hotels, accounting for 22.22% of all luxury hotels, and 40 are upscale hotels, approximately 11.24% of all upscale hotels in Texas. In comparison, the percentage of mid-price hotels to host a Tesla DC station is only 4.43% and the corresponding percentage for the economy and budget hotels is only 0.97% and 0.09%, respectively. In our empirical analysis, we combine the three segments of mid-price, economy, and budget hotels to function as a reference category and then investigate the heterogeneous effect in the upscale and luxury price segments.

Figure 5.

Distribution of all hotels and DC-hosting hotels across five price segments.

As presented in Table 3, we find that, with reference to the price segment of mid-priced and cheaper hotels, the interaction between the treatment upscale hotels and the post-DC period has a positive and significant coefficient (β = 0.338, p < .01). This implies that the upscale hotels benefited much more than the mid-priced and cheaper hotels from installing a Tesla DC station, and thus H2a is supported. In comparison, the interaction between the treatment luxury hotels and the post-DC period has a positive but insignificant coefficient (β = 0.125, p > .1), and thus H2b is not supported. In addition, we notice that the DID main effect (i.e. hosting hotels interaction with the post-DC period) is also positive but insignificant. Therefore, these results indicate an inverted U-shaped benefit for hotels at different price segments due to installing a Tesla DC station. That is, among the three price segments of hotels, only the upscale hotels experience significantly positive benefits from joining in the Tesla DC charging network, while the benefits for both luxury hotels and mid-priced and cheaper hotels are more trivial.

Table 3.

Heterogeneous Effect by Price Segments of Hotels.

	DV: Log of Monthly Revenue
Variables	Coefficient	Robust s.e.
Constant	6.829***	1.503
Post-DC installation	−0.049	0.037
Treated hotels × post-DC installation	0.111	0.111
Treated upscale hotels^b × post-DC installation	0.338**	0.127
Treated luxury hotels^b × post-DC installation	0.125	0.132
Log number of hotels in the same zip code	−1.275***	0.375
Number of rooms	−0.001	0.002
Hotel age by the end of 2018	0.149***	0.044
Hotel location: airport^a	7.284***	1.586
Hotel location: urban^a	2.533***	0.609
Hotel location: suburb^a	1.476	0.975
Hotel location: interstate^a	0.505	0.736
Hotel location: resort^a	0.611	1.014
Upscale hotels^b	2.417***	0.721
Luxury hotels^b	0.390	2.591
Number of floors	−0.835*	0.419
Restaurant	−0.220	0.468
Convention	−42.496	26.168
Total meeting space	0.489^†	0.296
Chain hotels^c	15.676**	4.980
Franchised hotels^c	6.289***	1.603
Cumulative sales of Tesla cars in US (in 1,000)	−0.009	0.006
The number of available Tesla models	1.080	0.752
Hotel fixed effect	Yes
Year-month fixed effect	Yes
No of observations	9,689
R-square	0.630
Adjusted R-square	0.619
Log-likelihood	−13,617.67
Akaike’s information criterion	27,765.34
Bayesian information criterion	29,667.71

The reference category is hotel location of Town.

The reference category of the hotel price segment is mid-price and cheaper hotels.

The reference category is independent hotels.

†

p < .1; *p < .05; **p < .01; ***p < .001.

Dynamic Effects with Introduction of Tesla Model 3

Elon Musk had a three-step master plan on product launching: from building sports cars to building more affordable mass-market models.⁶ Following this strategy, Tesla began with an all-electric sports car, Roadster, in 2008. Then Tesla designed a full-size luxury electric sedan, Model S, and made it available on the market in late 2012, followed by an electric sport utility vehicle, Model X, in 2015. As the last step in Elon Musk’s “Secret Master Plan,” Tesla introduced its mass-market model, Model 3, in 2016 and began sales in July 2017 in the US.

We thus further analyze the dynamic effect of the release of the Model 3 on the benefits for hosting hotels and any differences between the different price segments of the hosting hotels. First, as shown in Table 4, although the main DID effect remains positive and significant (β = 0.358, p < .001), the interaction term between the DID and the post-launch period has a negative coefficient (β = –0.110, p < .05). This implies that the treatment hotels’ revenue benefits from hosting a Tesla DC station decreased after the Model 3 became available on the market.

Table 4.

Dynamic Effect after the Introduction of Tesla Model 3.

	DV: Log of Monthly Revenue
Variables	Coefficient	Robust s.e.
Constant	−6.298***	1.438
Post-DC installation	−0.084*	0.038
Treated hotels × post-DC installation	0.358***	0.060
Post launch of Model 3	0.117	0.474
Treated hotels × post-DC installation × post-launch of Model 3	−0.110*	0.045
Control variables	Yes
Hotel fixed effect	Yes
Year-month fixed effect	Yes
No of observations	9,689
R-square	0.629
Adjusted R-square	0.619
Log-likelihood	−13,624.66
Akaike’s information criterion	27,777.32
Bayesian information criterion	29,672.51

p < .05; ***p < .001.

Second, as seen in Table 5, when differentiating price segments of hotels in the post-launch period, although the treatment upscale and luxury hotels had higher revenue than the mid-price and cheaper hotels after installing a Tesla DC station (β = 0.213, p < .10; β = 0.256, p < .05), their benefits changed in the opposite direction after the Model 3 became available. Specifically, the treatment of upscale hosting hotels had additionally significant benefits after the launch of the Model 3 (β = 0.171, p < .01), while the luxury hotels suffered a decrease in benefits (β = –0.198, p < .05) after the launch of the Model 3. This result suggests that the hotels’ revenue dynamics changed after Tesla stretched its product line from high-end vehicle models to more affordable product. That is, upscale hotels that had installed a Tesla DC station benefited most from Tesla’s product line stretching to the mass market, while the revenue benefits for the hosting luxury hotels decreased. Therefore, both H3a and H3b are supported.

Table 5.

Moderating Effects of Hotel Price Segments and Post-Launch of Model 3.

	Log of Monthly Revenue
Variable	Coefficient	Robust s.e.
Constant	7.076***	1.425
Post-DC installation	−0.091*	0.038
Treated hotels × post-DC installation	0.176*	0.088
Treated upscale hotels^a × post-DC installation	0.213^†	0.111
Treated luxury hotels^a × post-DC installation	0.256*	0.118
Post launch of Model 3	0.103	0.475
Treated hotels × post-DC installation × post-launch of Model 3	−0.084	0.058
Treated upscale hotels^a × post-DC installation × post-launch of Model 3	0.171**	0.056
Treated luxury hotels^a × post-DC installation × post-launch of Model 3	−0.198*	0.078
Control variables	Yes
Hotel fixed effect	Yes
Year-month fixed effect	Yes
No of observations	9,689
R-square	0.630
Adjusted R-square	0.620
Log-likelihood	−13,607.89
Akaike’s information criterion	27,751.78
Bayesian information criterion	29,675.68

The reference category is the price segment of mid-price and cheaper hotels.

†

p < .1; *p < .05; **p < .01; ***p < .001.

Moderating Effects of Popular Attractions in the Same City

We further examine the role of popular tourist attractions in moderating the financial benefits of nearby hotels after they host Tesla DC infrastructure. In this analysis, we count the number of popular attractions in the same city of every hotel out of the 150 most favorite attractions of Texas as ranked on the website of TripAdvisor, and then interact this number for every hotel with the DID term.

As presented in Table 6, this interaction term between the DID term and the number of popular attractions in the same city as the hotel is positive and significant (β = 0.094, p < .001). This effectively means that the number of popular attractions does amplify the financial benefits of DC-hosting hotels in the same city. That is, the treatment hotels’ revenue is significantly higher from hosting Tesla DC stations if these hotels are located in cities with more popular attractions. Therefore, H4 is supported.

Table 6.

Moderating Effect of Popular Attractions in the Same City.

	DV: Log of Monthly Revenue
Variables	Coefficient	Robust s.e.
Constant	6.404***	1.398
Post-DC installation	−0.047	0.037
Treated hotels × post-DC installation	0.288***	0.056
The number of popular attractions in the same city as the hotel^a	−0.213	0.157
Treated hotels × post-DC installation × the number of popular attractions in the same city as the hotel	0.094***	0.020
Control variables	Yes
Hotel fixed effect	Yes
Year-month fixed effect	Yes
No of observations	9,689
R-square	0.631
Adjusted R-square	0.621
Log-likelihood	−13,595.16
Akaike’s information criterion	27,718.33
Bayesian information criterion	29,613.52

The popular attractions in each city are extracted out of the 150 most favorite attractions of Texas in this analysis.

***

p < .001.

Robustness Checks

We conduct four additional analyses to verify the robustness of the results. The first two robustness checks use the hotels’ monthly revenue per available room (RevPAR) and the hotels’ monthly taxable rental, respectively, as the alternative dependent variable; the third robustness check excludes new hotels to consider only hotels that opened before the start of our observation horizon; and the fourth robustness check counts the number of popular attractions in the city of every hotel out of the top 100 or top 50 most favorite attractions of Texas. See the Appendix for the results of these robustness checks.

First, when using hotels’ monthly RevPAR as the dependent variable, we obtain consistent results in the main analysis. As presented in Appendix Table A1, model (1) indicates the significant and positive effect of installing a Tesla DC station on hotels’ monthly RevPAR (β = 0.219, p < .001). After considering the different price segments of hotels, model (2) shows that the upscale hotels had higher economic benefits per room after installing a Tesla DC station (β = 0.231, p < .01) than the base category of mid-priced and cheaper hotels, while luxury hotels did not have significant benefits over the hotels in the base category. In addition, when considering the effect of Tesla’s introduction of Model 3, model (3) in Appendix Table A1 reveals that overall, the hosting hotels had reduced revenue per room after the Tesla Model 3 was launched (β = –0.075, p < .05). Model (4) further verifies our earlier finding that upscale hosting hotels had increased RevPAR when the Tesla Model 3 was launched (β = 0.147, p < .001), but the luxury hosting hotels had relatively lower revenue after this launce (β = –0.126, p < .05), with reference to the mid-priced and cheaper hotels. Model (5) in Appendix Table A1 also shows the robust result that the number of popular attractions can strengthen the monthly RevPAR of treated hotels in the same city after they host the Tesla DC stations (β = 0.057, p < .001).

Second, our dataset of hotel income also includes each hotel’s taxable rental per month, which is defined as the rental of rooms of each hotel minus all charges not subject to tax. Using the taxable rental as the alternative dependent variable, we obtain results that are largely consistent with the findings in the main analysis, including the overall DID effect, the effect of differentiating price segments of treated hotels, the effect of Tesla’s introduction of Model 3, the effect of the introduction of the Tesla Model 3′s on the income of the different price segments of the treatment hotels, and importantly the effect of popular attractions in enhancing the same-city hotels’ financial benefits after they host Tesla DC stations.

Third, our main analysis consists of 106 treatment hotels that installed a Tesla DC station between 2015 and 2018, of which 74 were opened before 2015 and 32 were new hotels opened during our observation horizon. Thus, to exclude the effect of being a new hotel on the results, we conduct another robustness check to analyze effects on the 74 older hotels only. After removing the 32 new treatment hotels as well as their matched hotels from the dataset, our analysis reveals largely robust results (see Appendix Table A3). Specifically, we find that installing a Tesla DC station had a significant and positive effect on hotels’ room rentals (β = 0.127, p < .05). In addition, compared with the mid-priced and cheaper hotels, the treatment upscale and luxury hotels had the higher income from room rentals after installing a Tesla DC station (β = 0.349 and 0.379, respectively, p < .01). Further, Tesla’s introduction of the Model 3 resulted in a lower income for hotels that had installed a Tesla DC station (β = –0.134, p < .05). Moreover, the upscale hotels that installed DC facilities had increased revenue after introducing Tesla Model 3 (β = 0.292, p < .001). Finally, when the treated hotels are located in the city with more popular attractions, these hotels can achieve significantly higher financial benefits after installing the Tesla DC facility (β = 0.099, p < .001).

Fourth, the main analysis on the moderating effect of popular attractions in every city is based on Texas’s 150 most favorite attractions. In this robustness check, we change to count the number of popular attractions at the city level from the top 100 as well as the top 50 most favorite attractions of Texas. As summarized in Appendix Table A4, the analysis shows the robust results that the revenue of hotels after installing Tesla DC infrastructure is significantly higher for those in the cities with more popular attractions out of both the top 100 most favorite sites (β = 0.115, p < .001) and the top 50 most favorite sites (β = 0.145, p < .001). Compared with the coefficient of moderating effect in the main analysis with the top 150 most favorite sites in Texas (β = 0.094, p < .001), we find that the moderating effect is intensified when considering the attractions that are more popular among tourists. Then, by following Meng, Rosenthal, and Rubin (1992) to compare these coefficients, the moderating effect caused by the top 50 most favorite attractions is significantly higher than that from either the top 100 or top 150 attractions (p < .001), while there is no significant difference on the moderating effects generated by the latter two cases. This finding further suggests a “superstar” effect of the popular tourist attractions in increasing the demand for EV charging facilities in the same city or nearby region.

Discussion and Conclusion

This paper examines the effects of hosting EV charging infrastructure on the financial income of travel destinations. Specifically, in the case of 106 Texas hotels that installed a Tesla DC station between 2015 and 2018, we adopt the PSM and DID methods to quantify the positive effect of this partnership on the hosting hotels and importantly find that nearby attractions amplify such benefits. Further, we reveal the heterogeneity of this effect across different price segments of hotels, where the upscale hotels experienced the highest benefits from hosting Tesla DC infrastructure, followed by luxury hotels. Finally, we reveal the dynamics of the effect. That is, after Tesla extended its product line by introducing its Model 3 in the market, upscale hotels that had a Tesla DC station benefited more than luxury hotels that had a Tesla DC station, which experienced reduced income. Our findings make important theoretical contributions and have several practical implications.

Theoretical Contributions

This paper contributes to the literature in three ways. First, it contributes to the tourism and destination management literature by clearly quantifying the economic benefits of the destination properties from developing the brand alliance with brands in the EV market (i.e. by joining an EV charging network). Prior empirical studies on brand alliance or co-branding in tourism and hospitality are mostly based on surveys (e.g. Dioko and So 2012; Lin 2013). To the best of our knowledge, this is one of the first empirical studies that use real market data to provide evidence of the financial effects of destinations such as hotels developing brand alliances with brands in other markets. Note that in the empirical setting of this paper, the hospitality market is distinct from the EV market, so the two types of brands examined here may have different strengths and weaknesses in their functional attributes for serving customers. The quantified positive and significant economic benefits of hosting hotels suggest that such brand alliance may generate a complementary effect, strengthening performance by jointly offering additional value to hotel customers (Newmeyer, Venkatesh, and Chatterjee 2014). More importantly, we identify the moderating effect of popular attractions nearby to strengthen the economic benefits that the hotels can achieve from hosting EV charging facilities. This further contributes to the tourism literature on how different types of destinations (such as attractions and hotels) can co-create value for customers by jointly improving customer experience in the process of advocating destination sustainability (Ruiz-Ortega, Parra-Requena, and García-Villaverde 2021; Saraniemi and Kylänen 2011).

Second, this paper highlights the importance of brand congruency in a brand alliance and thus contributes to the literature of destination brand development (Saraniemi and Komppula 2019). Unlike the complementarity in utilitarian attributes that can improve the functional performance of a brand alliance, congruent or consistent hedonic attributes, which deliver a sensory or emotional feeling about a brand, are desirable between two partnering brands (Newmeyer, Venkatesh, and Chatterjee 2014). Our empirical analysis demonstrates that focal brands in different segments (i.e. the hotels in different price segments) benefit at very different levels, even after forming a brand alliance with the same partner brand (i.e. Tesla). In this brand alliance in our study, if the hotel is perceived to have congruent hedonic attributes such as market position in the respective industries, the brand alliance is likely to deliver a consistent and even stronger brand image so that there will be a more positive financial outcome for the focal brand with congruent hedonic attributes with Tesla. From the partner selection perspective in brand alliance, in addition to the complement functional attributes, our results emphasize that the focal brand (i.e. the hosting hotel) should assess whether the partnering brand (i.e. Tesla) can deliver a similar sensory or emotional feeling as it does, so that both brands can deliver a harmonious brand image (Newmeyer, Venkatesh, and Chatterjee 2014), thus leading to a positive financial outcome for the focal brand.

Third, this paper identifies the dynamics in the economic benefits of brand alliances for focal brands and thus extends the prior tourism literature that usually adopts a static perspective on brand alliances or co-branding (Hsiao 2018; Morgan et al. 2021). Perceived brand fit is a key factor in brand alliance (Ashton and Scott 2011; Decker and Baade 2016). Our study demonstrates that the brand fit of a brand alliance may change when the partner brand alters its product portfolio or extends its product line, and the change in brand fit may cause a behavior change on the consumer side. This dynamic was prominent when the hotels in our study partnered with a new technology firm in an emerging market, Tesla in the EV market. EV makers commonly began focusing on one specific market segment and then stretched the product line upwards or downwards. For example, Tesla first introduced its full-size luxury sedan Model S in 2012 when entering the EV market, followed by the luxury sport utility vehicle Model X in 2015, and then stretched its product line downwards to the mass market by introducing the compact sedan Model 3 in 2017. Therefore, this study contributes to the literature by highlighting the effects of brand fit dynamics in a brand alliance on the financial outcome for the focal brand.

Managerial Implications

This study has important implications for business practitioners. First, destinations such as hotels and attractions are highly encouraged to participate in EVs’ social transition toward low carbon mobility. Our analysis shows that hotels’ participation can not only demonstrate their corporate social responsibility (Radisson Hotels 2021; TopHotelNews 2018) but may also bring them additional revenues. Importantly, attractions and hotels nearby should collaborate to develop a destination brand identity (Saraniemi and Komppula 2019) that will bring more financial benefits.

Second, our analysis further suggests that destinations should evaluate brand congruency before forming a brand partnership. In the case of hosting EV charging infrastructure, our study identifies that upscale hotels, which have a similar brand position to Tesla in their respective markets, achieved significantly higher economic benefits than hotels in other price segments. That is, the economic benefits for mid-priced and cheaper hotels and luxury hotels would be limited if they chose to host a Tesla DC station because their target customers are misaligned with Tesla car drivers. Instead, these hotels may consider installing a general EV charging infrastructure independent of specific EV brands and thus may not lead to the problem of brand misalignment.

Third, our study reveals that the economic benefits of a brand alliance are dynamic. Destinations such as hotels should constantly monitor and evaluate the congruency with the partner brand. For example, when the partner in a brand alliance stretches its product line or even extends the brand, it may change the customers’ perception of its brand position and thus affect the congruency of the brand alliance. Therefore, hotels should consider congruency dynamics when developing a brand alliance with an EV brand. They may consider hosting the charging infrastructure of multiple EV brands to diversify the service offering. Further, hotel groups with multiple sub-brands of hotels may have the additional advantage of holistically managing the brand alliance with a specific EV brand (e.g. Tesla) by hosting charging infrastructure in hotels in different price segments in accordance with the EV product line extension.

Limitations and Future Research Directions

Our study has several limitations that provide directions for future research. While our research represents the pilot study for estimating the effect of brand fit between EV technologies and hotels, the data are limited to hotels in Texas. Researchers can adopt our research approach to verify the effects (as well as the heterogeneity and dynamics) in other markets. Second, Tesla is a leading firm that is enthusiastic about collaborating with hotels to expand its DC network, but other charging operators also seek such collaboration, which leads to the possibility that one hotel may host more than one brand’s charging infrastructure (i.e. multi-homing) (Belleflamme and Peitz 2019). Future research should examine how the economic benefits of hotels change if they choose to host more than one brand of EV charging infrastructure. Third, we examine only hotels’ economic benefits. Future research should examine customer and social benefits at the individual level.

Footnotes

Appendix: Robustness Check Results

Table A4.

Consider the Moderating Role of the More Popular Attractions.

	(1). Among top 100 attractions of Texas		(2). Among top 50 attractions of Texas
Variable	Coefficient	Robust s.e.	Coefficient	Robust s.e.
Constant	6.078***	1.325	6.590***	1.410
Post-DC installation	−0.048	0.037	−0.047	0.037
Treated hotels × post-DC installation	0.290***	0.056	0.267***	0.056
The number of popular attractions in the same city as the hotel	−0.226	0.156	−0.374	0.303
Treated hotels × post-DC installation × the number of popular attractions in the same city as the hotel	0.115***	0.023	0.145***	0.031
Control variables	Yes		Yes
Hotel fixed effect	Yes		Yes
Year-month fixed effect	Yes		Yes
No of observations	9,689		9,689
R-square	0.632		0.631
Adjusted R-square	0.622		0.621
Log-likelihood	−13590.36		−13600.13

***

p < .001.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is funded by the National Natural Science Foundation of China (Grant No. 71973107, 71871065, 72033003 and 71832002) and the Major Project of National Social Science Fund of China (21&ZD119).

ORCID iD

Lixian Qian

Notes

Author Biographies

Lixian Qian is a Senior Associate Professor in Marketing and Innovation at International Business School Suzhou, Xi’an Jiaotong-Liverpool University, China. His main research interests include adoption and diffusion of low-carbon innovations, data-driven marketing strategies, and energy and envionment sustainability. He has published research articles in renowed journals such as Journal of Business Ethics, Journal of Business Research, Technological Forecasting & Social Change, Transportation Research Part A & Part D, and Energy Policy.

Cheng Zhang is a professor at School of Management, Fudan University, China. His research interests include IT business value, digital platform and innovation. He has published his research in prestigious journals including Marketing Science, Journal of Marketing, MIS Quarterly, Journal of Management Information Systems, Journal on Computing and Production and Operations Management.

References

Aaker

Jennifer L.

1997. “Dimensions of Brand Personality.” Journal of Marketing Research 34 (3): 347–56.

Alternative Fuels Data Center of the US Department of Energy. 2018. “Alternative Fueling Station Locator.” Last modified September 24. http://www.afdc.energy.gov/locator/stations/.

Alternative Fuels Data Center of the US Department of Energy. 2021. “Developing Infrastructure to Charge Plug-In Electric Vehicles.” Last modified April 7. https://afdc.energy.gov/fuels/electricity_infrastructure.html.

Ashton

Ann Suwaree

Scott

Noel

. 2011. “Hotel Restaurant Co-branding: The Relationship of Perceived Brand Fit With Intention to Purchase.” Journal of Vacation Marketing 17 (4): 275–85.

Baumgarth

Carsten

. 2004. “Evaluations of Co-brands and Spill-over Effects: Further Empirical Results.” Journal of Marketing Communications 10 (2): 115–31.

Belleflamme

Paul

Peitz

Martin

. 2019. “Platform Competition: Who Benefits From Multihoming?” International Journal of Industrial Organization 64:1–26.

Buckley

Ralf

. 2012. “Sustainable Tourism: Research and Reality.” Annals of Tourism Research 39 (2): 528–46.

Carvell

Steven A.

Canina

Linda

Sturman

Michael C.

2016. “A Comparison of the Performance of Brand-Affiliated and Unaffiliated Hotel Properties.” Cornell Hospitality Quarterly 57 (2): 193–201.

Casidy

Riza

Wymer

Walter

O'Cass

Aron

. 2018. “Enhancing Hotel Brand Performance Through Fostering Brand Relationship Orientation in the Minds of Consumers.” Tourism Management 66:72–84.

10.

Cherubini

Sergio

Iasevoli

Gennaro

Michelini

Laura

. 2015. “Product-Service Systems in the Electric Car Industry: Critical Success Factors in Marketing.” Journal of Cleaner Production 97:40–9.

11.

CNN Business. 2021. “Hertz is buying 100,000 Teslas.” Last modified January 1. https://edition.cnn.com/2021/10/25/business/hertz-tesla-purchase/index.html.

12.

Decker

Carolin

Baade

Annika

. 2016. “Consumer Perceptions of Co-branding Alliances: Organizational Dissimilarity Signals and Brand Fit.” Journal of Brand Management 23 (6): 648–65.

13.

Denizci Guillet

Basak

Tasci

Asli D. A.

2010. “Travelers’ Takes on Hotel—Restaurant Co-branding: Insights for China.” Journal of Hospitality & Tourism Research 34 (2): 143–63.

14.

Dioko

Leonardo (Don) A. N

Siu-Ian (Amy)

. 2012. “Branding Destinations Versus Branding Hotels in a Gaming Destination—Examining the Nature and Significance of Co-branding Effects in the Case Study of Macao.” International Journal of Hospitality Management 31 (2): 554–63.

15.

Dong

Jing

Liu

Changzheng

Lin

Zhenhong

. 2014. “Charging Infrastructure Planning for Promoting Battery Electric Vehicles: An Activity-Based Approach Using Multiday Travel Data.” Transportation Research Part C: Emerging Technologies 38:44–55.

16.

Duval

David Timothy

. 2020. “Transport and Tourism: A Perspective Article.” Tourism Review 75 (1): 91–4.

17.

Fitt

Helen

. 2021. “Boring and Inadequate? A Literature Review Considering the Use of Electric Vehicles in Drive Tourism.” Current Issues in Tourism. doi:10.1080/13683500.2021.1937074.

18.

Fjelstul

Jill

. 2014. “Vehicle Electrification: On the “green” Road to Destination Sustainability.” Journal of Destination Marketing & Management 3 (3): 137–9.

19.

Gerdt

Sven-Olaf

Wagner

Elisa

Schewe

Gerhard

. 2019. “The Relationship Between Sustainability and Customer Satisfaction in Hospitality: An Explorative Investigation Using eWOM as a Data Source.” Tourism Management 74:155–72.

20.

Hall

C. Michael

. 2019. “Constructing Sustainable Tourism Development: The 2030 Agenda and the Managerial Ecology of Sustainable Tourism.” Journal of Sustainable Tourism 27 (7): 1044–60.

21.

Hao

Wang

Hewu

Lin

Zhenhong

Ouyang

Minggao

. 2020. “Seasonal Effects on Electric Vehicle Energy Consumption and Driving Range: A Case Study on Personal, Taxi, and Ridesharing Vehicles.” Journal of Cleaner Production 249:119403.

22.

Hopkins

Debbie

. 2020. “Sustainable Mobility at the Interface of Transport and Tourism.” Journal of Sustainable Tourism 28 (2): 129–43.

23.

Hsiao

Teng-Yuan

. 2018. “A Study of the Effects of Co-branding Between Low-Carbon Islands and Recreational Activities.” Current Issues in Tourism 21 (5): 529–46.

24.

Huang

Youlin

Qian

Lixian

. 2021. “Consumer Adoption of Electric Vehicles in Alternative Business Models.” Energy Policy 155:112338.

25.

Johan Lanseng

Even

Erling Olsen

Lars

. 2012. “Brand Alliances: The Role of Brand Concept Consistency.” European Journal of Marketing 46 (9): 1108–26.

26.

Kempton

Willett

. 2016. “Electric Vehicles: Driving Range.” Nature Energy 1 (9): 16131.

27.

Koschmann

Anthony

Bowman

Douglas

. 2018. “Evaluating Marketplace Synergies of Ingredient Brand Alliances.” International Journal of Research in Marketing 35 (4): 575–90.

28.

Lambert

Fred

. 2017a. “Tesla Expands its Destination Charging Network to Over 5,000 Locations.” Last modified June 5. https://electrek.co/2017/04/15/tesla-destination-charging-network/.

29.

Lambert

Fred

. 2017b. “Tesla to Expand Charging Station Program With Hilton Hotels to up to 200 Chargers.” Last modified May 23. https://electrek.co/2017/11/21/tesla-destination-charging-stations-hilton-hotels/.

30.

Lambert

Fred

. 2020. “Tesla pushes to Upgrade Destination Chargers with Gen 3 Wall Connectors, Enables Paid Charging.” Last modified July 21. https://electrek.co/2020/08/11/tesla-upgrade-destination-chargers-with-gen-3-wall-connector-enables-paid-charging/.

31.

Lim

Michael K.

Mak

Ho-Yin

Rong

Ying

. 2015. “Toward Mass Adoption of Electric Vehicles: Impact of the Range and Resale Anxieties.” Manufacturing & Service Operations Management 17 (1): 101–19.

32.

Lin

Yi-Chin

. 2013. “Evaluation of Co-branded Hotels in the Taiwanese Market: The Role of Brand Familiarity and Brand Fit.” International Journal of Contemporary Hospitality Management 25 (3): 346–64.

33.

Long

Zoe

Axsen

Jonn

Miller

Inger

Kormos

Christine

. 2019. “What Does Tesla Mean to Car Buyers? Exploring the Role of Automotive Brand in Perceptions of Battery Electric Vehicles.” Transportation Research Part A :Policy and Practice 129:185–204.

34.

Mangram

Myles Edwin

. 2012. “The Globalization of Tesla Motors: A Strategic Marketing Plan Analysis.” Journal of Strategic Marketing 20 (4): 289–312.

35.

Marquis

Christopher

Qiao

Kunyuan

. 2020. “Waking From Mao’s Dream: Communist Ideological Imprinting and the Internationalization of Entrepreneurial Ventures in China.” Administrative Science Quarterly 65(3):795–830.

36.

Meng

Xiao-li

Rosenthal

Robert

Rubin

Donald B.

1992. “Comparing Correlated Correlation Coefficients.” Psychological Bulletin 111 (1): 172–5.

37.

Morgan

Ashlee

Wilk

Violetta

Sibson

Ruth

Willson

Gregory

. 2021. “Sport Event and Destination Co-branding: Analysis of Social Media Sentiment in an International, Professional Sport Event Crisis.” Tourism Management Perspectives 39:100848.

38.

New Jersey Board of Public Utilities. 2021. “NJBPU Launches Grant Program for Electric Vehicle Charging at New Jersey Tourism Destinations.” Last modified January 1. https://www.bpu.state.nj.us/bpu/newsroom/2021/approved/20210930.html.

39.

Newmeyer

Casey E.

Venkatesh

Chatterjee

Rabikar

. 2014. “Cobranding Arrangements and Partner Selection: A Conceptual Framework and Managerial Guidelines.” Journal of the Academy of Marketing Science 42 (2): 103–18.

40.

Newmeyer

Casey E.

Venkatesh

Ruth

Julie A.

Chatterjee

Rabikar

. 2018. “A Typology of Brand Alliances and Consumer Awareness of Brand Alliance Integration.” Marketing Letters 29 (3): 275–89.

41.

Norman

Andrew T.

2016. “The Effects of Product Fit and Brand Fit on Memory Retention for Brand Alliances: When Less is More.” International Journal of Management and Marketing Research 9 (1): 13–28.

42.

Pan

Xiaofeng

Rasouli

Soora

Timmermans

Harry

. 2021. “Investigating Tourist Destination Choice: Effect of Destination Image From Social Network Members.” Tourism Management 83:104217.

43.

Park

C. Whan

Jun

Sung Youl

Shocker

Allan D.

1996. “Composite Branding Alliances: An Investigation of Extension and Feedback Effects.” Journal of Marketing Research 33 (4): 453–66.

44.

Park

C. Whan

Milberg

Sandra

Lawson

Robert

. 1991. “Evaluation of Brand Extensions: The Role of Product Feature Similarity and Brand Concept Consistency.” Journal of Consumer Research 18 (2): 185–93.

45.

Qian

Lixian

Grisolía

Jose M.

Soopramanien

Didier

. 2019. “The Impact of Service and Government-Policy Attributes on Consumer Preferences for Electric Vehicles in China.” Transportation Research Part A: Policy and Practice 122:70–84.

46.

Radisson Hotels. 2021. Think Planet: relax and recharge at Radisson Hotels” Last modified July 14. https://www.radissonhotels.com/en-us/partner/tesla.

47.

Rezvani

Zeinab

Jansson

Johan

Bodin

Jan

. 2015. “Advances in Consumer Electric Vehicle Adoption Research: A Review and Research Agenda.” Transportation Research Part D: Transport and Environment 34:122–36.

48.

Rosenbaum

Paul R.

Rubin

Donald B.

1983. “The Central Role of the Propensity Score in Observational Studies for Causal Effects.” Biometrika 70 (1): 41–55.

49.

Ruhanen

Lisa

Moyle

Char-lee

Moyle

Brent

. 2019. “New Directions in Sustainable Tourism Research.” Tourism Review 74 (2): 138–49.

50.

Ruiz-Ortega

María José

Parra-Requena

Gloria

García-Villaverde

Pedro Manuel

. 2021. “From Entrepreneurial Orientation to Sustainability Orientation: The Role of Cognitive Proximity in Companies in Tourist Destinations.” Tourism Management 84:104265.

51.

Saraniemi

Saila

Komppula

Raija

. 2019. “The Development of a Destination Brand Identity: A Story of Stakeholder Collaboration.” Current Issues in Tourism 22 (9): 1116–32.

52.

Saraniemi

Saila

Kylänen

Mika

. 2011. “Problematizing the Concept of Tourism Destination: An Analysis of Different Theoretical Approaches.” Journal of Travel Research 50 (2): 133–43.

53.

Simonin

Bernard L.

Ruth

Julie A.

1998. “Is a Company Known by the Company It Keeps? Assessing the Spillover Effects of Brand Alliances on Consumer Brand Attitudes.” Journal of Marketing Research 35 (1): 30–42.

54.

Tesla. 2016. “Annual Report For the Fiscal Year Ended December 31, 2015.” Tesla Motors Inc. Last modified May 7. https://www.sec.gov/Archives/edgar/data//1318605/000156459016013195/tsla-10k_20151231.htm

55.

Tesla. 2021. “Tesla Supercharger.” Last modified July 8. https://www.tesla.com/supercharger.

56.

Texas Comptroller of Public Accounts. 2019. “Hotel Tax Receipts.” Last modified February 26. https://data-secure.comptroller.texas.gov/.

57.

TopHotelNews. 2018. “Hitting the Road for the Holidays in an Electric Vehicle? Why Hilton Should be in Your GPS.” Last modified May 23. https://tophotel.news/hitting-the-road-for-the-holidays-in-an-electric-vehicle-why-hilton-should-be-in-your-gps/.

58.

U.S. Department of Energy. 2016. “Fact #939: August 22, 2016 All-Electric Vehicle Ranges Can Exceed Those of Some Gasoline Vehicles.” Last modified June 1. https://www.energy.gov/eere/vehicles/fact-939-august-22-2016-all-electric-vehicle-ranges-can-exceed-those-some-gasoline.

59.

van der Lans

Ralf

Van Den Bergh

Bram

Dieleman

Evelien

. 2014. “Partner Selection in Brand Alliances: An Empirical Investigation of the Drivers of Brand Fit.” Marketing Science 33 (4): 551–66.

60.

Völckner

Franziska

Sattler

Henrik

. 2006. “Drivers of Brand Extension Success.” Journal of Marketing 70 (2): 18–34.

61.

Wager

Guido

Whale

Jonathan

Braunl

Thomas

. 2016. “Driving Electric Vehicles at Highway Speeds: The Effect of Higher Driving Speeds on Energy Consumption and Driving Range for Electric Vehicles in Australia.” Renewable and Sustainable Energy Reviews 63:158–65.

62.

Wang

Xiao-kang

Peng

Juan-juan

Wang

Jian-qiang

. 2020. “The Differences in Hotel Selection Among Various Types of Travellers: A Comparative Analysis With a Useful Bounded Rationality Behavioural Decision Support Model.” Tourism Management 76:103961.

63.

Wang

Ying-Wei

. 2011. “Locating Flow-Recharging Stations at Tourist Destinations to Serve Recreational Travelers.” International Journal of Sustainable Transportation 5 (3): 153–71.

64.

Welford

Richard

Ytterhus

Bjarne

. 2004. “Sustainable Development and Tourism Destination Management: A Case Study of the Lillehammer Region, Norway.” The International Journal of Sustainable Development and World Ecology 11 (4): 410–22.

65.

Wheeler

Scott

Simonelli

Richard

. 2019. “CoStar Group to Acquire STR, a Global Leader in Benchmarking & Analytics for the Hospitality Industry.” Last modified September 25. https://www.businesswire.com/news/home/20191001005467/en/CoStar-Group-Acquire-STR-Global-Leader-Benchmarking.

66.

Yan

Shen

Huawen James

. 2021. “When to Get It is to Eat it—Assessing Brand Fit-Mediated Equity of Extended Restaurants of a High-End Automobile Vendor.” Journal of Hospitality and Tourism Management 47:202–9.

67.

Zervas

Georgios

Proserpio

Davide

Byers

John W.

2017. “The Rise of the Sharing Economy: Estimating the Impact of Airbnb on the Hotel Industry.” Journal of Marketing Research 54 (5): 687–705.