Introduction

Tourism industry plays a critical role in many countries in terms of prompting economic growth (Brida et al. 2020), supporting employment (Dogru et al. 2020), and reducing poverty (Croes 2014, Shang et al. 2023), particularly in developing nations (Tiwari et al. 2019). However, since 2020, the situation has changed due to the outbreak of the COVID-19 pandemic. Temporary lockdowns, border closures, social distancing, and other governmental interventions worldwide (Moosa 2020) have created great difficulties to the global economy, especially to the tourism industry.

Three years after the outbreak, analysts express optimism regarding the potential return of tourism to pre-pandemic levels in certain regions by 2023, such as Europe and the Middle East (UNWTO 2023). However, significant uncertainties persist due to economic slowdown and geopolitical tensions globally (such as the Russia-Ukraine conflict). Apart from economic and political factors, the psychological repercussions of the COVID-19 pandemic should not be overlooked (Gavurova et al. 2023). This is especially relevant to China, where quarantine measures are generally more extensive and prolonged compared to most other regions globally (Gan et al. 2022).

Nonetheless, the recovery of tourism following this unprecedented health crisis is complex and requires significant efforts from both the industry and governments (Assaf and Scuderi, 2020). In fact, governments worldwide have made a series of efforts on reviving the tourism industry. The policy report by the OECD (2020) suggests that a “joined up” approach is needed, including the lifting of restrictions, restoration of confidence, and stimulation of tourism demand. Sharma et al. (2021) review 35 papers related to the impact of COVID-19 on tourism. They highlight the need to build resilience for revitalize the global tourism industry, with government support playing a critical role. Fiscal and monetary policies, such as traveler subsidies (e.g., European Commission, 2020), financial support to the tourism companies (Collins-Kreiner and Ram 2021), and stimulus plans aimed at boosting tourism demand (Gidebo 2021), are commonly employed in recovery effort in numerous countries. A recent work by Şengel et al. (2023) documents a range of fiscal and monetary policies in response to the pandemic, advocating for long-term strategies and innovative approaches (e.g., digitalized technology) to reignite tourism demand.

The tourism industry has played an important role in the Chinese economy. Statistics show that the total revenue of domestic tourism exceeded 5.7 trillion yuan (RMB) in 2019. However, the number dropped substantially to 2.23 trillion yuan in 2020 as the consequence of the COVID-19 pandemic, underscoring the urgent need for government support (Wang et al. 2022). Like other countries in the world, the Chinese government introduced a combination of policies to support the industry during the pandemic period. With sustained governmental incentives, China’s tourism revenue in 2023 surged to 4.91 trillion RMB. Although a gap persists compared to the 2019 revenue, the trend has begun to reverse.

Among various means and policy tools, digital travel vouchers emerge as an innovative approach. Vouchers play an important role in incentivizing travel (Root 2001), and are regarded as a significant instrument of fiscal stimulus for boosting private consumption (Kan et al. 2017). Additionally, the recent trend of digitalization has reshaped the travel and tourism landscape (Jiang and Phoong 2023; Pencarelli, 2020). This is especially relevant for China, where smartphone-based digital transactions are ubiquitous, rendering digital vouchers a natural choice. According to our data, the total value of digital travel vouchers issued in China exceeded 4.9 billion yuan between 2020 and 2021. The question remains: to what extent does the issuance of these vouchers contribute to the recovery of domestic tourism in China?

Unlike many existing studies that rely on surveys or interviews to evaluate policy effects or comprehend travelers’ behaviors, this paper explores the corresponding issues by utilizing data-mining technology to extract data from online sources. We collected information on passenger flows over a span of 16 months across 306 cities in China, yielding over 1.4 million data points. The data are then used to examine the effectiveness of digital travel vouchers on domestic tourism. Employing city-level data for the empirical analysis further facilitates the exploration of regional spillover effects.

The COVID-19 pandemic has notably disrupted travel patterns (Borkowski et al. 2021), with individuals tending to travel less frequently, cover shorter distances, and opt for private transportation. Psychological barriers arising from the aftermath of the pandemic can exert significant impacts on behavior (Li et al. 2023; Gavurova et al. 2023). According to the Theory of Planned Behavior (TPB) model (Ajzen 1991) and its extensions (Bosnjak et al. 2020), human behavior is driven by subjective norms and perceived behavioral control, which ultimately dictate behavioral intention and actual behavior. When applying this theory to the present context, the impact of digital travel vouchers may alter tourists’ behavior through influencing their behavioral intentions.

Moreover, arguments regarding localization in the tourism industry are pertinent (Pappalepore and Gravari-Barbas 2022). Policies to combat the pandemic exhibited variations across regions. Compared to other countries like Italy and Slovenia, which also introduced vouchers to stimulate the tourism sector, China implemented more prolonged and stricter border controls. Given its vast geographical expanse compared to European counterparts, China also contends with regional barriers. For example, even within the country, enforcement of policies tends to diverge substantially among cities, engendering distinct localized patterns. These shifts underscore the imperative of exploring and understanding the spatial effects of supportive policies, motivating us to further investigate any potential spatial patterns in the relationship between digital travel vouchers and traveler flows.

The main contributions of this paper are as follows. First, we use a web-crawling tool to construct a rich dataset, facilitating the capture of actual traveler flows across cities in China. Additionally, we manually collect the information of digital travel vouchers issued during the studied period. These unique datasets can provide critical information to answer the questions raised above. Second, by examining the impact of digital travel vouchers on traveler flows and exploring the intricate characteristics of these vouchers, this study enriches the literature by combining digital transformation and policy supports during the pandemic to explore their impacts on the tourism sector. It offers critical evidence to authorities on how to effectively use digital platforms to implement policies. In the contemporary landscape of rapidly evolving digital economies globally, the evidence from China can offer valuable insights to other nations. Third, by utilizing a spatial regression model to address regional spillover effects, our analysis provides useful information regarding the argument of localization in the tourism industry.

Literature review and hypothesis development

Economic and behavioral logic of tourism stimulus policies

Essentially, the tourism market can be analyzed through the lens of classical economic theory, which examines tourism demand and supply (Eadington and Redman 1991). The role of the government is supposed to regulate the relationship between buyers and sellers and safeguard consumer interests. Intervention becomes imperative when the market encounters turmoil (Joppe 2018), as observed during the pandemic. As the COVID-19 pandemic spread globally, the tourism industry across the world faced severe repercussions. Clearly, the challenges confronting this industry cannot be tackled solely by internal mechanisms and thus necessitate robust policy backing from the government (Wang and Chou 2024).

Collins-Kreiner and Ram (2021) argue that the crisis gripping the tourism industry in the wake of the COVID-19 pandemic is unprecedented and has the potential to catalyze a global transformation within this sector. Strategies outlined by the UNWTO to combat the crisis, as summarized by Collins-Kreiner and Ram (2021), encompass three main categories: managing and mitigating the crisis, stimulating and accelerating recovery, and making ex ante preparations. All of these necessitate active engagement from authorities and include interventions from both the demand and supply sides.

On the demand side, the collapse in tourism demand stems not only from economic consequences but also from risk perceptions. Wang and Chou (2024), for example, demonstrate that heightened health risks due to the pandemic contribute to the downturn in tourism demand. Addressing such a multifaceted crisis is evidently challenging. According to the Theory of Planned Behavior (TPB) model, behavioral beliefs, normative beliefs, and control beliefs are three main factors shaping human behavior (Ajzen 1991; Bosnjak et al. 2020). Following this line of thought, Soliman (2019) extends the TPB model to predict tourist’s intentions. Liu et al. (2021) also attempt to understand the travel intentions of Chinese residents in the wake of the pandemic using the TPB framework. Both studies highlight the pivotal role of perception in influencing travel intentions.

Rebuilding consumer confidence and beliefs is a complicated process, wherein government support becomes critical. Government-issued vouchers have served as a mechanism for decades to stimulate consumer spending and bolster specific industries (Wilde et al. 2009). During the economic crisis of 1939, the United States established the world’s first food stamp program (Nestle and Guttmacher 1992), which was the prototype of consumption vouchers. Research on consumer behavior during economic recessions has found that government-issued consumption vouchers generally stimulate purchasing behavior among residents (Hsieh et al. 2010; Kan et al. 2017), especially increasing the purchasing desire of low-income groups. However, the stimulating effect varies across different types of goods.

As for travel vouchers, it is also found to stimulate demand. For example, Yan and Zhang (2012) find that China’s tourism consumption vouchers issued in 2009 greatly unleashed consumption potential. Zhao et al. (2014) suggest, however, that coupons do not invariably yield desired outcomes and should be carefully designed to be functional. Sigala (2020) argues that travel vouchers can be used to motivate or incentivize tourists. By combining these empirical results and insights from the TPB theory, it becomes evident that vouchers are important policy support for the recovery of the tourism industry. In summary, the following hypothesis can be proposed:

H1: Digital travel vouchers encourage more potential tourists to travel.

Characteristics of China’s digital travel vouchers

The above studies, however, mainly focus on paper vouchers issued rather long time ago, which can hardly provide useful references to the current situation during the post-COVID period. Due to the rapid development of digitalization, China has experienced fundamental changes in e-commerce, mobile payment systems, and digital services (Jiang and Murmann 2022; Zhou et al. 2024). People have started to enjoy browsing travel packages online and booking hotels and tickets in advance. This self-service online booking model makes it easier for travelers to find cost-effective travel packages and reduces the cost of changing itineraries (Perelygina et al. 2022). As a result, digital travel vouchers have been more attractive. The benefits of digitizing vouchers are broad. They are easy to use and often come with clear information advantages (Zimmer et al. 2021). Evidence presented in Joo et al. (2020) shows that social media plays a notable role in attracting tourists, also based on the TPB framework.

Digital travel vouchers are typically promoted through social media or third-party internet platforms, thereby conferring information advantages that enhance information sharing and influence consumer decisions (Toubes et al. 2021). So far, research on government digital vouchers post-COVID-19 mainly seeks to evaluate the effectiveness of individual vouchers within the cities where they are issued (Liu et al. 2021; Wang et al. 2022; Xing et al. 2023). These studies find that vouchers promote consumption, and that the characteristics of vouchers determine consumers’ purchasing preferences. For example, the size and creditability of the issuing platform significantly impact the effectiveness of vouchers (Cabiddu et al. 2013). In general, vouchers issued by larger platforms are expected to work more effectively, which leads to Hypothesis 2:

H2: The distinct characteristics of digital travel vouchers result in variations in their effectiveness in promoting local tourism.

Spatial spillover effect of China’s digital travel vouchers

In practice, digital travel vouchers in China are issued by local governments, often at the city level. There are clear regional imbalances both in terms of tourism destinations and infrastructure development (Pratt 2015). Moreover, regional economic development is strongly connected with regional tourism (Calero and Turner 2020). One of the most significant features of the tourism industry post-pandemic is the tendency for individuals to travel shorter distances (Borkowski et al. 2021). Some recent studies have also pointed out that tourists are more likely to explore local destinations in response to uncertainties stemming from the pandemic, potentially heralding a trend of localization within the tourism sector (Pappalepore and Gravari-Barbas 2022). However, digital consumption vouchers possess characteristics that transcend administrative geographical boundaries, thereby mitigating the psychological distance felt by tourists, reducing dependence of regional tourism economy on administrative geographical space, and broadening the spatial scope of tourist flow (Wang et al. 2014). The implication of these analyses suggests that digital travel vouchers, if effective, may exhibit spatial patterns. In other words, the impact of these vouchers may generate spatial spillovers, leading to Hypothesis 3:

H3: Local digital travel vouchers can benefit tourism in nearby cities.

Acknowledging the regional barriers imposed by the COVID-19 pandemic also emphasizes the need to study the determinants of inter-regional travel within a country. This is particularly relevant to China and some major economies across the world, where domestic tourism is a key contributor to total spending. For example, domestic tourism accounts for around 94% of total tourism spending in Brazil and over 80% in India, Germany, China, Japan, Mexico, the UK, and the US (WTTC 2018). Policies aimed at revitalizing the domestic/local tourism industry can create more employment opportunities, greater profits to small businesses, and generate overall economic benefits conducive to local economic recovery (Rocca and Zielinski 2022). In general, there is a clear gap in the literature regarding quantitative analyses of the role of digital travel vouchers in the tourism industry and understanding how to design relevant policies more effectively.

Sample selection and data sources

Explained variable

Obtaining high-quality traveler information is key to this research. To achieve this, a web-crawler tool is used to gather daily tourist inflow data from the Baidu Migration website. The points-of-interest data are associated with Baidu Map, a leading Internet navigator service provider in China. Technically, Python is used for the crawling process. Requests are sent in JavaScript Object Notation (JSON) format to acquire webpage URLs containing information about tourist inflows during each period. The information is then extracted and used to generate city-level tourist inflows. Monthly data from September 2020 to December 2021 for 306 cities are used for empirical analysis. To ensure the reliability of the data collected by web-crawler code, we randomly select tourist inflow data from 50 cities within our sample of 306 cities and manually compare it with data obtained by the crawler tool.

It should be noted that the Baidu Migration data on population inflows represents an index rather than a precise measure of actual travelers; therefore, the number does not reflect the exact number of individuals who traveled. Nevertheless, there are still many advantages to use the Baidu Migration Index as a proxy variable for passenger flows. The index facilitates the dynamic examination of population flow across the entire region of China (Deville et al. 2014) and exhibits strong reliability (Pei et al. 2014). Moreover, it provides a relatively uniform data type, enabling comparisons of inter-city passenger flow over an extended period (Wei and Wang 2020). This same dataset has been used by Huang et al. (2017) and Wei et al. (2018). Notably, population flow data may not perfectly align with the flow of tourists, as it is difficult to distinguish tourists from business travelers or daily commuters (Wu et al. 2020). However, Li et al. (2020) suggests that inter-city business travels exhibit relative stability, especially within a reasonably short timeframe. Based on this argument, the daily fluctuations in population inflows are likely primarily driven by tourists. Wu et al. (2021) also use similar migration data to study spatial characteristics and patterns of tourism inflows in China. Therefore, we opt to use the population inflows provided by the Baidu Migration Index as a proxy for tourist inflows.

Core explanatory variable

In previous literature, search engines have been widely utilized for data hand-collection (He et al. 2022; Shen et al. 2017), given the absence of an official release platform that comprehensively covers all digital travel vouchers at the city level. For the issuance of digital travel vouchers discussed in this paper, relevant news items are typically released on official or third-party platforms’ websites. To gather this information, we employ the Baidu search engine, using keywords such as “City i year x month t travel voucher”. We then manually collect relevant information from the links provided by Baidu. An example of such news items can be accessed at https://www.k1u.com/trip/92505.html, which provides details about digital travel vouchers in Weifang City, Shandong Province, in 2021. The webpage provides information on the vouchers’ validity period, their monetary value, the identities of businesses supporting voucher usage, and additional information.

These vouchers primarily serve to stimulate demand. Typically, non-local tourists can claim digital travel vouchers to defray expenses incurred at restaurants, hotels, tickets, and other relevant businesses. The receiving businesses, such as hotel proprietors, can then redeem these vouchers through the issuing platform.

Overall, a total of 171 issuances were identified, providing information on total volume, time frames, and issuing platforms. The monthly number of issuances over the entire sample period is depicted in Fig. 1. A clear seasonality in the time trend can be observed, with two peaks observed in October, corresponding to the China National Day holiday period.

Fig. 1
figure 1

Digital travel voucher issuance from September 2020 to December 2021.

Full size image

The issuance of digital travel vouchers also exhibits regional variation. To illustrate the distribution of issuances, we plot the total value of vouchers across each city during the entire sample period (up to the end of December 2021) in Fig. 2.

Fig. 2
figure 2

Map of digital travel voucher issuance in cities.

Full size image

Control variables

For the selection of control variables, there is a rich literature discussing the links between several city-level characteristics and the development of local tourism. Following Frick and Rodríguez-Pose (2018), Romão et al. (2018) and Ugolini et al. (2020), population, GDP, public facilities, fixed-asset investment in urban appearance, environmental sanitation, and the number of high-speed railway lines passing through a city are used as control variables. We also control for the influence of the COVID-19 pandemic by including the number of confirmed cases as an additional control variable.

Model specification

The baseline model

The key question of this paper is to investigate the impact of digital travel vouchers on tourist inflows. The issuance of travel vouchers, as a means of stimulus policy to the tourism industry, is expected to boost tourist demand to the issuance city. Consequently, the higher the value of vouchers issued, the more tourist inflow is expected. Typical OLS regression may not be appropriate as it requires homoscedastic errors (Larch et al. 2019). This is especially relevant to our case. Beine et al. (2016) recommend using the Poisson Pseudo-maximum Likelihood (PPML) model to study questions related to tourist flows. The PPML model with region-by-time dummies carefully controls for heteroscedasticity, measurement errors and zero bilateral flows. Thus, we use the PPML model here to estimate the effects of digital travel vouchers on tourist inflows.

The baseline model here is given in Eq. (1):

$${{Tour}{{\_}}{Arrival}}_{{it}}=\exp \left(\alpha {\mathrm{ln}{VALUE}}_{{it}}+\varGamma {\mathrm{ln}{Control}}_{{it}}+{\eta }_{i}+{\nu }_{t}\right)\times {\varepsilon }_{{it}}$$
(1)

where \({{Tour}{{\_}}{Arrival}}_{{it}}\) is the tourist inflow of city i in month t; the key independent variable \({\mathrm{ln}{VALUE}}_{{it}}\) represents the log-transformed value of the vouchers of city i in month t. Here, a positive value of α is expected. \({\mathrm{ln}{Control}}_{{it}}\) consists of six control variables that may affect tourism development, all in natural logarithm; \({\eta }_{i}\) and \({\nu }_{t}\) are the province and month fixed effects, respectively; \({\varepsilon }_{{it}}\) is the random error term. Detailed information of the variables used here is given in Table 1.

Table 1 Variable description and data sources.
Full size table

In addition to the value of digital travel vouchers, the decision to issue or not issue these vouchers can be considered as a signal, leading to changes in the attitude of potential tourists. According to the TPB framework, issuing vouchers should boost behavioral intention and eventually lead to actual behavior. In other words, cities issuing vouchers should expect to experience more tourist inflows compared to cities not issuing vouchers and relative to their own situations before issuing the vouchers. This allows us to set up a staggered difference-in-difference (DID) model to evaluate the effects of issuing digital travel vouchers.

Further analysis: the extended model

Besides the question of whether and how much the issuance of digital travel vouchers per se impacts tourism, we are interested in several additional practical aspects. Unlike paper vouchers, digital travel vouchers are issued online, so the first issue of interest is where to issue the vouchers. In practice, there are two main ways to issue these vouchers: one is on larger platforms such as Alipay, Ctrip and other mobile apps, and the other is through smaller WeChat Mini Apps. The latter often involves more steps and has less publicity compared to those on the larger platforms. Generally, we expect to observe stronger effects on attracting tourist inflows if the vouchers are issued on larger platforms.

The second influential factor is the frequency of issuance. According to the TPB, frequent issuance of vouchers can reinforce perceived behavioral control and boost the behavioral intention of tourists, which may lead to higher chance of them travelling to the issuing city. In our sample of cities issuing digital travel vouchers, the average frequency of issuance is 2.347 and the highest is 10 times over the entire sample period. A simple correlation between the frequency of issuance and tourist inflow is 0.108, indicating a possible stronger effect when cities issue vouchers more frequently.

Given the digital nature of the vouchers interested in this paper, we also argue that the effectiveness of the vouchers may be affected by the regional development of digital inclusiveness. A higher level of digital inclusiveness means more people have access to resources and services. The mechanism here is similar to the impacts of infrastructure development but with a greater emphasis on digital development. We use the number of Internet broadband access users as a proxy for a city’s Internet penetration or digital inclusiveness and expect to observe higher effectiveness of vouchers in cities with broader inclusiveness.

These hypotheses can be evaluated in the extended model with interaction terms, such as:

$${{Tour}{{\_}}{Arrival}}_{{it}}=\exp \left(\beta {\mathrm{ln}{VALUE}}_{{it}}+\gamma {{Factor}}_{{it}}+\,\delta {\mathrm{ln}{VALUE}\times {Factor}}_{{it}}+\varGamma {\mathrm{ln}{Control}}_{{it}}+{\eta }_{i}+{\nu }_{t}\right)\times {\varepsilon }_{{it}}$$
(2)

where \({{Factor}}_{{it}}\) represents the above-mentioned factors in city i, month t. To ensure the key coefficients are comparable in Eqs. (1) and (2), we use demeaned terms of \({\mathrm{ln}{VALUE}}_{{it}}\) and \({{Factor}}_{{it}}\), denoted by \({C\_}{\mathrm{ln}{VALUE}}_{{it}}\) and \({C\_}{{Factor}}_{{it}}\), in Eq. (2), following Balli and Sørensen (2013).

The spatial spillover of digital travel vouchers

Following the discussions above, the effect of digital travel vouchers may have spatial spillovers. In other words, the issuance of vouchers in one city may also affect tourist flows in neighboring cities. To investigate this possibility, we employ typical spatial regression models (i.e., the Spatial Durbin Model) in the empirical analysis.

The first step of this part is to test whether there exists any spatial autocorrelation among sample cities’ tourist inflows, which is a prerequisite for spatial econometric analysis. We conduct Moran’s test for spatial autocorrelation using a spatial adjacency weight matrix to test if tourist inflows exhibit a clustered pattern. In addition to the typical geographical adjacency weight matrix, other factors, such as economic development and social environment, may interact with each other. For instance, cities with similar levels of economic development can better utilize resources and increase returns to scale (Khan et al. 2019). Therefore, a nested weight matrix (e.g., Gu et al. 2022) will be used as an alternative measure.

A Spatial Durbin Model is then constructed to test if there exists any potential spillover effect of digital travel vouchers. As there is no Poisson Pseudo-maximum Likelihood model extension of the Spatial Durbin Model, this study takes the natural logarithm of the dependent variable and constructs a log-linear model to examine the spatial spillover effect. The model is specified as:

$${\mathrm{ln}{Tour}{{\_}}{Arrival}}_{{it}}=\rho \mathop{\sum }\limits_{j=1}^{N}{w}_{{ij}}\,{\mathrm{ln}{Tour}{{\_}}{Arrival}}_{{jt}}+{\alpha }_{1}{\mathrm{ln}{VALUE}}_{{it}}+\;{\varGamma }_{1}\mathop{\sum }\limits_{j=1}^{N}{w}_{{ij}}{\mathrm{ln}{VALUE}}_{{jt}}+{\varGamma }_{2}{\mathrm{ln}{Control}}_{{it}}+{\eta }_{i}+{\nu }_{t}+{\varepsilon }_{{it}}$$
(3)

where \({\mathrm{ln}{Tourism}}_{{it}}\) and \({\mathrm{ln}{Tourism}}_{{jt}}\) represent the logarithmic transformed values of the tourist inflows of cities i and j in month t, respectively; \({\mathrm{ln}{VALUE}}_{{it}}\) and \({\mathrm{ln}{VALUE}}_{{jt}}\) are the key independent variables, which are the log-transformed values of the vouchers issued in cities i and j in month t, respectively. \({\mathrm{ln}{Control}}_{{it}}\) is a vector of log-transformed control variables; \({\eta }_{i}\) and \({\nu }_{t}\) are the province and month fixed effects, respectively; \({\varepsilon }_{{it}}\) is the random error term.

\({w}_{{ij}}\) is the spatial weight matrix, which is normally defined as a geographical adjacency matrix (\({W}_{{ij}}^{a}\)) among the sample cities. The nested weight matrix (\({W}_{{ij}}^{{de}}\)) combines the inverse distance spatial weight matrix (\({W}_{{ij}}^{d}\)) and the economic weight matrix (\({W}_{{ij}}^{e}\)) to incorporate economic factors. It is defined as:

$${W}_{{ij}}^{{de}}=\left\{\begin{array}{l}{W}_{{ij}}^{d}\times {W}_{{ij}}^{e},\;i\,\ne\, j\\ 0\,,\qquad\qquad i=j\end{array}\right.$$
(4)

where \({W}_{{ij}}^{d}\) represents the inverse of the distance between city i and city j, and \({W}_{{ij}}^{e}\) represents the inverse of the absolute difference in GDP per capita between city i and city j.

Empirical results

Descriptive statistics

The descriptive statistics of the main variables are given in Table 2. Panel A presents the information about cities that issued digital travel vouchers, while Panel B presents information for cities that did not issue vouchers during the same sample period. Among all the cities in the sample, 69 issued vouchers, constituting approximately one-fifth of the total cities. The average influx of tourists into cities in Panel A is much higher than that in Panel B. On average, the difference between issuing cities and non-issuing cities is 0.128, indicating the potential impact of digital travel vouchers in attracting tourists and boosting local tourism.

Table 2 Descriptive statistics.
Full size table

In Table 2, it is shown that the average annual GDP of cities in Panel A is higher than that of cities in Panel B. These findings jointly suggest that governments in economically advantaged areas are more likely to support the tourism industry through issuing digital travel vouchers. The last row shows that, on average, cities issuing vouchers have a higher number of confirmed COVID cases compared to their counterparts in Panel B. This might imply that cities severely affected by the COVID-19 pandemic are more inclined to issue digital vouchers to revive their tourism sector.

The correlation matrix of the main variables is presented in Table 3 below. Clearly, some variables exhibit reasonably high correlations. For example, the values of issuance, publicity, and availability are highly correlated, which is not entirely surprising. It is often the case that a large volume of issuance attracts more attention and involves more resources from businesses. These two variables will be used for robustness analysis. To address concerns regarding multicollinearity, we conduct typical VIF tests, which reveal no significant issues of collinearity in the main model. The results are not reported here due to space constraints but will be available upon request.

Table 3 Correlation matrix for main variables.
Full size table

Baseline model results: the impacts of digital travel vouchers

Following the discussions above, we first investigate whether digital travel vouchers issuance and the associated values can affect tourist inflows. The results are reported in Table 4. This table presents the findings using six different specifications and sampling frames. In all models, we control for monthly and provincial fixed-effects. Along with the full sample analysis, we have included three sub-sample regression results: one excluding extreme months, another excluding major public holidays, and a third considering only the cities that issued digital travel vouchers throughout the entire sample period.

Table 4 The impacts of digital travel vouchers on tourist inflows.
Full size table

Based on the results presented in Table 4, clear evidence emerges indicating that issuing digital travel vouchers can significantly increase tourist inflows. Controlling for city-specific factors, the effect of issuing digital travel vouchers is 0.078 according to the DID regression results, and this result is statistically significant at the 5% level of significance. The issuance value also shows significant results (models (1) and (2)). These impacts are generally consistent when different samples are considered. Therefore, Hypothesis 1 is verified.

Regarding the control variables, the results generally align with intuition and economic logic. Larger cities, in terms of population, tend to attract more tourists, as do economically more developed cities. This trend extends to factors such as infrastructure and urban environment. Notably, the number of confirmed COVID-19 cases exerts a significant and negative impact on tourist inflows, indicating that a more severe wave of the pandemic reduces tourist visits and accelerates the deterioration of local tourism. It is interesting to observe that the contribution of high-speed railways is not significant. This finding is consistent with Gao et al.��s (2019) findings, suggesting that the effects of high-speed rail on tourism are not always positive, especially when factors such as population and other developmental considerations are controlled.

Extended models: factors that determine the vouchers’ effectiveness

While the above analysis shows the positive and significant impact of digital travel vouchers on promoting local tourism, our interest lies in the subsequent question of how to enhance the effectiveness of their issuance. To explore this, we extend the previous model (Eq. (2)) by incorporating issuing characteristics and assess the effects through interaction terms. The results are presented in Table 5. It should be noted that the DID model only considers the first issuance as the treatment; therefore, these additional characteristics could not be effectively utilized. We thus focus on the ‘value’ for the following study.

Table 5 Results of the extended models.
Full size table

In this section, three factors are considered. First, we demonstrate the significance of the choice of issuing platforms. Selecting larger platforms can significantly enhance the effects of issuance and increase the value effect. Larger platforms are generally more user-friendly and credible. According to the TPB model, the creditability of larger issuing platforms should boost behavioral intentions, thereby increasing the likelihood of visitors choosing issuing cities (Cabiddu et al. 2013). The results in model (1) of Table 5 confirm this hypothesis. Digital travel vouchers issued on larger platforms are generally associated with higher tourist inflows and can also amplify the value effect, as indicated by the significant interaction terms.

The second factor considered in the extended model is frequency; that is, how many times a city issues digital travel vouchers during the studied period. Repeated issuance of vouchers can reinforce consumers’ behavioral beliefs, thereby increasing the likelihood of visits. From the regression results in Column (2) of Table 5, the coefficient on frequency is positive and significant, confirming the projection that repeated issuance of digital travel vouchers can effectively attract more tourists. An intriguing aspect is the negative interaction terms with value. This result suggests that issuing frequency can diminish the value effect. In other words, issuing digital travel vouchers in smaller values many times could generate a stronger attraction for tourists.

To highlight the digital nature of the vouchers studied in this paper, we also consider regional internet penetration as a proxy for digital inclusiveness. Following the discussion in Section 3, we would expect to observe stronger effects of issuing digital travel vouchers in cities with higher levels of digital inclusiveness. Better digital development in a city makes the digital travel vouchers easier to use. Moreover, within the TPB framework, this could also be interpreted as stronger normative beliefs, leading to higher behavioral intention (Bosnjak et al. 2020). From Column (3), although internet penetration is statistically insignificant in the regression, the interaction term is positive and significant at the 5% level of significance. To summarize, H2 holds.

City-level heterogeneity analysis

This section examines whether cross-city heterogeneities, namely, city reputation and the level of a city’s digital financial inclusion, affect the effectiveness of digital travel vouchers. First, a dummy variable is created (Reputation), taking value 1 for well-known tourist cities and 0 otherwise according to the list of National Comprehensive Tourism Demonstration Zones released by the Ministry of Culture and Tourism of China. For digital finance development, we adopt the sub-indices—the coverage breadth and usage depth of digital finance—released by the Digital Finance Research Center of Peking University. The breadth of digital finance is mainly reflected in the coverage of digital financial accounts, payment business, and money fund business. The depth of digital finance is assessed based on the credit business, insurance business, investment business, etc.

Interaction terms are used in the regressions to reflect heterogenous effects, and the results are reported in Table 6. In column (1), although the coefficient on city reputation is positive, but it is not statistically significant, meaning well-known cities do not have advantage in attracting tourists. Due to the influence of the pandemic, people are less enthusiastic about travelling, applying to all type of cities (Jiang et al. 2022). The interaction term between city reputation and voucher issuance value is however, positive and significant at 10% level, implying that the popularity of voucher campaigns can be amplified by city reputation.

Table 6 Heterogeneity analysis (PPML model).
Full size table

Columns (2) and (3) in Table 6 report the influences of city-level digital financial inclusion. Both the breadth of digital finance coverage and its interaction term with the value of vouchers have positive impacts on tourist inflows, and they are significant at the 1% level of significance. As better digital finance coverage provides faster and easier access to digital financial products and services, it creates great convenience for customers using online travelling tools, meanwhile boosting the chance of using digital travel vouchers. In contrast, the depth of digital finance usage makes an insignificant contribution, though its interaction with value is positive and statistically significant.

These results emphasize the importance of the to-customer dimension of digital finance, rather than the to-business dimension, in promoting local tourism. Consumers tend to prefer cities with wide coverage of digital finance applications, while the extent of application by business-end users is not a major concern for them. We also notice that both interaction terms exert significant and positive impacts on tourist inflows at the 1% level of significance. This indicates that the effects of vouchers can be enhanced by both the breadth and depth of urban digital finance, and vice versa. Cities with developed digital finance technology are more experienced in and capable of advocating and facilitating the utilization of vouchers through diverse channels, thus making their vouchers more convenient, appealing, and effective.

Spatial spillover effect of digital travel vouchers

In this section, we investigate the potential spatial spillover effect of digital travel vouchers from the perspective of city clusters. The results of the Moran index test are presented in Table 7. The Moran index of tourist inflows for each month during the sample period is positive at the 1% level of significance. These findings strongly reject the null hypothesis of no spatial autocorrelation, indicating that our tourist inflow data exhibit a pattern of spatial autocorrelation and thus meet the prerequisite for spatial econometric analysis.

Table 7 Results of the Moran’s I test: September 2020 to December 2021.
Full size table

Then, we apply the Spatial Durbin Model based on the spatial weight matrix (W) to test the spillover effects of digital travel vouchers and other explanatory variables in model (3). The results reported in Tables 8 and 9 include results using the geographical adjacency matrix (\({W}_{{ij}}^{a}\)) and the nested weight matrix (\({W}_{{ij}}^{{de}}\)), respectively. First, in both cases, the results of R2 and Sigma2 suggest that the models are well-fitted. The magnitude of rho generally implies significant and positive spatial spillover effects.

Table 8 The spatial spillover effects: using the geographical adjacency matrix.
Full size table
Table 9 The spatial spillover effect: using the nested weight matrix.
Full size table

On average, the effect of the spatial lagged voucher value in one city (W×lnVALUE) on tourist inflows of its adjacent cities is shown to be significant and positive at the 1% level of significance. This indicates that the issuance of vouchers in one city contributes to boosting tourist inflows in neighboring cities, confirming our Hypothesis 3 that there exists a spillover effect of the benefits brought by the voucher campaigns based on our sample data. The direct effect refers to the change in tourist inflows caused by local issuance of vouchers, while indirect effect refers to the change in tourist inflows caused by vouchers issued by neighboring cities.

It is interesting to note that the spatial lagged city size, in terms of population, exerts significant and negative impacts on attracting tourists, meaning that a larger size of one city reduces tourist inflows to its neighboring cities. The negative coefficient of W×lnPOP directly reflects the pattern of tourist mobility in China and confirms the siphon effect discussed in the existing literature (Wang et al. 2020). Similar results are also found in infrastructure development, which also has negative impacts on neighboring cities. In general, the results using nested weight matrix (Table 9) are consistent.

Robustness checks

To further validate the results obtained in this paper, we conduct a series of robustness checks. First, we replace the value of issuance with alternative factors, namely, publicity and availability. Publicity of the vouchers can be proxied by the number of news items related to the issuance of the vouchers. We count the number of news articles published and included in Baidu related to “Digital tourism vouchers of city i in month t”. For availability, we use the number of businesses involved in a digital travel voucher campaign as a proxy. The more businesses involved, the more resources available for tourists.

Second, we present the parallel trend test and placebo test results for the DID regressions. We also use matched results to reproduce the DID regression results to address potential bias from non-random allocation or influence from unobserved factors. Specifically, we employ the nearest neighbor sampling method for matching. In addition, recognizing that the DID setup of this model might not be strictly satisfactory as cities may issue digital vouchers multiple times, we apply the approach suggested by Callaway and Sant’Anna (2021) as an additional robustness check (denoted as CSDID).

Third, to address potential endogeneity in the empirical analysis, we opt to use instrumental variable in a two-stage least squared (2SLS) model. The one-order lagged value of digital travel vouchers issued by other cities (excluding the studied city) in the same province is used as the instrumental variable. The rationale behind this is that neighboring cities, especially within the same provinces, could serve as examples or exert pressure on city governments to react similarly. Obviously, this decision to issue digital travel vouchers by other cities in the previous period would not be affected by tourist inflows in the studied city of the current period.

The results of these robustness checks are presented in Table 10. The results are generally consistent with those presented in the baseline models. Both publicity and availability show positive and significant effects at the 1% level of significance. The coefficients of the PSM-Staggered-DID and CSDID all support the same conclusions as before. Two IV regressions (2SLS and GMM) also point to the positive and significant impact of digital travel vouchers on tourist inflows. In both regressions, the KP LM test statistics and CD Wald F-statistics are significant, suggesting the appropriateness of the choice of instrumental variables.

Table 10 Robustness checks.
Full size table

Parallel trend tests

To confirm the validity of our DID regression results, the parallel trend test is presented in Fig. 3. In the parallel trend result, the difference in tourist inflows between the control and treatment groups before the issuance remains insignificant, but significant differences are found after the issuance of digital travel vouchers.

Fig. 3
figure 3

The Parallel trend test result.

Full size image

Placebo test

The results of the placebo test are shown in Fig. 4, where the horizontal axis represents the magnitudes of the estimated coefficients of the “pseudo policy dummy variable”, and the vertical axis represents the distribution density and p value. The curve is the kernel density distribution of the estimated coefficients, and the blue dots indicate the corresponding p values of these estimated coefficients. The vertical dotted line marks the true estimated coefficient of the key independent variable. The dashed horizontal line marks the 10% level of significance. Most estimated coefficients of the “pseudo policy dummy variable” are close to zero, and most of the corresponding p values are greater than 10%, indicating that the estimated results are not significant. This ensures that the probability of accidentally obtaining these estimated results is extremely low.

Fig. 4
figure 4

The Placebo test result.

Full size image

Conclusions and policy suggestions

A critical concern of many governments during the post-pandemic era is how to restore the severely impacted tourism industry and attract tourist inflows back. This unprecedented public health crisis has far-reaching impacts, involving a combination of social, economic, and psychological shocks to tourists, making the issue much more complicated to deal with. Despite numerous efforts made by the international community, the recover process remains slow. In light of the TPB theory used in related literature, this study empirically explores the issue in China, with a special focus on the issuance of digital travel vouchers. We collect tourist inflow data for 306 Chinese cities between September 2020 and December 2021 and investigate whether and how the issuance of digital travel vouchers affects tourist inflows.

We first confirm the positive role of digital travel vouchers, showing that their issuance can help bring tourists back. This conclusion remains robust after a series of robustness checks. In the extended model, we explore three characteristics closely related to the issuance of vouchers, especially concerning their digital nature. The results show that using larger and more reliable digital platforms leads to a more effective role of the vouchers. Repeatedly issuing vouchers can reinforce the behavioral intention, making the vouchers more effective. Using the internet penetration rate at the city level as a proxy for digital inclusiveness confirms that a higher penetration rate is associated with more effective issuance of vouchers. With the rapid progress of information technology and the digital economy, utilizing innovative digital methods to issue vouchers could be more effective. However, understanding the new features associated with digital transformation requires further study.

Individual cities’ characteristics may also play a role in affecting the vouchers’ effectiveness. We find evidence of the strengthening effect of city reputation on vouchers, though statistically marginal. The same voucher campaign should work better in more famous cities relative to other cities. Besides, more widespread coverage of digital finance directly increases tourist inflows. Wide digital finance coverage brings fast and easy access to related products and services, especially those in digital forms. The depth of digital finance usage, however, does not directly promote tourist inflows, as travelers care much less about how deep digital technology penetrates a destination city’s financial industries, relative to how wide and fast can the digital travel tools be applied there. Nevertheless, both the coverage and depth of digital finance enhance the effects of vouchers in attracting tourists.

In the spatial econometric analysis of the issuance effect, we confirm the existence of spatial spillover. Issuing digital travel vouchers not only affects the tourist inflows in the issuing city but also has positive impacts on the tourism industry of neighboring cities. These spatial effects suggest the need to explore regional synthetic policy designs.

Beyond employing web-crawling techniques to gather and study pandemic-era tourism data, this study contributes to the literature by providing important evidence supporting the TPB theory, which has been used recently in understanding tourist behaviors (e.g., Liu et al. 2021; Soliman, 2021). And this study explores the effect and details of digital voucher issuance, enriching the theoretical framework of the research on consumption vouchers. Policy designs to counter the negative impacts of the pandemic should not only focus on economic incentives but also take psychological factors and consumer behaviors into consideration.

This study offers several practical implications. First, based on the TPB theory, governments can play a more active role in reviving the troubled tourism sector. By issuing digital travel vouchers and providing economic incentives, they can attract more travelers, thereby helping the local tourism sector recover from the pandemic. Second, digital travel vouchers can be an effective tool, but careful attention is needed to make the issuance more effective. We demonstrate that local digital inclusiveness, issuance platforms, the frequency of issuance, and internet penetration all matter for their effectiveness. Authorities should consider these factors when designing their vouchers. Specifically, before issuing digital travel vouchers, local governments need to cooperate with larger travel or consumption platforms. They should also increase the frequency of voucher issuance and ensure that local attractions, hotels, and other tourism products have online purchasing options and convenient offline redemption channels.

Third, recognizing the potential spatial spillover effects is also critical for policy effectiveness. This necessitates the design of a regional synthetic plan instead of isolated issuance in each city. Supportive policies should be coordinated across local governments to achieve optimal effects. For instance, cities can collaborate to issue customized digital travel vouchers, thereby expanding the visibility and effectiveness of the vouchers. This can enhance the region’s attractiveness and foster a tourism consumption cluster. Lastly, for tourism industry stakeholders, the effectiveness of digital travel vouchers depends on the issuance forms. Therefore, when issuing vouchers, tourism industry stakeholders should use sound marketing methods, increase the publicity of the vouchers, and change the issuance mode to maintain the consumption enthusiasm of potential tourists.

In the current era of fast-developing digital economies worldwide, China’s evidence can offer valuable insights to other countries. The suggestions in this article regarding issuing platforms, frequency, and the extent of internet penetration provide empirical references for the issuance of vouchers in other regions worldwide. Additionally, the heterogeneity analysis in our study helps explain the differences in the effectiveness of travel vouchers between cities.

However, this paper has several limitations. Due to data availability, we use travel inflows as a proxy for tourist inflows for each city, which could introduce bias into the estimated results. Although we have tried our best to address these concerns through a series of robustness checks, future investigation using alternative available data sources could be helpful. For example, information of other types of vouchers issued to boost consumption may have an impact, but it is challenging to gather sufficiently representative data on that. In addition, the voucher data utilized in our analysis is the amount issued by local governments. However, the dissemination of digital travel vouchers across multiple mobile applications introduces fragmentation, and the associated payment details are safeguarded under personal privacy regulations. Consequently, we are unable to procure precise information regarding the utilization of these vouchers, thus constraining our ability to delve deeper into their operational efficacy. Third, our discussion is limited by the distinctions between different types of cities. Although we have separately explored the effectiveness of digital travel vouchers from the perspectives of voucher issuance and the characteristics of the cities themselves, there is still room for further discussion in characterizing city features. Last, it is necessary to acknowledge that the issue transcends the borders of China. Various nations may harbor distinct considerations and novel characteristics. Hence, from a broader perspective, conducting cross-country comparisons would significantly enrich the scope of this study.

The digital revolution presents an opportunity to reevaluate conventional theories pertaining to consumer behavior. More efforts are needed to gain a better understanding of these new features, which should be investigated in future research. In our subsequent research endeavors, we therefore intend to investigate the various forms and effects of travel vouchers across diverse nations worldwide. This investigation will entail comparative analyses to discern both similarities and disparities, with a particular focus on exploring the underlying drivers stemming from consumer psychology and socio-cultural dynamics. By identifying the nuanced distinctions, we aspire to formulate recommendations for the issuance of travel vouchers tailored to the unique characteristics of each respective country.