Digital financial inclusion and domestic tourism demand: Through the lens of spatial spillover

Abstract

Domestic tourism demand has dominated the global tourism market and has been more resilient during the COVID-19 pandemic; yet, it is under-studied compared to international tourism demand. Digital financial inclusion (DFI) enables the provision of formal financial products and services to small and medium-sized tourism enterprises and tourists through cost-effective digital means, potentially boosting domestic tourism demand. Integrating micro- and macro-level data from 335 prefecture-level regions in China, we investigate the impact of DFI on domestic tourism demand using spatial panel models. Our analysis reveals significant and positive effects, both direct and spillover, of DFI on domestic tourism demand within a region and across regions. Notably, among the three dimensions of DFI, the usage depth exhibits the most substantial spillover effects. Furthermore, our findings also highlight the crucial role of DFI in internalizing domestic tourism revenues. Our study provides practical implications for sustainable domestic tourism development in the digital era.

Keywords

digital financial inclusion domestic tourism demand spatial spillover spatial economics sustainable tourism development

Introduction

As the foundation of the national economy, the tourism industry plays an enormous role in promoting economic growth and generating employment opportunities. Before the COVID-19 pandemic, the tourism industry accounts 10.3% of global GDP (equivalent to USD 9.6 trillion) and supports 333 million jobs (WTTC, 2022). While the pandemic and economic downturn severely impacted tourism in recent years, global tourism demand is expected to recover strongly in 2023, driven by the attainment of herd immunity (Li et al., 2023; UNWTO, 2023). Notably, domestic tourism has traditionally dominated global tourism market even before the pandemic (Adamiak and Szyda, 2022).¹ Moreover, domestic tourism has shown faster resilience and recovery compared to international tourism in the post-pandemic period (UNWTO, 2020). Therefore, for countries like China, which possess abundant tourism resources and a large number of tourists, fostering and sustaining a strong domestic tourism demand is of paramount importance for ensuring the long-term sustainability of the tourism sector.²

Financial development has been widely recognized as a significant driver of tourism demand (Churchill et al., 2022; Katircioglu et al., 2018; Khanna and Sharma, 2021; Menegaki and Tiwari, 2022). In countries or regions with higher levels of financial development, the impact on tourism demand is twofold. First, tourism-related enterprises are more likely to access financial support and standardized governance services, fostering innovation in tourism products and business models (Khanna and Sharma, 2021). Second, tourists can alleviate household credit constraints and achieve smoother consumption through more abundant financial products, leading to an increase in tourism demand (Yang and Zhang, 2022). Particularly, the majority of tourism enterprises are small and medium-sized enterprises (SMEs) (Carrillo-Hidalgo and Pulido-Fernández, 2019), and domestic tourism is often preferred by low- and middle-income groups. In this context, financial inclusion, which aims to provide affordable formal financial services and products to companies and individuals sustainably (Célerier and Matray, 2019; Yang and Zhang, 2022), emerges as a crucial aspect of financial development in the literature (Hu et al., 2023; International Monetary Fund, 2023; Yang and Zhang, 2022).

However, existing literature towards financial development or financial inclusion and tourism demand primarily focuses on international tourism (Gopalan and Khalid, 2022, 2023; Khanna and Sharma, 2021; Shi et al., 2021). This narrow focus can be attributed to the limited availability of indicators or proxies for measuring financial development or financial inclusion at subnational levels. The emergence of digital financial platforms and the availability of micro-level data of digital financial inclusion (DFI) offer a promising solution to this data and research dilemma (Yang and Zhang, 2022). DFI, in particular, is more likely to achieve broad financial inclusion as it eliminates the reliance on physical network associated with traditional financial system and hence significantly lowers the costs (Demir et al., 2022). In the current era of booming digital economy and FinTech, DFI has gradually surpassed traditional financial inclusion. For instance, in China, the number of ATMs and bank branches, which are key indicators of traditional financial inclusion, has decreased significantly.³ It is therefore essential to consider digital rather than solely focusing on traditional financial inclusion when examining its relationship with tourism demand.

This study intends to investigate the effect of DFI on domestic tourism demand using micro- and macro-level data in China. China is a good case due to its status as the largest domestic tourism market globally and its notable advancements in DFI recently (Lin et al., 2022; Yang and Zhang, 2022). To assess the level of DFI, we utilize the DFI index, jointly developed by Peking University and Ant Group based on the massive user data from Alipay.⁴ Further details regarding the DFI index can be found in the third section. In addition, we adopt data measuring domestic tourism demand, including domestic tourist arrivals and tourism revenues, as well as other relevant data for 335 prefecture-level regions in China from 2011 to 2019.

Previous studies have documented significant spatial spillover effects on domestic tourism demand in China (see, e.g., Yang and Wong, 2012; Yang and Fik, 2014; Zhang et al., 2011). Simultaneously, the promotion and development of DFI also exhibit spatial dependence due to demonstration and competition effects among neighboring regions (Guo et al., 2020; Zhang et al., 2023a). To mitigate bias and obtain reliable results, it is necessary to apply spatial econometric modeling to measure the impact of DFI on domestic tourism demand. This study employs the approach presented by Gunter et al. (2019) to rigorously determine appropriate spatial econometric model and other key details. Furthermore, we explore the direct, indirect, and total effects of different dimensions of DFI, namely, coverage breadth, usage depth, and digitization level, on domestic tourism demand by taking them as surrogate variables. To ensure the robustness of our findings, a comprehensive set of robustness checks including substituting dependent variables, substituting spatial weight matrices, adjusting sample period, and lagging DFI indexes, are also conducted.

Our study contributes to the literature linking financial development and tourism demand. First, this is the first study to highlight the positive impact of digital financial inclusion on domestic tourism demand, to the best of our knowledge. While previous studies on tourism demand have examined various determinants of domestic tourism demand, such as meteorological factors (Otero-Giráldez et al., 2012), income (Yang et al., 2014), vacation policy (Zhang et al., 2016), and transportation (Tsui, 2017), financial inclusion, especially digital financial inclusion, has been largely ignored. Second, we employ a DFI index based on extensive micro-level data and a multi-tiered index system. In contrast to the isolated DFI indicators commonly used in a limited number of previous studies (Gopalan and Khalid, 2022, 2023), this index provides a more comprehensive reflection of the complex multidimensional characteristics of DFI. Third, our study offers novel insights into the relationship between DFI and domestic tourism demand from a spatial spillover perspective. The findings of this research emphasize the significance of DFI for different stakeholders, including tourism practitioners and governments, to prioritize the importance of DFI and fully internalize the spillover effects of enhanced DFI in neighboring regions.

The remainder of the article is organized as follows. The second section reviews literature on financial development and financial inclusion promoting tourism demand, as well as spatial spillover in tourism demand studies. The third section introduces related data and methodology involved in this study. The fourth section analyzes empirical estimation results and performs robustness checks. The last section discusses and concludes this study with suggestions for future research.

Literature review

In this section, we provide a comprehensive review of advances in financial inclusion, with a special focus on DFI, in the tourism and general (non-tourism) literature, as well as research on spatial spillovers in tourism demand studies, leading to the formulation of our research hypotheses.

Financial inclusion and tourism demand

A review of the nexus between financial inclusion and tourism demand could be traced to the relatively broader literature linking financial development and tourism demand. According to the World Bank, financial development can be defined as the reduction of information, enforcement, and transaction costs in the financials through the presence of financial institutions, intermediaries, and markets (World Bank, 2015). Numerous researches at the single-country and regional levels have demonstrated the positive impact of financial development on economic growth and tourism demand, highlighting the corresponding feedback effect (Cannonier and Burke, 2017; Churchill et al., 2022; Katircioglu et al., 2018; Menegaki and Tiwari, 2022). Notably, Khanna and Sharma (2021) present the solid cross-country evidence to assess the influence of financial development on tourism demand using a sample of 207 countries.

In recent years, the view of financial inclusion as an integral dimension of financial development has been given due attention (International Monetary Fund, 2023). Especially for the tourism industry, where most of the enterprises are SMEs, financial inclusion calls for financial development to bring financial support to many SMEs instead of just concentrating on the top enterprises (Carrillo-Hidalgo and Pulido-Fernández, 2019). Only a handful of initial attempts have been made to investigate the role of financial inclusion in boosting tourism demand. For example, Shi et al. (2021) first formulate the impact of financial inclusion on tourism demand and compare the differences in this impact between developed and emerging economies. By contrast, Gopalan and Khalid (2022) and Gopalan and Khalid (2023) focus solely on emerging economies. Nonetheless, these above studies mainly consider international tourism demand including inbound tourism and outbound tourism, ignoring the huge domestic tourism demand. Considering the actual beneficiaries of financial inclusion, it seems that domestic tourism demand is more affected by financial inclusion than international tourism demand.

It is widely believed that fintech represented by digital financial services has significantly improved the accessibility and affordability of financial services, thereby driving the development of financial inclusion (Demir et al., 2022; Hu et al., 2023; Yang and Zhang, 2022). Recent studies have identified and confirmed the positive effects of such DFI on SMEs financing (Babilla, 2023; Hu et al., 2023; Zhang et al., 2023b), (household) entrepreneurship (Gao et al., 2022; Wu and Wu, 2023), household income (Zhang et al., 2020), and household consumption (Yang and Zhang, 2022). Back to tourism industry, DFI is expected to benefit both tourism SEMs and domestic tourists to promote domestic tourism demand. On the one hand, DFI extends formal bank credit to tourism SMEs lacking collateral assets through the utilization of big data credit investigation, which benefits tourism entrepreneurial activities and tourism product innovation (e.g., online travel booking), and consequently stimulate tourism demand (Gopalan and Khalid, 2022). Digital platform giants often start by offering digital financial products, followed by a package of digital transformation services. The large number of consumers enjoying digital financial products also drive SMEs to embrace the trend of digital transformation. Therefore, DFI could support the digital transformation of the tourism SMEs, enhancing management and operational efficiency (Guo et al., 2023). On the other hand, DFI smooths the flow of funds between investments, savings, and payments, which facilitates the consumption of tourists during their travel activities. DFI also potentially increases tourists’ leisure time by saving their shopping time through convenient payments.⁵ Furthermore, tourists and tourism SMEs benefit from convenient, diverse, and transparent insurance products and services made possible through DFI. In view of the inherent variability and fragility of tourism demand, tourists and tourism enterprises can use insurance to dilute the risks in travel and business operations (Gopalan and Khalid, 2022), which also contributes to the sustainable development of domestic tourism demand. Considering that international tourism is expensive and mainly operated by large enterprises, the DFI’s mechanism for promoting domestic tourism demand is unsuitable for international tourism demand. Thus, the following hypothesis is proposed:

Hypothesis 1

There is a positive and significant direct effect on domestic tourism demand with a region.

Spatial spillover in tourism demand studies

The spatial spillover effect refers to the indirect or unintentional influence of tourism demand among neighboring regions (Jiao et al., 2021; Yang and Wong, 2012). Various factors contribute to tourism flows spillover, including productivity spillover, market access spillover, joint promotion, negative events, and tourists with multi-destination travel plans, as summarized by Yang and Wong (2012). Failing to take spatial spillovers into account in modeling tourism demand may lead to biased and misleading coefficient estimates, which is not conducive to untangling the determinants affecting regional tourism demand growth (Yang and Fik, 2014). Pioneering studies by Yang and Wong (2012) and Yang and Fik (2014), have formulate the spatial auto-regressive models and spatial Durbin models to estimate the spatial spillover of tourist flows and their growths in China. On the basis of these studies, Zhou et al. (2017) and Tian et al. (2022) identify the spillover effects of attractions and transport improvement on China’s inbound and domestic tourism demand, respectively. Deng and Hu (2019) examine the spillover effects of outbound tourism demand from Chinese tourists to the “Silk Road” countries from the perspective of geographical and cultural proximity. Xu et al. (2020) reveals spatial heterogeneity of the direct and spillover effects of haze pollution on inbound tourism in the Central and Eastern China. Beyond China, scholars have also observed spatial spillovers in tourism demand of other countries, such as Italy and Spain, identifying specifications and the extent of spillovers of determinants (Alvarez-Diaz et al., 2020; Marrocu and Paci, 2013).

The above literature basically adopts the spatial econometric models to estimate the spatial spillover of tourism demand. Furthermore, Gap model (Li et al., 2011), distance model (Yang et al., 2017), and agent-based modeling (Li et al., 2021) are all used to capture this spatial spillover of tourism demand. In terms of the research geographical scale, aside from the common national, provincial, and city levels, the spillover effects of tourism demand at the level of attraction (Kim et al., 2022; Zhou et al., 2017), community (Li et al., 2022), and vacation town represented by Disney Resort (Weng et al., 2021) are investigated.

Although the promotion mechanism of tourism demand by financial inclusion has been preliminarily discussed in the literature (Gopalan and Khalid, 2022, 2023; Shi et al., 2021), little is known about the spatial interaction of financial inclusion in tourism industry. Given that DFI spreads across geographic boundaries but is also spatially dependent at region level (Guo et al., 2020; Yang and Zhang, 2022; Zhang et al., 2023a), exploring the effect of DFI on tourism demand from the lens of spatial spillover is helpful to better understand the dynamic interaction between DFI and domestic tourism demand. Therefore, we develop the following hypothesis:

Hypothesis 2

There is a positive and significant spatial spillover on domestic tourism demand across regions.

Data and methodology

Digital financial inclusion in China

The emerging DFI can effectively serve the underserved long-tail users among SMEs and individuals, who cannot be covered by the traditional financial system. DFI is already booming in China. The digital financial platforms represented by Alipay and WeChat Pay not only provide convenient payment services, but also extend their provision to credit, insurance, funds, and various other financial services. For example, Yu’ E Bao, a popular monetary fund launched by Alipay, allows fund holders to make payments at any time while enjoying the fund’s earnings. With 741 million fund holders in the middle of 2022, Yu’ E Bao stands as the largest financial inclusion product in money management in China and even in the world.

To describe the level of DFI in China, we adopt the DFI index developed by Guo et al. (2020). As illustrated in the Table 1, the DFI index system consists of three primary indexes, which are in turn composed of a series of secondary indexes. Specifically, the primary indexes include coverage breadth, usage depth, and digitization level. The coverage breadth reflects the penetration of DFI and is described by Alipay account coverage. The usage depth considers the use of payment, insurance, monetary fund, investment, credit, and credit investigation. The digitization level measures the mobility, benefits, credit level, and facilitation of DFI.

Table 1.

Digital financial inclusion index system.

Primary index	Secondary index	Specific indicators
Coverage breadth	Alipay account coverage	Number of Alipay accounts per 10,000 people
		Proportion of Alipay accounts bound with bank cards
		Average number of bank cards bound to each Alipay account
Usage depth	Payment	Average number of payments by Alipay accounts
		Average amount of payments by Alipay accounts
		Proportion of Alipay active users (active at least 50 times a year)
	Insurance	Number of insured per 10,000 Alipay accounts
		Average number of insurance for Alipay accounts
		Average amount of insurance for Alipay accounts
	Monetary fund	Average number of purchases of monetary fund on Alipay accounts
		Average amount of purchases of monetary fund on Alipay accounts
		Number of monetary fund purchases per 10,000 Alipay accounts
	Investment	Number of Internet investment participators per 10,000 Alipay accounts
		Average number of investments by Alipay accounts
		Average amount of investments by Alipay accounts
	Credit	Number of accounts with Internet consumer loans per 10,000 Alipay accounts
		Average number of loans for Alipay accounts
		Average amount of loans for Alipay accounts
		Number of accounts with Internet small business loans per 10,000 Alipay accounts
		Average number of loans for small business Alipay accounts
		Average amount of loans for small business Alipay accounts
	Credit investigation	Average number of natural person credit calls for Alipay accounts
	Credit investigation	Number of accounts with credit-based services per 10,000 Alipay accounts
Digitization level	Mobility	Proportion of number of mobile payment
	Mobility	Proportion of amount of mobile payment
	Benefits	Interest rates on loans for Alipay accounts
	Benefits	Interest rates on loans for small business Alipay accounts
	Credit level	Proportion of number of consumer credit services
		Proportion of amount of consumer credit services
		Proportion of number of deposit-free services
		Proportion of amount of deposit-free services
	Facilitation	Proportion of number of QR code payment for Alipay accounts
	Facilitation	Proportion of amount of QR code payment for Alipay accounts

Note: This index system is constructed by Guo et al. (2020).

The DFI total index is synthesized in the following three steps (Guo et al., 2020). First, all the specific indicators are dimensionless using the logarithmic efficiency coefficient method. Next, the variance coefficient method is employed to determine the weight of each specific indicator on the corresponding secondary index. Finally, the weights of the secondary and primary indexes on the upper level indexes are determined by the analytic hierarchy process.

Figures 1 and 2 display the spatial distribution of the DFI index in 2019 and the growth rate of the DFI index from 2011 to 2019, respectively. Figures 3 and 4 depict the spatial distribution of the Chinese domestic tourist arrivals in 2019 and the growth rate of the Chinese domestic tourist arrivals from 2011 to 2019, respectively. Darker colors in the figures indicate higher levels or growth rates of DFI or domestic tourism demand. There is conspicuous agglomeration and consistency in the spatial distribution of both the DFI and domestic tourism demand. Although the level is higher in the Eastern region, the growth rate is faster in the Midwest. This intuitive finding also implies that spatial rather than non-spatial models are more suitable for estimating coefficients.

Figure 1.

Digital financial inclusion index in China, 2019.

Figure 2.

Growth rate of digital financial inclusion index in China from 2011 to 2019.

Figure 3.

Domestic tourist arrivals in China, 2019.

Figure 4.

Growth rate of domestic tourist arrivals in China from 2011 to 2019.

Data collect and descriptions

This study uses panel data of 335 prefecture-level regions in Mainland China from 2011 to 2019 to establish the spatial panel models. Specifically, the dependent variable is ln arrival (the logarithm of domestic tourist arrivals, in 1000) or ln revenue (the logarithm of domestic tourism revenues, in million RMB). The data for the dependent variables are from the China Premium Database in CEIC Data (https://insights.ceicdata.com) and the statistical yearbooks of prefecture-level regions in China. The primary explanatory variables consist of DFI indexes, namely, ln DFI_total (the logarithm of DFI total index), ln DFI_breadth (the logarithm of DFI coverage breadth index), ln DFI_depth (the logarithm of DFI usage depth index), and ln DFI_level (the logarithm of DFI digitization level index). With regards to other explanatory variables (control variables), we select the following ones based on the literature review and data availability. To control the influence of extreme observations, we winsorize the top and bottom 1% of all continuous variables.

(i) ln road_den is the logarithm of road density in a region (in km/km²), which indicates the capacity of land transport facilities in a region (Tian et al., 2022; Yang and Wong, 2012). Data on road density (road length divided by administrative area) are also obtained from the China Premium Database in CEIC Data.

(ii) ln flight is the logarithm of one plus the flight number at all civil aviation airports (CAAs) within a region, which measures the capacity of air transport facilities in a region (Chow et al., 2021; Tian et al., 2022; Yang and Wong, 2012). We first collect the flights number at each CAA from the annual CAA production statistics report issued by the Ministry of Transport of China (https://www.mot.gov.cn/), and then aggregate it by region.

(iii) ln resource is the logarithm of one plus the number of 5A-level scenic spots and national parks within a region. This variable captures the tourism resource endowment. We refer to Yang and Fik (2014) to construct this weight variable, in which 5A-level scenic spots and national parks each contribute half the weight. The World Heritage sites mentioned in the previous literature (Yang et al., 2014) are not considered as most of them fall under the categories of 5A-level scenic spots or national parks. The relevant data are from the websites of Ministry of Culture and Tourism of China (https://www.mct.gov.cn/) and Ministry of Housing and Urban-Rural Development of China (https://www.mohurd.gov.cn/).

(iv) ln GDP_per is the logarithm of the real GDP per capital in constant prices of 2011 (in RMB). This variable reflects the income level in a region (Tian et al. 2022; Yang and Wong 2012). Regions with higher income levels invest more in tourism, which means better tourism infrastructure and better experience for tourists. GDP data is also collected from the China Premium Database in CEIC Data.

The descriptive statistics for all variables are presented in Table 2. All statistics are close for the two dependent variables and the four DFI indexes. Among the control variables, the standard deviation of ln flight is relatively large, and its value is 4.862. This is because many prefecture-level regional aviation industries in China are in their infancy.

Table 2.

Descriptive statistics of variables.

Variable	Period	Obs	Mean	Std. Dev	Min	Max
ln arrival	2011–2019	3015	9.838	1.167	5.835	12.349
ln revenue	2011–2019	3015	9.718	1.353	5.447	12.601
ln DFI_total	2011–2019	3015	4.978	0.524	3.411	5.642
ln DFI_breadth	2011–2019	3015	4.887	0.596	2.583	5.631
ln DFI_depth	2011–2019	3015	4.954	0.526	3.274	5.663
ln DFI_level	2011–2019	3015	5.185	0.588	3.085	5.774
ln road_den	2011–2019	3015	−0.292	0.851	−3.120	0.761
ln flight	2011–2019	3015	5.123	4.862	0.000	13.604
ln resource	2011–2019	3015	0.627	0.622	0.000	2.833
ln GDP_per	2011–2019	3015	10.615	0.561	9.319	11.910

Note: Std. Dev denotes standard deviation.

The high correlation between explanatory variables may produce multicollinearity and thus affect the accuracy of model estimation. The variance inflation factor (VIF) analysis is conducted and its results are showed in Table 3. The VIF values for all explanatory variables remain below 1.5, which is far below the warning value of 10, indicating that there is no serious multicollinearity in our econometric models (Tian et al., 2022).

Table 3.

Variance inflation factor analysis.

Variable	Variance inflation factor
ln DFI_total	1.31
ln DFI_breadth		1.49
ln DFI_depth			1.27
ln DFI_level				1.05
ln road_den	1.13	1.14	1.16	1.13
ln flight	1.32	1.32	1.31	1.31
ln resource	1.24	1.24	1.25	1.24
ln GDP_per	1.40	1.54	1.32	1.21

The panel unit root tests are performed to avoid spurious regression due to the possible non-stationary data. Given the cross-sectional correlation, we follow Song et al. (2023a) to adopt the cross-sectionally augmented IPS (CIPS), the augmented Dickey-Fuller (Fisher-ADF), and the Fisher Phillips Perron (Fisher-PP). The results presented in Table 4 show that all variables are stationary.

Table 4.

Panel unit root tests.

Variable	CIPS test	Fisher-ADF test	Fisher-PP test
ln arrival	−2.035**	3064.549***	1934.938***
ln revenue	−3.336***	4327.066***	2810.408***
ln DFI_total	−11.658***	6509.388***	6850.483***
ln DFI_breadth	−11.437***	5478.985***	1.78 × 10⁴***
ln DFI_depth	−11.026***	4848.444***	4157.943***
ln DFI_level	−18.548***	7008.809***	7904.504***
ln road_denn	−8.133***	2187.879***	2420.505***
ln flight	−4.699***	2145.861***	1189.876***
ln resource	−13.816***	834.391***	2317.203***
ln GDP_per	−16.084***	2559.454***	1023.012***

Note: *** and ** indicate significance at the 0.01 and 0.05 level.

Spatial panel models

The common spatial panel models include spatial auto-regressive (SAR) model, spatial error model (SEM), and spatial Durbin model (SDM). The SAR model incorporates the spatially lagged dependent variable, which takes the following specification form:

Y_{i t} = ρ \sum_{j = 1}^{n} W_{i j} Y_{i t} + β X_{i t} + μ_{i} + ε_{i t}

(1)

where Y_it is the dependent variable of region i at time t, W is the row-normalized spatial weight matrix (Elhorst, 2014), X_it is explanatory variables, ρ is the coefficient for spatially lagged dependent variable, n is the number of regions, μ_i is the region fixed effects, ɛ_it is the disturbance terms.

The SEM contains the spatial lagged error term to reflect the spatial correlation of omitted and unobservable factors, which is in the form of

Y_{i t} = β X_{i t} + μ_{i} + ν_{i t}, ν_{i t} = λ \sum_{j = 1}^{n} W_{i j} ν_{t} + ε_{i t}

(2)

where λ is the coefficient for the spatial lagged error term.

The SDM is the extension of the SAR model, which includes the spatial lagged dependent and all explanatory variables. The SDM is specified as follows:

Y_{i t} = ρ \sum_{j = 1}^{n} W_{i j} Y_{i t} + β X_{i t} + θ \sum_{j = 1}^{n} W_{i j} X_{i t} + μ_{i} + ε_{i t}

(3)

where θ is the coefficient for spatially lagged explanatory variables.

Given the spatial lags of the dependent variable and explanatory variables in spatial panel models (except for SEM), the marginal effect of the explanatory variables on the dependent variable is not equal to the coefficients estimated by the models (Yang and Fik, 2014; Zhang et al., 2021). LeSage and Pace (2009) propose an approach to decompose complicated spatial effects to obtain direct, indirect, and total effects. The direct effect measures the impact of the explanatory variables in the focal region on the dependent variable in the same focal region, including the feedback effect via neighboring regions. The indirect effect, also known as spatial spillover effect, measures the influence of the explanatory variables in the focal regions on the dependent variable in the neighboring regions. The total effect is the sum of the direct effect and indirect effect, measuring the cumulative impact of the explanatory variables in the focal region on the dependent variable in all regions.

This study adopts the Stata command of xsmle, developed by Belotti et al. (2017), to establish these spatial panel models. In xsmle, all spatial panel models are fitted by quasi-maximum likelihood estimation. The spatial weight matrix specifies the connection form and influence degree between spatial regions. We follow the pioneers of spatial tourism demand research in China (Yang and Fik, 2014; Yang and Wong, 2012) and adopt the K-nearest neighbor matrix, in which W_ij equals 1 if region j is the one of the K-nearest regions to region i, otherwise W_ij equals 0. The K-nearest neighbor matrix is the composite of adjacency matrix (including bishop, queen, and rook) and distance-based matrix, which contributes to the balance of weight matrix structure and the stability of model convergence.

Empirical results

Spatial autocorrelation

We first conduct spatial autocorrelation tests on the dependent variables and DFI indexes. The Moran’s I is one of the most popular methods for measuring spatial autocorrelation of variables (Akamavi et al., 2022; Deng and Hu, 2019), which can be expressed in the following equation:

I = \frac{\sum_{i = 1}^{n} \sum_{j = 1}^{n} W_{i j} (V_{i} - \bar{V}) (V_{j} - \bar{V})}{\sum_{i = 1}^{n} {(V_{i} - \bar{V})}^{2}}

(4)

where W is the row-normalized spatial weight matrix, V is the variable to be tested,

\bar{V}

is the mean of the variable across all regions. As for the spatial weight matrix, we follow Elhorst et al. (2013) and Long et al. (2019) to adopt the K-nearest neighbor matrix with the smallest estimated residual variance, that is, the 5-nearest neighbor matrix. In the following model estimates, we continue to follow this setting. We also consider the 3- and 7-nearest neighbor matrixes as the robustness check.

Table 5 presents the Moran’s I values of these main variables. A positive value of the Moran’s I indicates positive spatial autocorrelation, and the larger the value, the stronger the spatial autocorrelation. In almost all years, the Moran’s I values of the dependent variables and the core explanatory variables related to DFI are positive and statistically significant at the 0.01 level. These statistics confirm the intuitive findings of spatial autocorrelation of these variables in Figures 1 and 3.

Table 5.

Moran’s I values of main variables.

	Moran’s I and statistical significance
Year	ln arrival	ln revenue	ln DFI_total	ln DFI_breadth	ln DFI_depth	ln DFI_level
2011	0.539***	0.521***	0.444***	0.441***	0.569***	0.290***
2012	0.541***	0.511***	0.488***	0.456***	0.652***	0.206***
2013	0.518***	0.512***	0.470***	0.436***	0.550***	0.005
2014	0.520***	0.514***	0.396***	0.432***	0.613***	0.289***
2015	0.532***	0.503***	0.465***	0.437***	0.666***	0.320***
2016	0.536***	0.499***	0.467***	0.432***	0.737***	0.276***
2017	0.538***	0.504***	0.470***	0.402***	0.693***	0.195***
2018	0.540***	0.507***	0.527***	0.423***	0.696***	0.533***
2019	0.507***	0.494***	0.545***	0.460***	0.690***	0.593***

Note: *** indicates significance at the 0.01 level.

Estimation results

Determining the appropriate model specification helps to obtain reliable and interpretable results (Gunter et al., 2019). Non-spatial and three spatial panel models are used to estimate the impact of DFI on domestic tourism demand in 335 prefecture-level regions in China from 2011 to 2019. The corresponding estimated results are shown in Table 6. Models 1-4 fit domestic tourist arrivals while Models 5-8 fit domestic tourism revenues.

Table 6.

Estimates of Non-spatial and spatial panel models.

Variable	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Variable	ln arrival	ln arrival	ln arrival	ln arrival	ln revenue	ln revenue	ln revenue	ln revenue
Model	OLS	SAR	SEM	SDM	OLS	SAR	SEM	SDM
ln DFI_total	0.486***	0.176***	0.570***	0.213***	0.626***	0.234***	0.725***	0.200***
ln DFI_total	(0.015)	(0.014)	(0.021)	(0.050)	(0.019)	(0.018)	(0.026)	(0.060)
ln road_den	0.215***	0.122***	0.083**	0.087**	0.237***	0.137***	0.092*	0.098**
ln road_den	(0.046)	(0.036)	(0.039)	(0.038)	(0.057)	(0.043)	(0.048)	(0.047)
ln flight	0.025***	0.014***	0.011***	0.013***	0.027***	0.014***	0.011***	0.013***
ln flight	(0.004)	(0.003)	(0.003)	(0.003)	(0.005)	(0.003)	(0.004)	(0.003)
ln resource	0.198***	0.127***	0.131***	0.125***	0.234***	0.161***	0.169***	0.154***
ln resource	(0.032)	(0.024)	(0.024)	(0.025)	(0.039)	(0.030)	(0.030)	(0.030)
ln GDP_per	0.625***	0.278***	0.353***	0.229***	0.676***	0.247***	0.296***	0.132***
ln GDP_per	(0.034)	(0.028)	(0.039)	(0.040)	(0.042)	(0.034)	(0.047)	(0.049)
W × ln DFI_total				−0.077				−0.015
W × ln DFI_total				(0.052)				(0.063)
W × ln road_den				0.196***				0.226***
W × ln road_den				(0.068)				(0.082)
W × ln flight				0.023***				0.028***
W × ln flight				(0.006)				(0.008)
W × ln resource				0.047				0.035
W × ln resource				(0.054)				(0.065)
W × ln GDP_per				0.107**				0.224***
W × ln GDP_per				(0.052)				(0.063)
ρ		0.599***		0.569***		0.610***		0.574***
ρ		(0.016)		(0.018)		(0.016)		(0.018)
λ			0.623***				0.625***
λ			(0.017)				(0.017)
σ		0.033***	0.034***	0.033***		0.049***	0.051***	0.049***
σ		(0.001)	(0.001)	(0.001)		(0.001)	(0.001)	(0.001)
R-squared	0.756	0.775	0.749	0.786	0.751	0.769	0.742	0.781
AIC	−477.393	−1458.830	−1341.349	−1476.154	764.967	−266.033	−129.094	−291.301
BIC	−441.325	−1416.751	−1299.270	−1404.018	801.035	−223.953	−87.014	−219.165
Hausman test	175.47***	40.15***		45.44***	127.00***	43.56***		57.22***

Note: Standard errors in parentheses. ***, **, and * indicate significance at the 0.01, 0.05, and 0.1 level, respectively.

The results of the robust Hausman test are all positive and significant, which supports the usage of fixed effect (FE) rather than random effect (RE) in panel models. Kim et al. (2021) suggest that FE is better suited to spatial data for all regions of a country because FE can mitigate potential bias caused by omitted variables.

Model 1 and 5 (OLS models) in Table 6 reveal the significant promoting of DFI and control variables on domestic tourism demand, among which the coefficient of ln GDP_per ranks first and the coefficient of ln DFI_total ranks second. Model 2–4 and 6–8 present the estimation results of spatial panel models for domestic tourist arrivals and domestic tourism revenues, respectively. The SDM specifications (Models 4 and 8) have the best fit by combining R-squared, Akaike information criterion (AIC), and Bayesian information criterion (BIC) values, compared with the other two spatial panel models and non-spatial panel models. This highlights that the inclusion of spatial lags of the independent variables favorably improves the goodness of fit of models. Consequently, the interpretation of the models focuses on the SDM specification, which also allows the investigation of spatial spillover effects of DFI.⁶

Given that the coefficients estimated by the SDM are not partial derivatives and cannot measure the marginal impact (Kim et al., 2021), the spatial effect of explanatory variables is further decomposed into direct, indirect, and total effects, as shown in Table 7. We replace the ln DFI_total with the three sub-indexes in turn to explore the mechanism of DFI promoting tourism demand. In almost all models, the direct and indirect effects of the DFI indexes are statistically significant and positive. Taking Model 13 as an example, the coefficients of the direct and indirect effects of ln DFI_total are 0.216 and 0.218, and both are significant at the 0.01 level. In other words, a 1% increase in the DFI total index leads to a 0.216% and 0.218% increase in domestic tourism revenue in the local region and neighboring regions, respectively. This result confirms the local promotion and spatial spillover mechanism of DFI on domestic tourism demand, and also indicates this mechanism can be interpreted from the coverage breadth, usage depth, and digitization level of DFI. The usage depth, in particular, has the most substantial spillover and total effects in almost all estimates. It is also found that the spillover effects of the DFI indexes in all the revenue models (Models 13-16) are larger than those in the corresponding arrival models (Models 9-12).

Table 7.

Decomposition of spatial effect.

Variable	Model 9	Model 10	Model 11	Model 12	Model 13	Model 14	Model 15	Model 16
Variable	ln arrival	ln arrival	ln arrival	ln arrival	ln revenue	ln revenue	ln revenue	ln revenue
Direct effect
ln DFI_total	0.221***				0.216***
ln DFI_breadth		0.106***				0.152***
ln DFI_depth			0.075**				0.074*
ln DFI_level				−0.003				0.003
ln road_den	0.120***	0.131***	0.113***	0.178***	0.136***	0.149***	0.135***	0.196***
ln flight	0.017***	0.018***	0.016***	0.021***	0.019***	0.020***	0.018***	0.023***
ln resource	0.142***	0.152***	0.129***	0.153***	0.172***	0.190***	0.163***	0.178***
ln GDP_per	0.262***	0.266***	0.274***	0.301***	0.175***	0.178***	0.193***	0.214***
Indirect effect
ln DFI_total	0.096*				0.218***
ln DFI_breadth		0.154***				0.178***
ln DFI_depth			0.311***				0.392***
ln DFI_level				0.104***				0.201***
ln road_den	0.531***	0.608***	0.369***	0.911***	0.617***	0.781***	0.511***	1.039***
ln flight	0.064***	0.071***	0.051***	0.089***	0.079***	0.090***	0.067***	0.107***
ln resource	0.266**	0.354***	0.206*	0.387***	0.282*	0.427***	0.250*	0.364**
ln GDP_per	0.513***	0.524***	0.487***	0.740***	0.657***	0.704***	0.688***	0.892***
Total effect
ln DFI_total	0.318***				0.434***
ln DFI_breadth		0.260***				0.330***
ln DFI_depth			0.386***				0.466***
ln DFI_level				0.101***				0.204***
ln road_den	0.651***	0.739***	0.483***	1.088***	0.753***	0.929***	0.646***	1.235***
ln flight	0.082***	0.089***	0.067***	0.109***	0.098***	0.110***	0.085***	0.130***
ln resource	0.408***	0.506***	0.336***	0.540***	0.454***	0.618***	0.413***	0.542***
ln GDP_per	0.775***	0.790***	0.761***	1.041***	0.832***	0.882***	0.881***	1.105***

Note: ***, **, and * indicate significance at the 0.01, 0.05, and 0.1 level, respectively.

In addition, the direct and spillover effects of control variables such as ln road_den, ln flight, ln resource, and ln GDP_per are significantly positive with the expected signs. Among them, the coefficient of ln GDP_per is the largest in almost all the models, both the direct and spillover effects, which highlights the crucial role of economic development and income increase in increasing the domestic tourism demand.

Robustness checks

To confirm the findings of this study are robust, a battery of checks are implemented. First, the measures of tourism demand typically include tourist arrivals and tourism revenues (Song et al., 2023b). In our study, the consistent estimation results of models with the above two variables as dependent variables can support each other.

Second, it’s challenging to determine the spatial weight matrix appropriate to the underlying spatial data generation process (Gyódi and Nawaro, 2021). Although the pioneer of spatial measurement tourism demand research in China recommends K-nearest neighbor matrix, the determination of K mainly relies on experience (Yang and Fik, 2014). We choose the 3- and 7-nearest neighbor matrixes on both sides of the baseline to check the robustness of models, as presented in Table 8. Compared with the baseline results, the estimated signs and significance of the DFI indexes are basically consistent, but their coefficients are reduced.

Table 8.

Estimations under alternative spatial weight matrices.

Spatial weight	Variable	ln arrival	ln arrival	ln arrival	ln arrival	ln revenue	ln revenue	ln revenue	ln revenue
W_3nn	Direct effect
	ln DFI_total	0.295***				0.339***
	ln DFI_breadth		0.157***				0.241***
	ln DFI_depth			0.133***				0.151***
	ln DFI_level				0.017				0.039*
	Indirect effect
	ln DFI_total	0.052				0.175***
	ln DFI_breadth		0.117***				0.140***
	ln DFI_depth			0.270***				0.397***
	ln DFI_level				0.122***				0.221***
	Total effect
	ln DFI_total	0.347***				0.514***
	ln DFI_breadth		0.274***				0.381***
	ln DFI_depth			0.403***				0.547***
	ln DFI_level				0.139***				0.260***
W_7nn	Direct effect
	ln DFI_total	0.194***				0.209***
	ln DFI_breadth		0.097***				0.160***
	ln DFI_depth			0.062*				0.026
	ln DFI_level				−0.020				−0.007
	Indirect effect
	ln DFI_total	0.106*				0.244***
	ln DFI_breadth		0.147***				0.164***
	ln DFI_depth			0.309***				0.476***
	ln DFI_level				0.108***				0.231***
	Total effect
	ln DFI_total	0.300***				0.453***
	ln DFI_breadth		0.243***				0.324***
	ln DFI_depth			0.371***				0.502***
	ln DFI_level				0.088***				0.223***

Note: *** and * indicate significance at the 0.01 and 0.1 level.

Third, the sample period is also critical for model estimation. The DFI index survey began in 2011, and the data available for this study ended in 2019 due to the COVID-19 pandemic in early 2020. Therefore, the sample period in the model estimation is 2011-2019. Two sub-sample periods, 2011-2018 and 2012-2019, are selected to evaluate the robustness of the model estimates. As Table 9 demonstrates, the estimations results under sub-sample periods are almost consistent with those in the full sample period.

Table 9.

Estimations in the sub-sample periods.

Period	Variable	ln arrival	ln arrival	ln arrival	ln arrival	ln revenue	ln revenue	ln revenue	ln revenue
2011-2018	Direct effect
	ln DFI_total	0.182***				0.195***
	ln DFI_breadth		0.084***				0.144***
	ln DFI_depth			0.082***				0.079**
	ln DFI_level				0.000				0.001
	Indirect effect
	ln DFI_total	0.138***				0.217***
	ln DFI_breadth		0.149***				0.131***
	ln DFI_depth			0.309***				0.365***
	ln DFI_level				0.104***				0.189***
	Total effect
	ln DFI_total	0.320***				0.412***
	ln DFI_breadth		0.233***				0.275***
	ln DFI_depth			0.392***				0.444***
	ln DFI_level				0.104***				0.190***
2012-2019	Direct effect
	ln DFI_total	0.418***				0.348***
	ln DFI_breadth		0.291***				0.376***
	ln DFI_depth			0.073				0.019
	ln DFI_level				0.059*				0.005
	Indirect effect
	ln DFI_total	0.256***				0.454***
	ln DFI_breadth		0.343***				0.343***
	ln DFI_depth			0.507***				0.623***
	ln DFI_level				0.178***				0.365***
	Total effect
	ln DFI_total	0.673***				0.802***
	ln DFI_breadth		0.634***				0.719***
	ln DFI_depth			0.580***				0.642***
	ln DFI_level				0.237***				0.370***

Note: ***, **, and * indicate significance at the 0.01, 0.05, and 0.1 level, respectively.

Finally, to mitigate the possibility of reverse causality, we lag the DFI indexes by one year while keeping other variables constant, in which case the data for 2012-2019 is available. The results in Table 10 show that the vast majority of estimates are still consistent with the original estimates.

Table 10.

Estimations under the DFI indexes with a one-year lag.

Variable	ln arrival	ln arrival	ln arrival	ln arrival	ln revenue	ln revenue	ln revenue	ln revenue
Direct effect
L_ ln DFI_total	0.166***				0.181***
L_ ln DFI_breadth		0.086***				0.129***
L_ ln DFI_depth			0.072**				0.028
L_ ln DFI_level				0.001				0.025
Indirect effect
L_ ln DFI_total	0.178***				0.236***
L_ ln DFI_breadth		0.188***				0.177***
L_ ln DFI_depth			0.326***				0.422***
ln DFI_level				0.146***				0.197***
Total effect
L_ ln DFI_total	0.344***				0.417***
L_ ln DFI_breadth		0.274***				0.307***
L_ ln DFI_depth			0.399***				0.450***
L_ ln DFI_level				0.147***				0.222***

Note: *** and ** indicate significance at the 0.01 and 0.05 level. L_variable indicates the variable with a one-year lag.

Discussion and conclusion

Domestic tourism has been praised for reducing spatial inequality, mitigating seasonality and leakage, and being more sustainable (Canavan, 2013; Stone and Nyaupane, 2019). For example, Goh et al. (2015) and Li et al. (2016) state that compared with international tourism, domestic tourism can better promote balanced regional development and reduce regional income inequality in China. A recent study also suggests the importance of domestic tourism in reducing economic vulnerability (Canh and Thanh, 2020). Tourism enterprises have also been benefited from domestic tourism. During the COVID-19 pandemic, many tourism enterprises have turned to domestic tourism to stay afloat. The low entry barrier of domestic tourism has allowed household/unregistered SMEs to participate, thereby improving livelihoods and boosting social welfare (Seyfi et al., 2022).

DFI, which involves providing formal financial services to previously excluded individuals and businesses through digital means, has attracted increasing attention from governments, academics, and practitioners in recent years (Demir et al., 2022; Hu et al., 2023; Yang and Zhang, 2022). In the context of global tourism, especially domestic tourism, recovering from the COVID-19 pandemic, it is of great significance to understand the mechanism of DFI in boosting domestic tourism demand. In this study, we estimate and analyze the direct and spatial spillover effects of DFI on domestic tourism demand with region-level balanced panel data.

The main findings and implications are as follows. During the sample period, the total index of DFI and its sub-indices in three dimensions exhibit significant spatial autocorrelation, which are consistent with the two measures of domestic tourism demand, namely, domestic tourist arrivals and tourism revenues. DFI and its three dimensions have significant and positive direct and spillover effects on domestic tourism demand, among which the spillover effects and total effects of the usage depth are larger than those of the other two dimensions. These findings emphasize the importance for regional planners and policy makers to not only develop DFI comprehensively and in a coordinated manner but also focus on enhancing the usage depth of DFI. Moreover, the spillover effects of DFI on domestic tourism demand in the tourist revenues models are larger than that in the corresponding tourism arrivals models, highlighting the effectiveness of DFI in internalizing regional tourism economic benefits.

The current digital economy is fundamentally reshaping the tourism industry (Perelygina et al., 2022; Sharma et al., 2020). As a crucial infrastructure for digital economy development, DFI connects financial institutions, tourism enterprises, and tourists through the establishment of digital financial platforms and the application of digital technologies, promoting the allocation and flow of financial resources and enhancing the efficiency and vitality of the tourism economy. Governments should prioritize the implementation of digital platforms and technologies, such as DFI, and incorporate them into strategic planning and policy frameworks. Tourism enterprises, especially SMEs, should actively embrace the digital trend, fostering business model innovation, improving operational efficiency, and providing tourists with more convenient and comfortable services.

A study closely related to our research is Gopalan and Khalid (2022), which mainly focuses on the impact of (traditional) financial inclusion on (international) tourism demand in the context of emerging markets and developing economies. They also investigated the role of DFI through several isolated indicators. In contrast, we adopt a composite DFI index system and limit the study scope to domestic tourism demand by fully considering the applicability of this index system.⁷ Our findings support the pending hypothesis of Gopalan and Khalid (2022) about DFI. Moreover, from a methodological perspective, we also identify the spatial spillover effect of DFI with the help of spatial econometric techniques.

A recent authoritative review graphically illustrates the determinants of tourism demand that have been revealed in research over the past five years (Song et al., 2023b). Silva et al. (2017) and Silva and Hassani (2022) comprehensively summarize the leading indicators related to tourism demand forecasting explored by the study published during the period of 1997 to 2021. These determinants and leading indicators mainly include economic variables (e.g., income-like and price-like variables), transport variables (e.g., expressway mileage), environment and climate variables (e.g., air quality), political stability variables, and Internet data (e.g., Google search intensity index). The determinants are leading indicators in the specific forecast scenarios, while the leading indicators are often reflectors to tourism demand rather than determinants (Song et al., 2023b). In this study, we identify DFI as a determinant of domestic tourism demand. It is foreseeable that DFI has the potential to serve as a candidate leading indicator to help improve the performance of annual tourism demand forecasting in the digital economy era. In the context of the COVID-19 pandemic, tourism demand forecasting has shifted to a fine-grained forecasting paradigm (i.e., weekly, daily, or even hourly forecasting) to support tourism stakeholders in implementing flexible and actionable measures to cope with high uncertainty (Bi et al., 2023; Li et al., 2023; Yang et al., 2022). Future research should investigate high-frequency leading indicators related to DFI (e.g., online sales of tourism tickets or insurance).

Nevertheless, this study has several limitations. First, due to the ongoing COVID-19 pandemic, data is only available until 2019. While we acknowledge the unprecedented role of DFI in boosting domestic tourism demand during the COVID-19 pandemic, quantifying the implications is outside the scope of this study. Second, the data used in this study is limited to China. Future research can investigate the impact of DFI on domestic tourism demand in other popular domestic tourism destinations, such as the United States, India, Spain, and Australia. In addition, in a destination where both international and domestic tourists have access to the same digital financial services (e.g., Eurozone), it is interesting to compare the difference in the impact of DFI on international and domestic tourism demand. Finally, the local spatial models, such as the geographically weighted regression model and its variants, can further reveal the spatial heterogeneity of the effect of DFI on domestic tourism demand. We leave the discussion of these questions for future research.

Footnotes

Acknowledgements

We would like to thank the editor and two anonymous referees for their valuable feedback and suggestions.

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 work was supported by the Construction and Demonstration of Shaanxi Science and Technology Holding Group Financial Service Platform (2019ZDLGY11-03), the Science and Technology Project of China Huaneng Group (HNKJ20-H87), the China Scholarship Council (202206280179), the Natural Science Basic Research Program of Shaanxi (2023-JC-QN-0785), Singapore Ministry of Education Academic Research Fund Tier 1 (RG143/19), and the Society of Actuaries Education Institution Grant.

ORCID iDs

Yanzhao Li

Ju-e Guo

Wenjun Zhu

Notes

Author biographies

Yanzhao Li is a Ph.D. candidate at School of Management, Xi’an Jiaotong University, China. He is currently visiting Nanyang Business School at Nanyang Technological University with funding from the China Scholarship Council. His research interest mainly lies in the financial inclusion, tourism management, and machine learning forecasting. He has published several papers in Tourism Economics, Current Issues in Tourism, Asia Pacific Journal of Tourism Research, Financial Innovation, and others.

Ju-e Guo is a Full Professor at School of Management, Xi’an Jiaotong University, China. She is the executive deputy director of Research Center of Chinese Management, which is the soft science research base of the Ministry of Education, China. Her research interests include investment and financing decision, tourism economics, and energy economics. She has published more than 50 papers in leading journals, such as Omega, International Journal of Production Economics, Tourism Economics, and Energy Economics.

Wenjun Zhu is an Assistant Professor at Nanyang Business School, Nanyang Technological University, Singapore. She is a Fellow of the Society of Actuaries (FSA) and a Chartered Enterprise Risk Analyst (CERA). Her research spans diverse areas, encompassing statistical modeling in finance and insurance, climate risk management, predictive analytics with machine learning, and others. She has published over 20 papers in leading journals, such as Management Science and Journal of Banking and Finance.

References

Adamiak

Szyda

(2022) Combining conventional statistics and big data to map global tourism destinations before COVID-19. Journal of Travel Research 61(8): 1848–1871.

Akamavi

Ibrahim

Swaray

(2022) Tourism and troubles: effects of security threats on the global travel and tourism industry performance. Journal of Travel Research 62(8): 1755–1800.

Alvarez-Diaz

D’Hombres

Ghisetti

, et al. (2020) Analysing domestic tourism flows at the provincial level in Spain by using spatial gravity models. International Journal of Tourism Research 22(4): 403–415.

Babilla

TUK

(2023) Digital innovation and financial access for small and medium-sized enterprises in a currency union. Economic Modelling 120: 106182.

Belotti

Hughes

Mortari

(2017) Spatial panel-data models using stata. The Stata Journal 17(1): 139–180.

Han

Yao

(2023) Fine-grained tourism demand forecasting: a decomposition ensemble deep learning model. Tourism Economics 29(7): 1736–1763.

Canavan

(2013) The extent and role of domestic tourism in a small island: the case of the Isle of Man. Journal of Travel Research 52(3): 340–352.

Canh

Thanh

(2020) Domestic tourism spending and economic vulnerability. Annals of Tourism Research 85: 103063.

Cannonier

Burke

(2017) Tourism and financial development in small states: evidence from Caribbean countries. Tourism Economics 23(6): 1369–1377.

10.

Carrillo-Hidalgo

Pulido-Fernández

(2019) Is the financing of tourism by international financial institutions inclusive? A proposal for measurement. Current Issues in Tourism 22(3): 330–356.

11.

Célerier

Matray

(2019) Bank-branch supply, financial inclusion, and wealth accumulation. The Review of Financial Studies 32(12): 4767–4809.

12.

Chow

CKW

Tsui

WHK

(2021) Airport subsidies and domestic inbound tourism in China. Annals of Tourism Research 90: 103275.

13.

Churchill

Cai

Erdiaw-Kwasie

, et al. (2022) Financial development and tourism: a century of evidence from Germany. Applied Economics 55(1): 1–14.

14.

Demir

Pesqué-Cela

Altunbas

, et al. (2022) Fintech, financial inclusion and income inequality: a quantile regression approach. The European Journal of Finance 28(1): 86–107.

15.

Deng

(2019) Modelling China’s outbound tourist flow to the ‘Silk Road’: a spatial econometric approach. Tourism Economics 25(8): 1167–1181.

16.

Elhorst

(2014) Spatial econometrics: from cross-sectional data to spatial panels. Heidelberg: Springer, Vol. 479.

17.

Elhorst

Zandberg

De Haan

(2013) The impact of interaction effects among neighbouring countries on financial liberalization and reform: a dynamic spatial panel data approach. Spatial Economic Analysis 8(3): 293–313.

18.

Gao

Wang

(2022) Does digital inclusive finance promote entrepreneurship? Evidence from Chinese cities. Singapore Economic Review. doi: 10.1142/S0217590822500618.

19.

Goh

Zhang

(2015) Achieving balanced regional development in China: is domestic or international tourism more efficacious? Tourism Economics 21(2): 369–386.

20.

Gopalan

Khalid

(2022) How does financial inclusion influence tourism demand? Empirical evidence from emerging markets and developing economies. Tourism Recreation Research. doi: 10.1080/02508281.2022.2028084.

21.

Gopalan

Khalid

(2023) Promoting outbound tourism through financial inclusion: evidence from emerging markets and developing economies. Annals of Tourism Research Empirical Insights 4(1): 100086.

22.

Gunter

Önder

Smeral

(2019) Scientific value of econometric tourism demand studies. Annals of Tourism Research 78: 102738.

23.

Guo

Wang

, et al. (2020) Measuring China’s digital financial inclusion: index compilation and spatial characteristics. China Economic Quarterly 19(4): 1401–1418.

24.

Guo

Feng

Lin

(2023) Digital inclusive finance and digital transformation of enterprises. Finance Research Letters 57: 104270.

25.

Gyódi

Nawaro

(2021) Determinants of airbnb prices in european cities: a spatial econometrics approach. Tourism Management 86: 104319.

26.

Fang

DiGiovanni

(2023) Technological progress, financial constrains, and digital financial inclusion. Small Business Economics 61(4): 1693–1721.

27.

International Monetary Fund (2023) Financial development and financial inclusion (FDFI). Available at: https://www.imf.org/en/CapacityDevelopment/Training/ICDTC/Schedule/JV/2023/FDFIJV2317 (accessed 20 April 2023).

28.

Jiao

Chen

(2021) Forecasting tourism demand: developing a general nesting spatiotemporal model. Annals of Tourism Research 90: 103277.

29.

Katircioglu

Katircioğlu

Altinay

(2018) Interactions between tourism and financial sector development: evidence from Turkey. The Service Industries Journal 38(9-10): 519–542.

30.

Khanna

Sharma

(2021) Does financial development raise tourism demand? A cross-country panel evidence. Journal of Hospitality & Tourism Research 47(6): 1040–1070.

31.

Kim

Williams

Park

, et al. (2021) Spatial spillovers of agglomeration economies and productivity in the tourism industry: the case of the UK. Tourism Management 82: 104201.

32.

Kim

Liu

Stienmetz

, et al. (2022) Visitor flow spillover effects on attraction demand: a spatial econometric model with multisource data. Tourism Management 88: 104432.

33.

LeSage

Pace

(2009) Introduction to Spatial Econometrics. New York: CRC Press.

34.

Xiao

Zhang

, et al. (2011) Measuring tourism spillover effects among cities: improvement of the gap model and a case study of the Yangtze River Delta. Journal of China Tourism Research 7(2): 184–206.

35.

Chen

, et al. (2016) Tourism and regional income inequality: evidence from China. Annals of Tourism Research 58: 81–99.

36.

Yang

Zhong

, et al. (2021) Agent-based modeling of spatial spillover effects in visitor flows. Journal of Travel Research 60(3): 546–563.

37.

Guo

Zhao

, et al. (2022) Research on the spatial-temporal pattern and spatial spillover effect of tourism based on mobile signaling and pois data: a case study of Xiamen city, southeast China. Asia Pacific Journal of Tourism Research 27(10): 1052–1070.

38.

Yang

Guo

, et al. (2023) Daily tourism demand forecasting before and during covid-19: data predictivity and an improved decomposition-ensemble framework. Current Issues in Tourism. doi: 10.1080/13683500.2023.2202308.

39.

Lin

Qin

, et al. (2022) Multiple effects of “distance” on domestic tourism demand: a comparison before and after the emergence of COVID-19. Annals of Tourism Research 95: 103440.

40.

Long

Liu

Song

(2019) Pooling in tourism demand forecasting. Journal of Travel Research 58(7): 1161–1174.

41.

Marrocu

Paci

(2013) Different tourists to different destinations. Evidence from spatial interaction models. Tourism management 39: 71–83.

42.

Menegaki

Tiwari

(2022) The stability of interaction channels between tourism and financial development in 10 top tourism destinations: evidence from a Fourier Toda-Yamamoto estimator. Tourism Economics 28(7): 1914–1942.

43.

Otero-Giráldez

Álvarez-Díaz

González-Gómez

(2012) Estimating the long-run effects of socioeconomic and meteorological factors on the domestic tourism demand for Galicia (Spain). Tourism Management 33(6): 1301–1308.

44.

Perelygina

Kucukusta

Law

(2022) Digital business model configurations in the travel industry. Tourism Management 88: 104408.

45.

Seyfi

Hall

Saarinen

(2022) Rethinking sustainable substitution between domestic and international tourism: a policy thought experiment. Journal of Policy Research in Tourism, Leisure and Events. doi: 10.1080/19407963.2022.2100410.

46.

Sharma

Chaudhary

(2020) Are small travel agencies ready for digital marketing? Views of travel agency managers. Tourism Management 79: 104078.

47.

Shi

Swamy

Paramati

(2021) Does financial inclusion promote tourism development in advanced and emerging economies? Applied Economics Letters 28(6): 451–458.

48.

Silva

Hassani

(2022) ‘modelling’UK tourism demand using fashion retail sales. Annals of Tourism Research 95: 103428.

49.

Silva

Ghodsi

, et al. (2017) Cross country relations in european tourist arrivals. Annals of Tourism Research 63: 151–168.

50.

Song

Liu

, et al. (2023a) Tourism development and urban housing prices: evidence from China. Tourism Economics. doi: 10.1177/13548166231186906.

51.

Song

Qiu

Park

(2023b) Progress in tourism demand research: theory and empirics. Tourism Management 94: 104655.

52.

Stone

Nyaupane

(2019) The tourist gaze: domestic versus international tourists. Journal of travel research 58(5): 877–891.

53.

Tian

Yang

Jiang

(2022) Spatial spillover of transport improvement on tourism growth. Tourism Economics 28(5): 1416–1432.

54.

Tsui

KWH

(2017) Does a low-cost carrier lead the domestic tourism demand and growth of New Zealand? Tourism Management 60: 390–403.

55.

UNWTO (2020) UNWTO highlights potential of domestic tourism to help drive economic recovery in destinations Worldwide. Available at: https://www.unwto.org/news/unwto-highlights-potential-of-domestic-tourism-to-help-drive-economic-recovery-in-destinations-worldwide (accessed 14 April 2023).

56.

UNWTO (2023) UNWTO World Tourism Barometer. Available at: https://www.e-unwto.org/toc/wtobarometereng/21/1 (accessed 17 April 2023).

57.

Weng

Ding

(2021) Measuring demand spillover of vacation town–a case of Shanghai Disney Resort. Asia Pacific Journal of Tourism Research 26(2): 95–108.

58.

World Bank (2015) Global financial development report 2015/2016: long term-finance. https://www.worldbank.org/en/publication/gfdr/gfdr-2016/background/financial-development (accessed 13 April 2023).

59.

WTTC (2022) Travel and tourism economic impact global trends 2022. Available at: https://wttc.org/research/economic-impact (accessed 17 April 2023).

60.

(2023) Impacts of digital inclusive finance on household entrepreneurship. Finance Research Letters 56: 104114.

61.

Huang

Hou

, et al. (2020) The spatial spillover effects of haze pollution on inbound tourism: evidence from mid-eastern China. Tourism Geographies 22(1): 83–104.

62.

Yang

Fik

(2014) Spatial effects in regional tourism growth. Annals of Tourism Research 46: 144–162.

63.

Yang

Wong

(2012) A spatial econometric approach to model spillover effects in tourism flows. Journal of Travel Research 51(6): 768–778.

64.

Yang

Zhang

(2022) Fintech adoption and financial inclusion: evidence from household consumption in China. Journal of Banking & Finance 145: 106668.

65.

Yang

Liu

(2014) Domestic tourism demand of urban and rural residents in China: does relative income matter? Tourism Management 40: 193–202.

66.

Yang

Fik

Zhang

(2017) Designing a tourism spillover index based on multidestination travel: a two-stage distance-based modeling approach. Journal of Travel Research 56(3): 317–333.

67.

Yang

Fan

Jiang

, et al. (2022) Search query and tourism forecasting during the pandemic: when and where can digital footprints be helpful as predictors? Annals of Tourism Research 93: 103365.

68.

Zhang

Zhuang

(2011) The spatial relationship of tourist distribution in Chinese cities. Tourism Geographies 13(1): 75–90.

69.

Zhang

Singh

(2016) Does it matter? examining the impact of China’s vacation policies on domestic tourism demand. Journal of Hospitality & Tourism Research 40(5): 527–556.

70.

Zhang

Wan

, et al. (2020) Fintech, growth and inequality: evidence from China’s household survey data. Singapore Economic Review 65(supp01): 75–93.

71.

Zhang

Wang

Yang

, et al. (2021) Host facial features and Airbnb listing prices: a spatial hedonic pricing approach. Journal of Travel & Tourism Marketing 38(9): 991–1003.

72.

Zhang

Chen

Wang

, et al. (2023a) Spatial-temporal evolution and influencing factors of digital financial inclusion: county-level evidence from China. Chinese Geographical Science 33(2): 221–232.

73.

Zhang

Xiang

, et al. (2023b) Digitalization, financial inclusion, and small and medium-sized enterprise financing: evidence from China. Economic Modelling 126: 106410.

74.

Zhou

Yang

, et al. (2017) The spillover effect of attractions: evidence from Eastern China. Tourism Economics 23(4): 731–743.