Relationship between ICT and international tourism demand: A study of major tourist destinations

Abstract

In this article, we study the effect of ICT on tourism demand in nine major tourist destinations based on visitor arrivals. Mobile and broadband subscriptions are used to proxy for ICT. Additionally, we account for price, source country’s income, and the destination’s income. Balanced panels for the period 1995–2017 and 2002–2017 are used for mobile and broadband subscriptions, respectively. The pooled mean group approach is used for estimation. The results indicate a 1% increase in mobile subscriptions and broadband would increase international visitor arrivals by 0.04% and 0.11%, respectively. The elasticity coefficients of price and income are −0.71 and 1.58, respectively, based on the mobile subscription model, and −0.88 and 1.83, respectively, based on broadband subscription. The destination’s income has only a short-run positive association with tourism demand. The causality results indicate that ICT cause tourism demand, and support for technology-led growth hypothesis in the major tourist destinations.

Keywords

causality cointegration ICT major tourist destinations pooled mean group tourism demand

Introduction

Technology continues to revolutionize tourism industries in many countries, especially in the context of digital tourism. Digital tourism refers to a suite of real world and online contents aimed at improving experiences of tourists (Adeola and Evans, 2019a). For tourists, technology can be an important tool, especially in terms of finding information about the intended destinations, climate and weather patterns, accommodations, sceneries, geopolitical and economic situations, making travel bookings, online shopping and payments, and capturing memories, among other things (Law et al., 2018).

Several studies have shown a close link between ICT and economic growth (Jorgenson, 2001; Jorgenson and Vu, 2011; Kumar, 2014; Kumar and Kumar, 2012; Kumar et al., 2015, 2016; Madden and Savage, 1998; Pohjola, 2002; Stiroh, 2002; Strobel, 2016; Ward and Zheng, 2016). The high-income and highly developed provinces, countries, and regions can benefit from ICT developments, especially in terms of productivity and economic growth (Lam and Shiu, 2010). Because tourism development and economic activity are closely linked (Lanza and Pigliaru, 2000; Narayan et al., 2010; Pablo-Romero and Molina, 2013; Sheldon, 1993), interest in exploring key drivers of tourism demand is growing. Studies have shown that developments in the tourism sector can result in the improvements in infrastructure including technology and public facilities (Andereck et al., 2005; Yoon et al., 2001), the reduction in poverty (Croes, 2014; Vanegas et al., 2015), additional employment (Andereck and Nyaupane, 2011), and the creation of local businesses (Dyer et al., 2007).

In this study, we examine the association between ICT, measured by mobile and broadband usage, and tourism demand in nine major tourist destinations ranked by visitor arrivals. Drawing insights from the gravity model and demand theory, we estimate the tourism demand model that includes leading tourist destinations. Our sample includes China, France, Germany, Italy, Mexico, Russia, Spain, the United Kingdom, and the United States.¹ Interestingly, these countries possess a well-developed ICT infrastructure (Lam and Shiu, 2010) and are experiencing growth in tourism (Shahzad et al., 2017). The pooled mean group (PMG; Pesaran et al., 1999) approach is used to estimate the long-run and short-run models, and the Dumitrescu–Hurlin (2012) heterogeneous panel test is used to examine causation. The model with mobile usage as an indicator of ICT comprises data from 1995 to 2017, and the model with broadband usage as an indicator comprises data from 2002 to 2017, which amounts to 207 and 144 observations, respectively. The rest of the article is set as follows: in the second section, we discuss the literature on ICT and tourism. The third section is on data and methods. The fourth section 4 is on the results, and the fifth section concludes with some policy implications.

Literature review

Tourism demand is often measured by visitor arrivals, tourism receipts or expenditure, nights spent by a tourist in the destination country or distance traveled, Song et al. (2010) categorize these measures into quantity, pecuniary, time consumed, and distance criterion. In any case, visitor arrivals are the most commonly used measure of tourism demand (Crouch, 1994; Li et al., 2005; Lim, 1997). The popularity of the use of visitor arrivals as a measure is because they provide information to tourism service suppliers regarding the scope of new investments in hotel, aircraft, and tourism infrastructure (Sheldon, 1993; Song et al., 2010).

A tourism demand model consists of variables like the tourist’s price and the tourists’ income, among other variables. The income variable is usually measured by the real GDP per capita of the source markets, industrial production index, and the real consumption expenditure (Dogru et al., 2017; Kim and Lee, 2017). However, when the focus is on estimating models that examine overall tourism demand from all destinations’ source markets, then the use of the world real GDP can represent the income variable (Algieri, 2006).

The price variable is commonly measured by the real exchange rate. This measure also minimizes the risk of obtaining biased price elasticities when relative prices and exchange rates are individually incorporated into the model (Seetaram et al., 2016). The real exchange rate allows a prospective tourist to compare the cost of living in the destination against the local currency (Song and Li, 2008). Moreover, it has been noted that leisure travelers are more price elastic than business travelers because the latter can postpone their travel (Crouch, 1994). Depending on the availability of data, studies have considered additional variables to account for transportation costs (Narayan, 2004), substitute prices (Seetaram, 2012), seasonality effects (Ridderstaat et al., 2014), and structural breaks (Smeral, 2017) in a tourism demand model.

Further, the causal models for tourism demand have followed either the demand or gravity theories. Some earlier studies like Anderson and Van Wincoop (2003) suggest that such an approach overcomes the large data requirements of the demand-based model that follows the almost ideal demand system. The gravity model specification includes trade which can act as the tourism price indicator. Durbarry (2008) uses a gravity model to explain the inbound international tourism demand in the United Kingdom. The regression method employed was the fixed and random effects. The results are consistent with the demand theory in that tourism price, measured by the real exchange rate, has a negative association with tourism demand. Moreover, the source country’s income has a positive effect on the United Kingdom’s tourism demand. Interestingly, the study notes that the income of the destination is negatively associated with tourism demand, and that travel distance has a positive association.

Digital tourism refers to the application of ICT to create a substantive user experience for the tourist (Adeola and Evans, 2019a; Benyon et al., 2014). Recent studies have shown that ICT like smartphones and the Internet can have positive impacts on the tourism industry, in terms of, among other things, easing communication channels, providing efficient payment mechanisms, and access to useful information for decision-making (Law et al., 2018; Tan et al., 2017; Wang et al., 2018).

Focusing on the e-commerce industry, Werthner and Ricci (2004) highlight that the tourism sector can be a positive driver. This and some other studies (Buhalis, 1998; Buhalis and Law, 2008; Navío-Marco et al., 2018) have established a plausible link between tourism and technology. The ICT-tourism is characterized by consumer demand, technological innovations, and industry functions. From the consumer demand perspective, ICT enables tourists to access up-to-date information on reservations at a fraction of the cost and provides relevant information regarding destinations, resorts and hotels, and suite of activities (Buhalis and Law, 2008; Li and Buhalis, 2005; Mills and Law, 2004; Niininen et al., 2007; O’Connor, 1999; O’Connor et al., 2001). In this regard, ICT can promote word-of-mouse effect through review and post-travel reports on websites and travel forums (Gelb and Sundaram, 2002). Subsequently, ICT can reduce time lags due to lengthy information search, and the uncertainty that may arise from expensive travels and bad experiences (Fodness and Murray, 1997; Harrison-Walker, 2001; Song et al., 2003; cf. Spencer, 2019). In an earlier study, Buhalis (1998) notes that many service providers use ICT to communicate prices and special offers to prospective tourists. Savings from lower prices, commissions, or reduced airfares means tourists have more to spend at a prospective destination (Clemons et al., 2002; Luo et al., 2004). Additionally, through easy access to information supported by ICT, a well-informed tourist is better able to interact with the locals, find special offers and other deals that meet his or her tastes, and add a personal touch to the trip (Kotiloglu et al., 2017). The use of smartphones has greatly facilitated these aspects among other things like travel apps to help in route mapping, booking tickets and accommodation, and searching for packaged tours (Adeola and Evans, 2019a).

While the theoretical link between ICT and tourism is clear, there are a few studies that have empirically examined the relationship between the two (Adeola and Evans, 2019a, 2019b). Adeola and Evans (2019a) examine the linear and nonlinear effects of mobile phones and Internet in Africa over the period 1996–2017 using the systems generalized method of moments approach. They note a U-shaped response between mobile penetration, Internet usage, and tourism. Their results indicate a unidirectional causality from mobile usage to tourism demand. In another study, Adeola and Evans (2019b) relate ICT infrastructure to tourism development in African countries over the period 1996–2016 using the dynamic gravity model. They find that ICT infrastructure has a positive and statistically significant effect on visitor arrivals. Moreover, the study notes that additional drivers of tourism demand are the bilateral real exchange rate and the real per capita GDP of the origin countries. However, it was noted that distance and visitor arrivals are negatively correlated which supports the view that greater travel distance could be due to higher travel costs.

Methods and data

Cointegration, long run, and short run

The basic model specification is consistent with earlier studies (Durbarry, 2008; Lin et al., 2015) and follows the autoregressive distributed lag (ARDL) framework

\begin{array}{l} Δln Q_{i, t} = μ_{1} + \sum_{j = 1}^{q_{1}} ϑ_{1 j} Δln Q_{i, t - j} + \sum_{j = 0}^{q_{2}} ϑ_{2 j} Δln P_{i, t - j} + \sum_{j = 0}^{q_{3}} ϑ_{3 j} Δln Y_{i, t - j} \\ + \sum_{j = 0}^{q_{4}} ϑ_{4 j} Δln Y D_{i, t - j} + \sum_{j = 0}^{q_{5}} ϑ_{5 j} {Δln ICT}_{i, t - j} + φ_{1} ln Q_{i, t - 1} + φ_{2} ln P_{i, t - 1} + φ_{3} ln Y_{i, t - 1} \\ + φ_{4} ln Y D_{i, t - 1} + φ_{5} ln {ICT}_{i, t - 1} + u_{i, t} \end{array}

where $Q_{i, t}$ is visitor arrivals in destination i, $μ_{1}$ is the deterministic component which includes the intercept term and time trend, $P_{i, t}$ is the tourism price in destination i, $Y_{i, t}$ is the tourist’s income, $Y D_{i, t}$ is the income in the destination, ${ICT}_{i, t} \in {{MOB}_{i, t}, {BB}_{i, t}}$ is either the number of mobile subscriptions ${MOB}_{i, t}$ or the number of broadband subscriptions ${BB}_{i, t}$ , $α^{L R} = - φ_{2} / φ_{1}$ is the price elasticity of tourism demand, $β^{L R} = - φ_{3} / φ_{1}$ is the income elasticity of tourism demand, $θ^{L R} = - φ_{4} / φ_{1}$ is the destination income elasticity of tourism demand, $φ^{L R} = - φ_{5} / φ_{1}$ is the ICT elasticity of demand, $- 1 < φ_{1} < 0$ is the error correction term’s (ECT) coefficient, and $u_{i, t}$ is the residual term.

For estimation, we use the PMG estimator developed by Pesaran et al. (1999). This is a panel ARDL approach and the model is estimated with maximum likelihood with the short-run individual heterogeneous effects (Pesaran and Smith, 1995) and the long-run homogenous effects. The procedure is an intermediate case between the mean group and dynamic fixed panel estimator (cf. Khoshnevis-Yazdi and Dariani, 2019). The procedure can be applied with $I (0)$ and $I (1)$ data. Moreover, it avoids the need to examine cross-sectional dependence and the need to examine cointegration due to the long-run homogeneity (Goh et al., 2018).

The Hausman homogeneity test is an integral part of the PMG analysis. It tests the null hypothesis of that pooling of the long-run coefficients is appropriate against the alternative that long-run effects poolability is invalid (Asafu-Adjaye et al., 2016). The null hypothesis requires that the difference between the mean group and PMG is not systematic (Asafu-Adjaye et al., 2016). Additional tests carried out are the unit roots test which assumes common AR effects (Levin et al., 2002) and individual AR effects (Im et al., 2003), and the tests for robustness in the presence of cross-sectional dependence (Pesaran, 2007). Additionally, we test for cross-sectional dependence using the Breusch and Pagan Lagrange multiplier (LM), Pesaran’s scaled LM, and Pesaran’s bias-corrected scaled LM tests. Cointegration test follows the Pedroni’s (2004) and Kao’s (1999) tests. The two cointegration tests assume the null hypothesis of no cointegration.

Causality

We apply two tests to assess the direction of causality. To test for bivariate causality, we use the Granger causality test developed by Dumitrescu and Hurlin (2012) with insights from some recent tourism demand studies (Dogru and Bulut, 2018; Tugcu, 2014). The approach is a non-causality test for heterogeneous panels with fixed coefficients in a bivariate vector autoregression (VAR) model (Tugcu, 2014). Under the null hypothesis, there is no causality for any cross-sectional units of the panel, termed as the homogenous non-causality (HNC) hypothesis. Under the alternative hypothesis, at least one cross-sectional subgroup has a causal relationship from the explanatory to the independent variable (Tugcu, 2014). The Wald statistic is averaged for each cross-sectional unit to develop the panel version of the test static to test the HNC hypothesis (Dumitrescu and Hurlin, 2012). The approach has good statistical properties even in the presence of cross-sectional dependence and can be applied irrespective of the structure of the panel (Alam and Paramati, 2016; Dogru and Bulut, 2018; Tugcu, 2014).

Additionally, to test for multivariate causality, we apply the Granger causality test of Toda and Yamamoto (1995). The approach enables causality to be determined with a mixture of $I (0)$ , $I (1)$ , and fractionally integrated variables. Moreover, causality can be examined in the absence of cointegration. For this purpose, the following VAR models are specified

\begin{array}{l} In Q_{t} = α_{0} + \sum_{i = 1}^{k} α_{1 i} In Q_{t - i} + \sum_{j = k + 1}^{δ_{max}} α_{2 j} In Q_{t - j} + \sum_{i = 1}^{k} α_{3 i} In P_{t - i} \\ + \sum_{j = k + 1}^{δ_{max}} α_{4 j} In P_{t - j} + \sum_{i = 1}^{k} α_{5 i} In Y_{t - i} + \sum_{j = k + 1}^{δ_{max}} α_{6 j} In Y_{t - j} + \sum_{i = 1}^{k} α_{7 i} In Y D_{t - i} \\ + \sum_{j = k + 1}^{δ_{max}} α_{8 j} In Y D_{t - j} + \sum_{i = 1}^{k} α_{9 i} {In ICT}_{t - i} + \sum_{j = k + 1}^{δ_{max}} α_{10 j} {In ICT}_{t - j} + u_{1 t} \end{array}

\begin{array}{l} In P_{t} = β_{0} + \sum_{i = 1}^{k} β_{1 i} In Q_{t - i} + \sum_{j = k + 1}^{δ_{max}} β_{2 j} In Q_{t - j} + \sum_{i = 1}^{k} β_{3 i} In P_{t - i} \\ + \sum_{j = k + 1}^{δ_{max}} β_{4 j} In P_{t - j} + \sum_{i = 1}^{k} β_{5 i} In Y_{t - i} + \sum_{j = k + 1}^{δ_{max}} β_{6 j} In Y_{t - j} + \sum_{i = 1}^{k} β_{7 i} In Y D_{t - i} \\ + \sum_{j = k + 1}^{δ_{max}} β_{8 j} In Y D_{t - j} + \sum_{i = 1}^{k} β_{9 i} {In ICT}_{t - i} + \sum_{j = k + 1}^{δ_{max}} β_{10 j} {In ICT}_{t - j} + u_{2 t} \end{array}

\begin{array}{l} In Y_{t} = ζ_{0} + \sum_{i = 1}^{k} ζ_{1 i} In Q_{t - i} + \sum_{j = k + 1}^{δ_{max}} ζ_{2 j} In Q_{t - j} + \sum_{i = 1}^{k} ζ_{3 i} In P_{t - i} + \sum_{j = k + 1}^{δ_{max}} ζ_{4 j} In P_{t - j} \\ + \sum_{i = 1}^{k} ζ_{5 i} In Y_{t - i} + \sum_{j = k + 1}^{δ_{max}} ζ_{6 j} In Y_{t - j} + \sum_{i = 1}^{k} ζ_{7 i} In Y D_{t - i} + \sum_{j = k + 1}^{δ_{max}} ζ_{8 j} In Y D_{t - j} \\ + \sum_{i = 1}^{k} ζ_{9 i} {In ICT}_{t - i} + \sum_{j = k + 1}^{δ_{max}} ζ_{10 j} {In ICT}_{t - j} + u_{3 t} \end{array}

\begin{array}{l} In Y D_{t} = η_{0} + \sum_{i = 1}^{k} η_{1 i} In Q_{t - i} + \sum_{j = k + 1}^{δ_{max}} η_{2 j} In Q_{t - j} + \sum_{i = 1}^{k} η_{3 i} In P_{t - i} \\ + \sum_{j = k + 1}^{δ_{max}} η_{4 j} In P_{t - j} + \sum_{i = 1}^{k} η_{5 i} In Y_{t - i} + \sum_{j = k + 1}^{δ_{max}} η_{6 j} In Y_{t - j} + \sum_{i = 1}^{k} η_{7 i} In Y D_{t - i} \\ + \sum_{j = k + 1}^{δ_{max}} η_{8 j} In Y D_{t - j} + \sum_{i = 1}^{k} η_{9 i} {In ICT}_{t - i} + \sum_{j = k + 1}^{δ_{max}} η_{10 j} {In ICT}_{t - j} + u_{4 t} \end{array}

\begin{array}{l} {In ICT}_{t} = ϑ_{0} + \sum_{i = 1}^{k} ϑ_{1 i} In Q_{t - i} + \sum_{j = k + 1}^{δ_{max}} ϑ_{2 j} In Q_{t - j} + \sum_{i = 1}^{k} ϑ_{3 i} In P_{t - i} \\ + \sum_{j = k + 1}^{δ_{max}} ϑ_{4 j} In P_{t - j} + \sum_{i = 1}^{k} ϑ_{5 i} In Y_{t - i} + \sum_{j = k + 1}^{δ_{max}} ϑ_{6 j} In Y_{t - j} + \sum_{i = 1}^{k} ϑ_{7 i} In Y D_{t - i} \\ + \sum_{j = k + 1}^{δ_{max}} ϑ_{8 j} In Y D_{t - j} + \sum_{i = 1}^{k} ϑ_{9 i} {In ICT}_{t - i} + \sum_{j = k + 1}^{δ_{max}} ϑ_{10 j} {In ICT}_{t - j} + u_{5 t} \end{array}

The null hypothesis of the absence of causality is rejected if the probability value (p value) is less than 10%. Hence, in equation (2), Granger causality from ${ICT}_{t}$ to $Q_{t}$ is noted if $α_{9 i} \neq 0 \forall i$ , and in equation (6) Granger causality from Q_t to ${ICT}_{t}$ is noted if $ϑ_{1 i} \neq 0 \forall i$ . The stability of the VAR model is confirmed if the inverse roots of the characteristic polynomial lie within the positive and negative unit circle. Stability of the model can be achieved by the use of either optimal lags in the VAR specification or the inclusion of the appropriate exogenous instruments such as the trend term, structural break dummies, or seasonality dummies.

Data

The data for visitor arrivals, real effective exchange rate, destination per capita GDP, mobile subscriptions, and fixed broadband subscriptions are gathered from the World Development Indicators and Global Development Finance database (World Bank, 2019). The data for visitor arrivals are from 1995 to 2017. The per capita real GDP data are from 1960 to 2017 for China, France, Italy, Mexico, Spain, the United Kingdom, and the United States. For Germany, data on per capita real GDP are from 1970 to 2017, and for Russia, they are from 1989 to 2017. The data for mobile subscriptions are from 1987 to 2017 for China, from 1986 to 2017 for France, from 1985 to 2017 for Germany, Italy, and the United Kingdom, from 1988 to 2017 for Mexico, from 1991 to 2017 for Russia, from 1986 to 2017 for Spain, and from 1984 to 2017 for the United States. Data for the real effective exchange rate are from 1980 to 2017 for China, France, Italy, Mexico, Spain, and the United States, and for Germany and the United Kingdom, they are from 1979 to 2017. The data for world per capita GDP are used as a proxy for source market income effects (Algieri, 2006). These data are extracted from the Federal Reserve Economic Database (FRED, 2019) and covers the period 1960–2017. A balanced panel is created with 23 years of annual data for the nine countries, thus a total of 207 observations with mobile subscriptions as a proxy for ICT. However, due to lack of data on broadband subscription as a proxy for ICT, we have 16 years of data for nine countries, thus a total of 144 observations. All variables are transformed into logarithms to interpret the coefficients as elasticities and minimize standard errors.

Results

Descriptive statistics and correlation matrix

In Table 1, we present the descriptive statistics and correlation matrix. A positive correlation is noted between mobile subscriptions and visitor arrivals, and broadband subscriptions and visitor arrivals.

Table 1.

Descriptive statistics and correlation matrix.

Statistics	Visitor arrivals	Real effective exchange rate	World real GDP per capita	Destination real GDP per capita	Mobile subscriptions	Broadband subscriptions
Panel A: Descriptive statistics
Mean	41,185,816	99.37	8997.06	27,960.38	$1.39 \times 10^{8}$	34,747,224
Median	36,513,000	100	9184.82	34,318.50	66,681,000	15,725,000
Maximum	86,861,000	131.11	10634	53,128.54	$1.47 \times 10^{9}$	$3.94 \times 10^{8}$
Minimum	10,290,000	46.75	7389.72	1227.556	88,526.00	11,000
Standard deviation	19,657,424	13.26	976.90	15,659.58	$2.43 \times 10^{8}$	57,320,858
Skewness	0.62	−0.58	−0.04	−0.42	3.67	3.698647
Kurtosis	2.29	4.96	1.75	1.65	17.14	19.06
Normality	17.92 (<0.01)*	45.09 (<0.01)*	13.49 (<0.01)*	22.05 (<0.01)*	2192.97 (<0.01)*	1875.95 (<0.01)
Panel B: Correlation matrix
Visitor arrivals	1.00
Real effective exchange rate	0.03 (0.67)	1.00
World real GDP per capita	0.29 (<0.01)*	0.02 (0.74)	1.00
Destination real GDP per capita	0.36 (<0.01)*	0.26 (<0.01)*	0.06 (0.44)	1.00
Mobile subscriptions	0.18 (0.02)**	0.29 (<0.01)*	0.24 (<0.01)*	−0.39 (<0.01)*	1.00
Broadband subscriptions	0.29 (<0.01)*	0.37 (<0.01)*	0.35 (<0.01)*	−0.16 (0.05)**	0.93 (<0.01)*	1.00

Source: Authors’ estimation in Eviews 10.

Note: p Value in parenthesis.

*Significant at 1%.

**Significant at 5%.

Cross-sectional dependence and unit root

Results in Table 2 indicate the presence of cross-sectional dependence. Subsequently, we examine the presence of unit roots, and the results are reported in Table 3. We note from the unit root test results that the maximum order of integration is one, and a mix of order of integration exists between series.

Table 2.

Cross-sectional dependence test.

Null hypothesis	Model	Breusch–Pagan LM	Pesaran scaled LM	Pesaran CD
No cross-sectional dependence in residuals	Mobile	182.56 [36] (<0.01)*	17.27 [36] (<0.01)*	9.06 [36] (<0.01)*
	Broadband	139.43 [36] (<0.01)*	12.19 [36] (<0.01)*	7.79 [36] (<0.01)*

Source: Authors’ estimation in Eviews 10.

Note: LM: Lagrange multiplier; CD: Cross-section Dependence. Degrees of freedom in square parenthesis, p value in round parenthesis.

*Significant at 1%.

Table 3.

Unit root test.

Variables	LLC		IPS		Fisher-ADF		Fisher-PP		CADF
Variables	Level	1st diff.	Level	1st diff.	Level	1st diff.	level	1st diff.	level	1st diff.
$ln Q_{i, t}$	0.41 [1](0.66)	−8.10 [0](<0.01)*	2.82 [1](0.99)	−7.92 [0](<0.01)*	10.06 [1](0.93)	90.31 [0](<0.01)*	23.46 [1](0.17)	89.78 [0](<0.01)*	−1.59 [2](0.71)	−2.36 [1](0.03)**
$ln P_{i, t}$	−1.99 [0](0.02)**	−9.14 [1](<0.01)*	−1.89 [0](0.02)**	−7.47 [1](<0.01)*	29.37 [0](0.04)**	84.19 [1](<0.01)*	25.20 [0](0.12)	80.06 [1](<0.01)*	−2.43 [1](0.01)**	−2.57 [1](<0.01)*
$ln Y_{i, t}$	−2.05 [0](0.02)**	−11.32 [0](<0.01)*	2.38 [0](0.99)	−8.81 [0](<0.01)*	4.00 [0](0.99)	99.76 [0](<0.01)*	6.79 [0](0.99)	116.22 [0](<0.01)*	2.61 [1](0.99)	−2.31 [1](0.03)**
$ln Y D_{i, t}$	−1.84 [2](0.03)**	−5.20 [1](<0.01)*	−1.29 [2](0.09)***	−4.79 [1](<0.01)*	25.79 [2](0.11)	56.69 [2](<0.01)*	29.83 [2](0.04)**	72.31 [2](<0.01)*	−2.59 [1](<0.01)*	−2.42 [2](0.02)**
$ln {MOB}_{i, t}$	−7.51 [4](<0.01)*	−9.19 [4](<0.01)*	−14.12 [4](<0.01)*	−7.12 [4](<0.01)*	178.23 [4](<0.01)*	176.68 [4](<0.01)*	511.63 [4](<0.01)*	23.81 [4](0.17)	−2.87 [1](<0.01)*	−3.07 [2](<0.01)*
$ln {BB}_{i, t}$	−8.34 [9](<0.01)*	−14.39 [9](<0.01)*	−8.10 [9](<0.01)*	−11.85 [9](<0.01)*	89.18 [9](<0.01)*	106.33 [9](<0.01)*	186.09 [9](<0.01)*	116.09 [9](0.17)*	−3.54 [1](<0.01)*	−2.54 [1](0.01)**

Source: Authors’ estimation in Eviews 10 and STATA 13.

Note: Lag length in square parenthesis, p value in round parenthesis. LLC: Levin–Lin–Chu; IPS: Im–Pesaran–Shin; FADF: Fisher–Augmented Dickey–Fuller test; FPP: Fisher–Phillips–Perron test; CADF: Covariate augmented Dickey–Fuller test. Test assumes intercept only. CADF test is based on Pesaran.

*Stationary at 1% level.

**Stationary at 5% level.

***Stationary at 10% level.

Cointegration and poolability

Based on the Hausman test results, we note that the series are cointegrated. This is confirmed via the Pedroni and Kao cointegration tests (Table 4). Additionally, the null hypothesis of long-run homogeneity cannot be rejected, and thus, the PMG method is appropriate for further estimation (Table 5).

Table 4.

Cointegration tests.

Mobile subscriptions model
Panel A: Pedroni cointegration
Alternative hypothesis: common AR coefficients (within-dimension)
	Statistic	Prob.	Weighted statistic	Prob.
Panel v statistic	1.29***	0.09	1.25***	0.09
Panel ρ statistic	1.08	0.86	0.65	0.74
Panel PP statistic	−0.39	0.34	−1.57**	0.05
Panel ADF statistic	−3.16*	<0.01	−3.03*	<0.01
Alternative hypothesis: individual AR coefficients (between-dimension)
	Statistic	Prob.
Group ρ statistic	1.94	0.97
Group PP statistic	−1.09	0.13
Group ADF statistic	−3.23*	<0.01
Panel B: Kao cointegration
ADF statistic	−2.63*	<0.01
Broadband subscriptions model
Panel A: Pedroni cointegration
Alternative hypothesis: common AR coefficients (within-dimension)
	Statistic	Prob.	Weighted statistic	Prob.
Panel v statistic	−1.31	0.91	−0.59	0.72
Panel ρ statistic	2.94	0.99	2.46	0.99
Panel PP statistic	−0.49	0.31	−2.44*	<0.01
Panel ADF statistic	0.29	0.62	−1.51***	0.07
Alternative hypothesis: individual AR coefficients (between-dimension)
	Statistic	Prob.
Group ρ statistic	3.62	0.99
Group PP statistic	−5.68*	<0.01
Group ADF statistic	−2.65*	<0.01
Panel B: Kao cointegration
ADF statistic	−2.53*	<0.01

Source: Authors’ estimation using Eviews 10.

*Cointegrated at 1% level.

**Cointegrated at 5% level.

***Cointegrated at 10% level.

Table 5.

Hausman poolability test.

Null hypothesis	Model	Test statistic	p Value
Long-run poolability is appropriate	Mobile	$χ^{2} (4)$ = 0.28	0.99
	Broadband	$χ^{2} (4)$ = 0.37	0.98

Source: Authors’ estimation using Stata 13.

Long run and short run

The results with mobile and broadband subscriptions are reported in Tables 6 and 7, respectively. In both cases, the optimal specification is based on the Akaike information criterion and the lag length of 1 is applied.

Table 6.

PMG long-run and short-run estimates—mobile subscriptions model.

Variable	Coefficient	Standard error	T statistic	p Value
Panel A: Long run
$ln P_{i, t}$	−0.71***	0.1220	−5.7960	<0.01
$ln Y_{i, t}$	1.58***	0.3946	3.9948	<0.01
$ln Y D_{i, t}$	−0.41	0.2824	−1.4551	0.15
$ln {MOB}_{i, t}$	0.04**	0.0152	2.4481	0.02
Panel B: Short run
${ECT}_{i, t - 1}$	−0.42***	0.0933	−4.4530	<0.01
$Δ ln P_{i, t}$	0.17**	0.0911	1.9450	0.05
$Δ ln Y_{i, t}$	0.70	0.4580	1.5421	0.12
$Δ ln Y D_{i, t}$	0.57***	0.2660	2.1595	0.03
$Δ ln {MOB}_{i, t}$	0.06	0.0587	1.1174	0.26
Constant	4.05***	0.9720	4.1679	<0.01
Trend	0.01	0.0054	1.0206	0.30
Panel C: Model statistics
Overall R ² = 0.9999; adj. overall R ² = 0.9988; SER = 0.0566; $χ_{N}^{2}$ = 7.87; p value = 0.03. LL = 393.9802; AIC = −3.1592; Specification: ARDL (1,1,1,1)

Note: N-normality test. AIC: Akaike information criterion; PMG: pooled mean group; ARDL: autoregressive distributed lag; SER: standard error of regression; LL: log likelihood.

Source: Authors’ estimation using Eviews 10 and Stata 13.

*** and ** refers to significance level at 1% and 5%, respectively.

Table 7.

PMG long-run and short-run estimates—broadband subscriptions model.

Variable	Coefficient	Standard error	T statistic	p Value
Panel A: Long run
$ln P_{i, t}$	−0.88***	0.2033	−4.3374	<0.01
$ln Y_{i, t}$	1.83***	0.2414	7.5926	<0.01
$ln Y D_{i, t}$	0.04	0.1261	0.3176	0.75
$ln {BB}_{i, t}$	0.11**	0.0436	2.5928	0.01
Panel B: Short run
${ECT}_{i, t - 1}$	−0.38***	0.1175	−3.2467	<0.01
$Δ ln P_{i, t}$	0.21	0.1811	1.1663	0.24
$Δ ln Y_{i, t}$	0.41	0.5329	0.7750	0.44
$Δ ln Y D_{i, t}$	0.66**	0.3029	2.2106	0.02
$Δ ln {BB}_{i, t}$	0.03	0.0303	1.0790	0.28
Constant	0.98***	0.3227	3.0624	<0.01
Panel C: Model statistics
Overall R ² = 0.9853; adj. overall R ² = 0.9700; SER = 0.0498; $χ_{N}^{2}$ = 3.73; p value = 0.15. LL = 301.4915; AIC = −3.3818; Specification: ARDL (1,1,1,1)

Note: N-normality test. AIC: Akaike information criterion; PMG: pooled mean group; ARDL: autoregressive distributed lag; SER: standard error of regression; LL: log likelihood.

Source: Authors’ estimation using Eviews 10 and Stata 13.

*** and ** refers to significance level at 1% and 5%, respectively.

In the mobile subscriptions model, the elasticity of mobile subscriptions with respect to visitor arrivals is 0.04. This implies that a 1% increase in mobile subscriptions would increase international arrivals by 0.04%, ceteris paribus. The price coefficient is −0.71, which indicates an inelastic price. The income elasticity is 1.58 which means that the countries are luxury destinations (Crouch, 1994). The ECT indicates that about 42% of disequilibrium errors are corrected each year and that equilibrium is achieved in about 2.4 (1/0.42) years. Notably, the destination’s income coefficient (0.57) is statistically significant in the short run only (Table 6).

In Table 7, broadband subscriptions is used as a proxy for ICT. We note its elasticity with respect to visitor arrivals is 0.11. Thus, a 1% increase in mobile subscriptions would increase international arrivals by 0.11%, ceteris paribus. The price elasticity is estimated at −0.88, which is also price inelastic, and the income elasticity is estimated at 1.83. Based on the coefficient of the ECT, about 38% of disequilibrium is corrected each year. Further, the destinations’ income has a short-run effect only. Interestingly, the outcomes are consistent with the results obtained in Table 6, where mobile subscriptions is used as a proxy for ICT.

Causality

Causality based on the Dumitrescu and Hurlin (2012) and Toda and Yamamoto (1995) are reported in Table 8. In both cases, mobile and broadband subscriptions as proxies for ICT indicate a unidirectional causality, which means that ICT cause visitor arrivals and hence tourism demand. Additionally, it is noted that ICT cause destination’s real GDP, which duly supports technology-led growth hypothesis.

Table 8.

Causality tests.

X variables	Toda–Yamamoto (1995)					Dumitrescu–Hurlin (2012)
X variables	$ln Q_{i, t}$	$ln P_{i, t}$	$ln Y_{i, t}$	$ln Y D_{i, t}$	$ln I C T_{i, t}$	$ln Q_{i, t}$	$ln P_{i, t}$	$ln Y_{i, t}$	$ln Y D_{i, t}$	$ln I C T_{i, t}$
Panel A: Mobile subscriptions model
$ln Q_{i, t}$	—	$χ^{2} (5)$ = 16.54[<0.01]*	$χ^{2} (5)$ = 6.10[0.29]	$χ^{2} (5)$ = 3.02[0.69]	$χ^{2} (5)$ = 4.95[0.41]	—	$χ^{2} (3)$ = 4.73[0.32]	$χ^{2} (3)$ = 3.20[0.77]	$χ^{2} (3)$ = 5.97[0.04]**	$χ^{2} (3)$ = 3.66[0.92]
$ln P_{i, t}$	$χ^{2} (5)$ = 3.61[0.60]	—	$χ^{2} (5)$ = 13.26[0.02]**	$χ^{2} (5)$ = 15.78[<0.01]*	$χ^{2} (5)$ = 20.14[<0.01]*	$χ^{2} (3)$ = 4.18[0.59]	—	$χ^{2} (3)$ = 1.36[0.07]***	$χ^{2} (3)$ = 2.76[0.52]	$χ^{2} (3)$ = 5.09[0.19]
$ln Y_{i, t}$	$χ^{2} (5)$ = 21.92[<0.01]*	$χ^{2} (5)$ = 1.77[0.87]	—	$χ^{2} (5)$ = 53.04[<0.01]*	$χ^{2} (5)$ = 13.86[0.02]**	$χ^{2} (3)$ = 7.24[<0.01]*	$χ^{2} (3)$ = 3.75[0.86]	—	$χ^{2} (3)$ = 4.63[0.37]	$χ^{2} (3)$ = 2.90[0.59]
$ln Y D_{i, t}$	$χ^{2} (5)$ = 11.83[0.03]**	$χ^{2} (5)$ = 3.15[0.67]	$χ^{2} (5)$ = 2.31[0.81]	—	$χ^{2} (5)$ = 27.98[<0.01]*	$χ^{2} (3)$ = 3.02[0.66]	$χ^{2} (3)$ = 6.68[<0.01]*	$χ^{2} (3)$ = 3.85[0.26]	—	$χ^{2} (3)$ = 4.07[0.66]
$ln I C T_{i, t}$	$χ^{2} (5)$ = 9.44[0.04]***	$χ^{2} (5)$ = 2.83[0.73]	$χ^{2} (5)$ = 3.69[0.38]	$χ^{2} (5)$ = 12.86[0.02]*	—	$χ^{2} (3)$ = 6.87[<0.01]*	$χ^{2} (3)$ = 8.83[<0.01]*	$χ^{2} (3)$ = 1.84[0.15]	$χ^{2} (3)$ = 8.08[<0.01]*	—
All	$χ^{2} (20)$ = 32.32[0.04]**	$χ^{2} (20)$ = 25.91[0.17]	$χ^{2} (20)$ = 32.78[0.04]*	$χ^{2} (20)$ = 76.73[<0.01]*	$χ^{2} (20)$ = 73.84[<0.01]*	—	—	—	—	—
Panel B: Broadband subscriptions model
$ln Q_{i, t}$	—	$χ^{2} (1)$ = 0.04[0.83]	$χ^{2} (1)$ = 0.24[0.62]	$χ^{2} (1)$ = 0.58[0.44]	$χ^{2} (1)$ = 3.16[0.05]**	—	$χ^{2} (1)$ = 2.96[<0.01]*	$χ^{2} (1)$ = 0.20[0.11]	$χ^{2} (1)$ = 5.89[<0.01]*	$χ^{2} (1)$ = 1.83[0.34]
$ln P_{i, t}$	$χ^{2} (1)$ = 0.02[0.88]	—	$χ^{2} (1)$ = 0.64[0.42]	$χ^{2} (1)$ = 9.10[<0.01]*	$χ^{2} (1)$ = 0.47[0.49]	$χ^{2} (1)$ = 1.19[0.89]	—	$χ^{2} (1)$ = 0.20[0.11]	$χ^{2} (1)$ = 7.14[<0.01]*	$χ^{2} (1)$ = 6.04[<0.01]*
$ln Y_{i, t}$	$χ^{2} (1)$ = 7.39[<0.01]*	$χ^{2} (1)$ = 0.22[0.64]	—	$χ^{2} (1)$ = 19.72[<0.01]*	$χ^{2} (1)$ = 4.07[0.04]**	$χ^{2} (1)$ = 1.57[0.43]	$χ^{2} (1)$ = 2.56[0.01]**	—	$χ^{2} (1)$ = 2.07[0.11]	$χ^{2} (1)$ = 3.67[<0.01]*
$ln Y D_{i, t}$	$χ^{2} (1)$ = 0.22[0.64]	$χ^{2} (1)$ = 0.46[0.49]	$χ^{2} (1)$ = 0.13[0.82]	—	$χ^{2} (1)$ = 20.80[<0.01]*	$χ^{2} (1)$ = 2.43[0.02]**	$χ^{2} (1)$ = 4.12[<0.01]*	$χ^{2} (1)$ = 0.72[0.49]	—	$χ^{2} (1)$ = 8.67[<0.01]*
$ln I C T_{i, t}$	$χ^{2} (1)$ = 3.46[0.03]**	$χ^{2} (1)$ = 4.67[0.03]**	$χ^{2} (1)$ = 1.25[0.26]	$χ^{2} (1)$ = 20.80[<0.01]*	—	$χ^{2} (1)$ = 3.57[<0.01]*	$χ^{2} (1)$ = 4.27[<0.01]*	$χ^{2} (1)$ = 0.18[0.12]	$χ^{2} (1)$ = 2.47[0.05]***	—
All	$χ^{2} (1)$ = 10.05[0.05]**	$χ^{2} (1)$ = 5.46[0.24]	$χ^{2} (1)$ = 1.35[0.85]	$χ^{2} (1)$ = 41.08[<0.01]*	$χ^{2} (1)$ = 32.81[<0.01]*	—	—	—	—	—

Source: Authors’ estimation using Eviews 10.

Note: p Value in square parenthesis.

*Significant causality at 1%.

**Significant causality at 5%.

***Significant causality at 10%.

Discussion and conclusions

The study set out to examine the effect of ICT on tourism demand in nine major tourist destinations, based on the largest number of visitor arrivals. It was noted that cointegration exists between ICT and tourism demand. Mobile and broadband subscriptions were used to proxy for ICT. Additionally, price, income, and destination’s income were included in the estimation. Our sample consisted of a balanced panel with 207 observations in the case of mobile subscription and 144 observations in the case of broadband subscription. Cross-sectional dependence, unit root, and cointegration tests were carried out, accordingly.

We note consistent results in both cases. The short- and the long-run results with respect to visitor arrivals indicate that ICT is positively associated, price is negatively associated, and source country income is positively associated—the latter implies that the major tourist destinations are of luxury type. Moreover, it is noted that the destination country’s growth has only a short-run positive effect on tourism demand. Further, we note a unidirectional causality from ICT to tourism demand and destination country’s income.

The study has some clear contributions. First, the study reinforces the role of technology as a driver of tourism demand and confirms that the major countries are luxury destinations. Second, the study examines major tourism destinations from the perspective of tourism-ICT literature, which was previously absent. Third, consistent results based on the two measures of ICT indicate that both mobile and broadband subscriptions are correlated, and thus can be used as reliable indicators of ICT. Additionally, the study reinforces the ICT-led growth hypothesis in the major tourist destinations. Our analysis underscores that ICT development as a subset of the broader definition of infrastructure improves visitor arrivals to the major destinations, holding all other things constant.

Future research can consider the links between the various digital platforms and visitor arrivals. The role of ICT in the context of platforms like social media can be analyzed in relation to tourism demand. Moreover, given that ICT is a broad term, additional indicators to proxy ICT development can be identified and examined in relation to tourism demand. Of course, with a larger data set, it is plausible to explore nonlinearity and asymmetries in tourism demand, namely, ICT, and subsequently examine the threshold levels of ICT.

Our results clearly indicate a strong link between ICT and tourism demand in major tourist destinations (China, France, Germany, Italy, Mexico, Russia, Spain, the United Kingdom, and the United States). Therefore, by enhancing the quality and coverage of technology infrastructure, these destinations can effectively promote their tourism sector at a greater scale. Using ICT tools to scale up digital tourism will be an effective strategy to improve tourism demand in the future. Additionally, for future consideration, we propose the possibility of implementing artificial intelligence and virtual tourism experience in tourism industry, especially in countries where technology and connectivity are par excellence.

Footnotes

Acknowledgements

The authors sincerely thank the editors, Professors Rodolfo Baggio and Davide Provenzano, and the anonymous reviewers for their comments and suggestions. The usual disclaimer applies.

Declaration of conflicting interests

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

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Ronald Ravinesh Kumar

Note

References

Adeola

Evans

(2019a) Digital tourism: Mobile phones, internet and tourism in Africa. Tourism Recreation Research 44(2): 190–202.

Adeola

Evans

(2019b) ICT, infrastructure, and tourism development in Africa. Tourism Economics. Available at: https://doi.org/10.1177/1354816619827712 (accessed 10 March 2019).

Alam

Paramati

(2016) The impact of tourism on income inequality in developing economies: Does Kuznets curve hypothesis exist? Annals of Tourism Research 61: 111–126.

Algieri

(2006) An econometric estimation of the demand for tourism: The case of Russia. Tourism Economics 12(1): 5–20.

Andereck

Nyaupane

(2011) Exploring the nature of tourism and quality of life perceptions among residents. Journal of Travel Research 50(3): 248–260.

Andereck

Valentine

Knopf

, et al. (2005) Residents’ perceptions of community tourism impacts. Annals of Tourism Research 32(4): 1056–1076.

Anderson

Van Wincoop

(2003) Gravity with gravitas: A solution to the border puzzle. American Economic Review 93(1): 170–192.

Asafu-Adjaye

Byrne

Alvarez

(2016) Economic growth, fossil fuel and non-fossil consumption: A Pooled Mean Group analysis using proxies for capital. Energy Economics 60: 345–356.

Benyon

Quigley

O’Keefe

, et al. (2014) Presence and digital tourism. AI & Society 29(4): 521–529.

10.

Buhalis

(1998) Strategic use of information technologies in the tourism industry. Tourism Management 19(5): 409–421.

11.

Buhalis

Law

(2008) Progress in information technology and tourism management: 20 years on and 10 years after the Internet—The state of e-Tourism research. Tourism Management 29(4): 609–623.

12.

Clemons

Hann

Hitt

(2002) Price dispersion and differentiation in online travel: An empirical investigation. Management Science 48(4): 534–549.

13.

Croes

(2014) The role of tourism in poverty reduction: An empirical assessment. Tourism Economics 20(2): 207–226.

14.

Crouch

(1994) The study of international tourism demand: A survey of practice. Journal of Travel Research 32(4): 41–55.

15.

Dogru

Bulut

(2018) Is tourism an engine for economic recovery? Theory and empirical evidence. Tourism Management 67: 425–434.

16.

Dogru

Sirakaya-Turk

Crouch

(2017) Remodeling international tourism demand: Old theory and new evidence. Tourism Management 60: 47–55.

17.

Dumitrescu

Hurlin

(2012) Testing for Granger non-causality in heterogeneous panels. Economic Modelling 29(4): 1450–1460.

18.

Durbarry

(2008) Tourism taxes: Implications for tourism demand in the UK. Review of Development Economics 12(1): 21–36.

19.

Dyer

Gursoy

Sharma

, et al. (2007) Structural modeling of resident perceptions of tourism and associated development on the Sunshine Coast, Australia. Tourism Management 28(2): 409–422.

20.

Federal Reserve Economic Database (2019) FRED Economic Data. Federal Reserve Bank of St. Louis. USA. Available at: https://fred.stlouisfed.org/ (accessed 28 December 2018).

21.

Fodness

Murray

(1997) Tourist information search. Annals of Tourism Research 24(3): 503–523.

22.

Gelb

Sundaram

(2002) Adapting to “word of mouse.” Business Horizons 45(4): 15–20.

23.

Goh

Wong

Yew

(2018) Are merchandise exports and FDI inflows an engine of growth for ASEAN-10 countries? New evidence using the Pooled Mean Group estimation method. Margin: The Journal of Applied Economic Research 12(4): 414–430.

24.

Harrison-Walker

(2001) The measurement of word-of-mouth communication and an investigation of service quality and customer commitment as potential antecedents. Journal of Service Research 4(1): 60–75.

25.

Pesaran

Shin

(2003) Testing for unit roots in heterogeneous panels. Journal of Econometrics 115(1): 53–74.

26.

Jorgenson

(2001) Information technology and the US economy. The American Economic Review 91(1): 1–32.

27.

Jorgenson

(2011) The rise of developing Asia and the new economic order. Journal of Policy Modeling 33(5): 698–716.

28.

Khoshnevis-Yazdi

Dariani

(2019) CO2 emissions, urbanisation and economic growth: Evidence from Asian countries. Economic Research-Ekonomska Istraživanja 32(1): 510–530.

29.

Kao

(1999) Spurious regression and residual-based tests for cointegration in panel data. Journal of Econometrics 90(1): 1–44.

30.

Kim

Lee

(2017) Role of tourism price in attracting international tourists: The case of Japanese inbound tourism from South Korea. Journal of Destination Marketing & Management 6(1): 76–83.

31.

Kotiloglu

Lappas

Pelechrinis

, et al. (2017) Personalized multi-period tour recommendations. Tourism Management 62: 76–88.

32.

Kumar

(2014) Exploring the role of technology, tourism and financial development: An empirical study of Vietnam. Quality & Quantity 48(5): 2881–2898.

33.

Kumar

(2012) Exploring the nexus between information and communications technology, tourism and growth in Fiji. Tourism Economics 18(2): 359–371.

34.

Kumar

Patel

(2015) Accounting for telecommunications contribution to economic growth: A study of Small Pacific Island States. Telecommunications Policy 39(3–4): 284–295.

35.

Kumar

Stauvermann

Samitas

(2016) The effects of ICT* on output per worker: A study of the Chinese economy. Telecommunications Policy 40(2): 102–115.

36.

Lanza

Pigliaru

(2000) Why are tourism countries small and fast-growing? In: Fossati

and Panella

(eds) Tourism and Sustainable Economic Development. Boston: Springer, pp. 57–69.

37.

Lam

Shiu

(2010) Economic growth, telecommunications development and productivity growth of the telecommunications sector: Evidence around the world. Telecommunications Policy 34(4): 185–199.

38.

Law

Chan

ICC

Wang

(2018) A comprehensive review of mobile technology use in hospitality and tourism. Journal of Hospitality Marketing & Management 27(6): 626–648.

39.

Levin

Lin

Chu

CSJ

(2002) Unit root tests in panel data: Asymptotic and finite-sample properties. Journal of Econometrics 108(1): 1–24.

40.

Buhalis

(2005) Predicting internet usage for travel bookings in China. Information and Communication Technologies in Tourism 2005: 429–439.

41.

Song

Witt

(2005) Recent developments in econometric modeling and forecasting. Journal of Travel Research 44(1): 82–99.

42.

Lim

(1997) Review of international tourism demand models. Annals of Tourism Research 24(4): 835–849.

43.

Lin

Liu

Song

(2015) Modeling and forecasting Chinese outbound tourism: An econometric approach. Journal of Travel & Tourism Marketing 32(1–2): 34–49.

44.

Luo

Feng

Cai

(2004) Information search behavior and tourist characteristics: The internet vis-à-vis other information sources. Journal of Travel & Tourism Marketing, 17(2/3): 15–25.

45.

Madden

Savage

(1998) CEE telecommunications investment and economic growth. Information Economics and Policy 10(2): 173–195.

46.

Mills

Law

(2004) Handbook of Consumer Behaviour, Tourism and the Internet. New York: Harworth Hospitality Press.

47.

Narayan

(2004) Fiji’s tourism demand: The ARDL approach to cointegration. Tourism Economics 10(2): 193–206.

48.

Narayan

Prasad

, et al. (2010) Tourism and economic growth: A panel data analysis for Pacific Island countries. Tourism Economics 16(1): 169–183.

49.

Navío-Marco

Ruiz-Gómez

Sevilla-Sevilla

(2018) Progress in information technology and tourism management: 30 years on and 20 years after the internet—Revisiting Buhalis & Law’s landmark study about eTourism. Tourism Management 69: 460–470.

50.

Niininen

Buhalis

March

(2007) Customer empowerment in tourism through consumer centric marketing (CCM). Qualitative Market Research: An International Journal 10(3): 265–281.

51.

O’Connor

(1999) Electronic information distribution in tourism and hospitality. Wallingford: CAB International.

52.

O’Connor

Buhalis

Frew

(2001) The transformation of tourism distribution channels through information technology. In: Laws

Buhalis

(eds) Tourism Distribution Channels: Practices, Issues and Transformations. London: Thompson, pp. 332–350.

53.

Pablo-Romero

MDP

Molina

(2013) Tourism and economic growth: A review of empirical literature. Tourism Management Perspectives 8: 28–41.

54.

Pedroni

(2004) Panel cointegration: Asymptotic and finite sample properties of pooled time series tests with an application to the PPP hypothesis. Econometric Theory 20(3): 597–625.

55.

Pesaran

(2007) A simple panel unit root test in the presence of cross-section dependence. Journal of Applied Econometrics 22(2): 265–312.

56.

Pesaran

Shin

Smith

(1999) Pooled mean group estimation of dynamic heterogeneous panels. Journal of the American Statistical Association 94(446): 621–634.

57.

Pesaran

Smith

(1995) Estimating long-run relationships from dynamic heterogeneous panels. Journal of Econometrics 68(1): 79–113.

58.

Pohjola

(2002) The new economy in growth and development. Oxford Review of Economic Policy 18(3): 380–396.

59.

Ridderstaat

Oduber

Croes

, et al. (2014) Impacts of seasonal patterns of climate on recurrent fluctuations in tourism demand: Evidence from Aruba. Tourism Management 41: 245–256.

60.

Seetaram

(2012) Estimating demand elasticities for Australia’s international outbound tourism. Tourism Economics 18(5): 999–1017.

61.

Seetaram

Forsyth

Dwyer

(2016) Measuring price elasticities of demand for outbound tourism using competitiveness indices. Annals of Tourism Research 56: 65–79.

62.

Shahzad

SJH

Shahbaz

Ferrer

Kumar

(2017) Tourism-led growth hypothesis in the top ten tourist destinations: New evidence using the quantile-on-quantile approach. Tourism Management 60: 223–232.

63.

Sheldon

(1993) Forecasting tourism: Expenditures versus arrivals. Journal of Travel Research 32(1): 13–20.

64.

Smeral

(2017) Tourism forecasting performance considering the instability of demand elasticities. Journal of Travel Research 56(7): 913–926.

65.

Song

(2008) Tourism demand modelling and forecasting—A review of recent research. Tourism Management 29(2): 203–220.

66.

Song

Witt

, et al. (2010) Tourism demand modelling and forecasting: How should demand be measured? Tourism Economics 16(1): 63–81.

67.

Song

Wong

Chon

(2003) Modelling and forecasting the demand for Hong Kong tourism. International Journal of Hospitality Management 22(4): 435–451.

68.

Spencer

(2019) ICT and Caribbean tourism. In: Spencer

(ed), Travel and Tourism in the Caribbean – Challenges and Opportunities for Small Island Developing States. Cham: Palgrave Macmillan, pp. 95–114.

69.

Stiroh

(2002) Are spillovers driving the new economy? Review of Income and Wealth 48(1): 33–57.

70.

Strobel

(2016) ICT intermediates and productivity spillovers—Evidence from German and US manufacturing sectors. Structural Change and Economic Dynamics 37(C): 147–163.

71.

Tan

GWH

Lee

Lin

, et al. (2017) Mobile applications in tourism: The future of the tourism industry? Industrial Management & Data Systems 117(3): 560–581.

72.

Toda

Yamamoto

(1995) Statistical inference in vector autoregressions with possibly integrated processes. Journal of Econometrics 66(1–2): 225–250.

73.

Tugcu

(2014) Tourism and economic growth nexus revisited: A panel causality analysis for the case of the Mediterranean Region. Tourism Management 42: 207–212.

74.

Vanegas

Sr Gartner

Senauer

(2015) Tourism and poverty reduction: An economic sector analysis for Costa Rica and Nicaragua. Tourism Economics 21(1): 159–182.

75.

Wang

de Almeida Correia

van Arem

, et al. (2018) Understanding travellers’ preferences for different types of trip destination based on mobile internet usage data. Transportation Research Part C: Emerging Technologies 90: 247–259.

76.

Ward

Zheng

(2016) Mobile telecommunications service and economic growth: Evidence from China. Telecommunications Policy 40: 89–101.

77.

Werthner

Ricci

(2004) E-commerce and tourism. Communications of the ACM 47(12): 101–105.

78.

World Bank (2019) World development indicators and global development finance. Washington: Author.

79.

Yoon

Gursoy

Chen

(2001) Validating a tourism development theory with structural equation modeling. Tourism Management 22(4): 363–372.