Examining the relationship between tourism, trade, and income: evidence from Vietnam

Vietnam's tourism industry has been one of the economy's key growth drivers for years. This study examines the relationship between international tourist arrivals, bilateral trade, and income in Vietnam by applying a cointegration model with the panel data of Vietnam and its major global tourism markets. The results show that export and per capita income significantly influence Vietnam's international tourist arrivals from 2008 to 2019. The study also uses causality techniques to test for the direction of causality and found a two-way causal effect from exports and income to international tourist arrivals. These findings suggest that future economic policy should focus more on tourism and trade-related to sustain the growth of the tourism industry and generate more foreign exchange earnings for Vietnam.


Introduction
Tourism is one of the largest industries in the world, driving socioeconomic development and creating jobs. In 2020, the tourism industry contributed one in 10 jobs globally, 10.4% of the total global GDP, and created one in four new jobs. Tourism plays a vital role in promoting prosperity and empowering women, youth, and other groups in society. The benefits of tourism directly affect GDP and employment, and indirect benefits through supply chain linkages with other sectors and accompanying impacts (WTTC, 2020).
For Vietnam, since the Tourism Development Strategy (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010) was approved, the tourism industry has achieved impressive results. The international visitors increased from about 2 million in 2000 to 5 million in 2010, and the annual average growth rate in the corresponding period was over 9% (ITDR, 2021). In 2019, according to the Ministry of Culture, Sports, and Tourism, Vietnam welcomed 18 million international visitors and served about 85 million domestic tourists. The average growth rate of the tourism industry for three consecutive years is 22%. Vietnam is one of the ten fastest-growing tourism countries in the world. In addition, the competitiveness of Vietnam's tourism has continuously improved in the World Economic Forum rankings (WEF) rankings. Vietnam's tourism competitiveness increased from 75/141 economies (in 2015) to 63/140 (in 2019). Thus, it can be affirmed that tourism is a fast-growing economic sector and a spearhead of the Vietnamese economy.
There have been several studies on the determinants of tourism demand. Among these studies, an analysis of the relationship between tourism demand and macroeconomic factors such as exchange rate, exports, imports, income, economic growth, and foreign direct investment has received much attention. Existing studies on the macroeconomic determinants of tourism demand have mainly considered trade, income, and other macroeconomic variables. The question of how both trade and income of a particular country impact on its tourism demand has not received much attention. This study contributes to the existing literature by evaluating the linkage between tourism demand, trade, and income, with a case study of Vietnam. This allows us to assess whether trade and income help explain the tourism demand of a particular country such as Vietnam.

IJBES VOL 5 NO 3 (2023) ISSN: 2687-2293
Furthermore, although the tourism industry has such an important role in Vietnam's economy, studies on Vietnam's tourism are limited, particularly at the macro level. Some typical studies are Si & Bang (2020) Shih & Do (2016); etc. These studies focus on analyzing the destination or firm's specific factors of the tourism demand, e.g., consumer behavior, revenue, marketing, destination image, and business efficiency. A few studies use macroeconomic variables such as income, exchange rate, and economic growth. None of the studies on Vietnam's tourism demand considers trade as a determinant. Therefore, this is the first study that accounts for trade as a determinant of tourism demand in Vietnam.
Several reasons make Vietnam become an interesting case study. Firstly, tourism has played an important role in Vietnam's economy. According to the Vietnam National Administration of Tourism, the tourism industry contributed over 9.2% to the country's GDP and created 2.9 million jobs, including 927 thousand direct jobs in 2019 (VNAT, 2020). Secondly, Vietnam is often considered a tourism star in Southeast Asia due to its stunning natural landscapes, rich history and culture, friendly people, and delicious cuisine. Finally, when compared to other Asian countries, Vietnam's tourism industry is still relatively small, but it has been growing rapidly. Vietnam ranked 7th in international tourist arrivals among Asian countries, behind China, Japan, Thailand, Malaysia, Hong Kong, and South Korea. However, Vietnam's growth rate was the highest among these countries, with a 16.2% increase in arrivals compared to the previous year (WTTC, 2019).
The study structure includes: section 2 presents the review of literature and research methods; section 3 analyzes the relationship between tourism and selected macroeconomic variables; and section 4 concludes the study.

Literature review Theoretical and Conceptual Background
Tourism demand can be influenced by multiple factors, including social media attachment (Song & Abukhalifeh, 2021), safety, hygiene, connectivity, availability of information, currency exchange rates (Gupta, Shukla, & Pandiya, 2022), gross domestic product, and the number of beds (Borrego-Domínguez, Isla-Castillo, & Rodríguez-Fernández, 2022) at the tourist destination. In addition, other well-known country-specific factors including culture, infrastructure, natural beauty, people's attitude, population, education, trade, income, commodity prices, and languages, can positively and negatively impact the tourism industry in the world (Khan et al., 2020).
Among the country-specific factors, macroeconomic factors such as exchange rate, exports, imports, economic growth, economic size, and income have been widely used as determinants of tourism demand in a number of studies. For instance, Katircioglu (2011) indicates that there is a long-run relationship between international tourism and net FDI flows in Turkey based on an ARDL model. Tarik Dogru et al. (2019) study the impact of the exchange rate on the balance of tourism trade through the cointegration model. Their results show that a stronger US dollar will worsen the US bilateral travel trade balance with Canada and the UK. Suresh & Tiwari (2017) use a Granger-causality model to analyze the relationship between international tourist arrivals, trade, and economic growth in India. Their findings indicate a two-way causal relationship between trade tourism and economic growth tourism. The result is consistent with the study of Borrego-Domínguez et al. (2022), who find that gross domestic product affects tourism demand in Spain. Santana-Gallego et al. (2016) analyzed the relationship between the numbers of international arrivals and the exports of 195 countries in 2012. They find that a 1% increase in tourist arrivals increases the likelihood of exports by 1 .25% and exports by 9%.
Similarly, Leitão (2010) assesses whether trade affects tourism demand in Portugal. He also finds that bilateral trade, along with immigration, border, and distance between Portugal and its country partners, is a key driver of tourism in Portugal, based on static and dynamic. Using Granger-causality, Kadir and Jusoff (2010) also find a unidirectional causality running from exports, imports, and total trade to international tourism receipts for the case of Malaysia. Lee's study (2012) examines the short-and long-term dynamic interactions between exports, imports, international tourism, and economic growth in Singapore using annual data from 1980 to 2007. He finds that the dynamic interactions of these four variables are complex. Another study by Keum (2010) on Korean trade and tourism indicates that bilateral international tourist flows will take place more when the income or economic size of the two countries is larger, and the distance between the two partners is closer.

Empirical Review
Very few studies assess the interaction between tourism demand and macroeconomic variables in Vietnam. The limited studies include Selvanathan et al. (2021), Puah et al. (2019), Ngo & Ha (2022), and Le, Nguyen, and Tran (2022). In particular, Selvanathan et al. (2021) explain why Vietnam is a favorite destination using exchange rate variables and commodity price differentials between Vietnam and other countries. Puah et al. (2019) apply the gravity model in trade to identify the determinants of tourism demand in Vietnam. The results show that the income of the destination country and the country of origin, the price of tourism, and travel costs are significant and critical macroeconomic factors determining tourism demand in Vietnam. Using bivariate and multivariate wavelet frameworks, Ngo & Ha (2022) investigate the relationship between tourism revenue, the real exchange rate, and economic development in Vietnam from 1995 to 2019. They find a positive and robust interaction between tourism demand, economic growth, and the real exchange rate. A study by Le et al. (2022) evaluates the triangular causal relationship between economic growth, tourism, and economic growth in central Vietnam and finds that there are two-way short-and long-term causal interactions between foreign tourist arrival and economic growth. Although the previous studies have used macroeconomic variables as determinants of tourism demand for Vietnam, none of them has accounted for trade as a determinant of tourism demand in Vietnam. Therefore, it is important to have a study evaluating the interaction between tourism demand, trade and income in Vietnam.
To summarize, the connection between macroeconomic factors and tourism demand has been explored in multiple studies. However, these studies typically only consider either income or trade in conjunction with other variables. This study examines both income and trade as drivers of tourism demand, making a novel contribution to the literature on this topic. Additionally, it's worth noting that this is the first study to include trade as a factor influencing tourism demand in Vietnam.

Research and Methodology
Based on the literature review, to investigate the relationship between trade and tourism demand in Vietnam, this study applies a cointegration model with panel data of Vietnam's international arrivals from its major origin countries between 2008 and 2019 (Appendix 1). Many published studies on tourism demand show the importance of using econometric models to analyze tourist behaviors by estimating demand determinants (Ouerfelli, 2010). Variables in the model include exports, imports, and per capita income of Vietnam and its partner. Specifically, the model is presented as follows: where β is the regression coefficient, i and j denote Vietnam and its partners, t represents the time, and e is the error term. The explanatory variables are explained as follows: Export and import are bilateral export and import between Vietnam and country j. Tourism activities are closely linked to different economic sectors, including trade; thus, studies on trade and tourism also confirmed a positive relationship between these two industries (Santana-Gallego et al., 2016;Lee, 2012;Keum, 2010). Therefore, the coefficients of export and import are expected to have positive signs.
Income is the sum of GDP per capita between Vietnam and country j. The relationship between income and tourism is bi-directional.
In the first direction, tourism development will contribute to economic growth and increase per capita income. In another direction, when income rises, demand for tourism also increases. Hence, previous studies concluded a bi-directional causality between income and tourism in the short term or long term (e.g., De Vita & Kyaw, 2016;Ivanov & Webster, 2007;Brida et al., 2010;Apergis & Payne, 2012). Thus, the sum of GDP per capita is expected to affect the number of international visitors to Vietnam positively.
Data regarding the number of international visitors was obtained from the Vietnam National Administration of Tourism. Information on bilateral trade, including exports and imports, was sourced from the United Nations Comtrade database. The GDP per capita figures for Vietnam and other countries were obtained from the World Bank. All the data series encompass the time frame from 2008 to 2019 and include major originating countries (based on the total number of visitors in the corresponding period), such as Australia, China, Germany, France, the United Kingdom, Indonesia, Japan, Korea, Malaysia, the Philippines, Russia, Singapore, Thailand, and the United States. Variable descriptions, study hypotheses, and data sources are presented in Table 1. The descriptive statistics of the sample are reported in Table 2.  Much economic time series exhibit trending behavior or non-stationarity in the mean. Nonstationary time series data can lead to spurious results, such as an overestimation of the persistence of the series. They can also make it difficult to detect meaningful relationships between variables. Thus, it is important to examine the stationarity characteristic of variables before proceeding to the estimation. Unit root tests are employed in the analysis of time series data to assess the stationarity of a series. These tests determine the existence of a unit root, representing a random trend in the data that leads to non-stationarity. One can establish if a first difference or regression on deterministic time functions is necessary to make the trending data stationary by conducting unit root tests. Several panel unit root tests are widely used in econometric and statistical studies. Most studies often use a combination of multiple panel unit root tests to ensure the robustness of the results. Therefore, this study first uses panel unit root tests of Dickey and Fuller (ADF, 1981); Phillips and Perron (PP, 1988); Levin, Lin, and Chu (LLC, 2002); and Im, Pesaran, and Shin (IPS, 1997) to test if there are unit roots in panel data sets.
Cointegration is a long-run relationship between two or more time series variables, and it is an essential concept in many areas of economics, finance, and international trade. Various methods and techniques have been developed to test for cointegration. The Engle- Granger (1987) two-step approach is an early and popular test for cointegration, which involves examining the stationarity of a regression model's residuals. Despite its widespread use, this approach has drawbacks, such as its susceptibility to the selection of the regression model and the existence of deterministic trends in the data. In response to these shortcomings, numerous alternative cointegration tests have been created, such as the Johansen (1991) maximum likelihood approach, the Phillips-Ouliaris (1990) residual-based test, and the Pedroni (1999Pedroni ( , 2004 panel cointegration test. These methods are more robust and flexible for cointegration testing and have been widely used in many studies. Thus, Pedroni's (2001) panel cointegration test is used in the second step to examine the cointegrating relationship. And finally, to examine the causal relationship between the variables, we use the Granger causality test. According to Granger (1988), if two series are cointegrated, there must be Granger causation in at least one direction. In this study, the hypotheses of the relationship between variables are as follows: (i) Does export and import cause international tourism arrival and vice versa? (ii) Does income cause international tourism arrival and vice versa?

Findings and Discussions
The stationarity test results of all variables in level and first differences are shown in Tables 3 and 4 To check whether these variables are integrated of order one, I(1), the same tests are carried out for the variables in first difference. As shown in Table 4, all the tests suggest that all variables are integrated into one order (I(1)).  In the next step, we proceed with the test for the cointegration between the variables. Several tests have been proposed for panel cointegration, like Pedroni, Kao, and a Fisher-type test using an underlying Johansen methodology (Maddala and Wu, 1999). The Fisher test is simply the combined Johansen test (as for the time series). The Pedroni and Kao tests are based on Engle- Granger's (1987) two-step (residual-based) cointegration tests. Pedroni tests for cointegration allow for heterogeneous intercepts and trend coefficients across cross-sections. Thus, we use both Pedroni and Kao tests to check whether there is a long-run relationship between the selected variables. The null hypothesis of the tests is no cointegration, or the long-run relationships between variables do not exist. The test results are presented in Table 5. As shown in Table 5, all the variables are cointegrated in the long run because most test statistics indicate the rejection of the null hypothesis of no cointegration.
Given the evidence of panel cointegration, the next step is to estimate the long-run relationship among variables. Table 6 provides estimated results for four models (FMOLS, DOLS, fixed effect model, and GMM). Based on the expected signs, all three independent variables are expected to have a positive relationship with the dependent variable. The estimated results show that exports and income impact the number of international visitors to Vietnam at a 1% significance level ( Table 6). The effect of these two variables is similar in both models (FMOLS and DOLS), in which income has a significant impact with a magnitude of 1,363 and 1.618. As for imports, neither model has a relationship with tourist arrivals. This result contradicts research expectations and can be explained as the countries with the largest number of tourist arrivals to Vietnam (China, Korea, and Japan) are located in Asia, with a distance of geographical proximity and a similar culture to Vietnam. Meanwhile, Vietnam has diverse import markets from both Europe and the US. Thus imports and tourism may not have a long-run relationship.
The Panel Generalized Method of Moments (GMM) is frequently utilized in econometrics to verify the robustness of panel data analysis. GMM can effectively manage individual and time effects, address endogenous variables, and produce consistent estimates under diverse assumptions. This makes it a dependable tool for conducting robustness checks and more. Therefore, we use panel GMM for robustness checks in the study. The results of panel GMM are presented in Table 6. As can be seen in table 6, the coefficients estimated from panel GMM are consistent with FMOLS and DOLS models. The coefficients' magnitude, sign, and statistical significance provide the same relationships between the variables of interest.  While cointegration implies causality in at least one direction, it says nothing about the direction of the causal relationship between the variables, as discussed above. Causality may run in either direction, from independent variables to dependent variables or dependent variables to independent variables, or in both directions. We apply the Granger causality tests to test the direction of longrun causality. Results presented in Table 7 show that the pairs of variables tourism -export, tourism -income, and income -export and import-have a two-way relationship with significance levels of at least 10%. This result is consistent with previous studies in this area. Notably, the pair of variables tourism -import is not statistically significant to reject the hypothesis, so we do not present the results. However, it is consistent with the cointegration estimation results in Table 6.

Conclusion
This article studied the interrelationship among tourism arrivals, trade, and income in Vietnam from two perspectives. First, we quantified the impacts of trade and income on the tourism industry from 2008 to 2019 and then estimated the long-term relationship between variables. The empirical evidence suggests a cointegration relationship between export, GDP per capita, and tourism arrivals in Vietnam. Second, the Granger causality test indicated that causality goes positively from export and income to tourism, confirming the hypothesis that trade and income positively impact long-term tourism demand in Vietnam.
Theoretically, this study contributes to the existing literature on tourism demand by first accounting for both income and trade as tourism demand determinants. It has deployed a four-variable model to analyze the dynamic interaction of tourism arrivals, export, import, and GDP per capita in the short and long run. This is also the first study in Vietnam that incorporated trade as a determinant of tourism demand.
Practically, the causal relationship between export and tourist arrivals suggests that an increase in exports will attract international visitors to Vietnam and vice versa. This result might imply that most tourist arrivals are related to business tourism. Hence, Vietnam should focus more on tourism and trade-related policy, for example, MICE (Meetings, Incentives, Conferences, and Exhibitions) tourism. In addition, the income-tourism relationship finding suggests that tourism promotion policy should target countries with good economic performance. It should focus on major markets and divert attention to other potential countries to sustain the tourism industry's performance.
This study still has some limitations. Although we have analyzed trade and income, other factors can still influence the number of tourist arrivals. Further studies can examine economic-related factors as well as non-economic factors in the model. In addition, this study only uses Vietnam as a case. Future studies should include several countries. Moreover, the time series could be expanded longer for a more validated result.