Impacts of Terrorism in the United Kingdom and Europe on Tourism in the United Kingdom

Terrorist attacks occur more and more frequently. Therefore, people from all over the world pay more and more attention to them. This paper will focus on the relationship between terrorism and tourism in the United Kingdom, especially focusing on the relationship between terrorism in the United Kingdom or in Europe and tourism in the United Kingdom. This paper ap-plies the Auto Regressive Distributed Lag (ARDL) model to figure out their relationship. The data used in this article are from the website of the United Kingdom’s office for national statistics and the website of the global terrorism database respectively. Firstly, using the ADF test to figure out whether all these three variables, tourists arrivals, terrorist attacks happened in the U.K. and terrorist attacks happened in Europe continent, are stationary. Secondly, use the ARDL model to test whether terrorism in the United Kingdom and Europe respectively has an impact on British tourism. The results show that terrorism in the United Kingdom has a negative impact on British tourism while terrorism in Europe has a positive impact on it. The results are unex-pected. Then analyzing the reasons why terrorism in Europe has a positive impact on tourism in the United Kingdom. But whether these reasons are true needs to explore furtherly.

. However, there is still no research conducting into the relationship between terrorism and tourism in the United Kingdom. Studying the relationship between them is helpful to policy-making and minimizes the impact of terrorist attacks on tourism destinations. in which the research object and method are discussed. The fourth part is divided into two parts, the first shows the estimated results and the second discusses why are these results. The fifth part shows the paper's major findings, implications and some recommendations for future research.

Research on Tourism
The range of research on tourism is wide, including cultural tourism (Richards, 2018), sustainable development (Tölkes, 2018), volunteer tourism (Wearing & McGehee, 2013), urban tourism (Cohen & Hopkins, 2019) and so on. There are also many researchers devoting themselves in exploring factors having impacts on tourism demand (see Dritsakis, 2004;Chaisumpunsakul & Pholphirul, 2018; Stauvermann & Kumar, 2017 among many others). Dritsakis (2004) used the data of tourist arrivals, real income per capita, transportation cost and the real exchange rate of demand as key variables to examine tourism demand for Greece. Chaisumpunsakul & Pholphirul (2018) found international trade as an important factor affecting inbound tourism demand. Stauvermann & Kumar (2017) included economic growth and the increase in labor productivity as factors that promoting tourism demand. Saha & Yap (2014) recruited a set of panel data from 139 countries from 1999-2009 to analyze whether political instability and terrorism as factors affecting tourism. They characterized that the political instability factor is much more significant than one-off terrorist attacks. While the two factors working together can seriously damage the tourism demand.
Terrorist attacks actually increase the tourism demand in those much safer countries.
There are a lot of research papers based on British tourism. The themes of British tourism mainly focus on tourist behavior, tourism demand, tourism impact and so on in recent years (see Thurnell-Read, 2012;Cater, 2017;Curtin, 2013

Terrorism and Tourism
There are various definitions of tourism among different governments and institutions. The US Department of State defined terrorism as "premeditated, politically motivated violence perpetrated against non-combatant targets by subnational groups or clandestine agents" and international terrorism as "involving citizens or the territory of more than one country" (Annual Country Reports on There are also many pieces of research relating to the relationship between tourism and terrorism among different countries or regions (see Goodrich, 2002;Pizam & Fleischer, 2002;Pizam, 1999;Chu, 2008 among many others). Goodrich (2002) used the September 11 terrorist attack as a case study and analyzed its impact on the tourism industry in the USA. Pizam & Fleischer (2002), as well as Pizam (1999) resulted that terrorism with high frequency but whatever degree of severity had a much greater negative impact on tourism demand than a high severity but low frequency. If the terrorist attacks happen frequently and no matter how severe they are, the tourism demand will gradually decline and eventually

Data and Methodology
This section briefly introduces data sources and the methods adopted in this study. The results obtained by using the model and its analysis will be elaborated in the next section.

Data Resources
The data used in this paper are all secondary data. Among which, the data of overseas tourist arrivals in the United Kingdom comes from the website of the

Methodology
There are many factors affecting tourism, the impacts of terrorist attacks may not be as important as the major factors such as economic factors, distance fac- v is pure white noise error term.
Since the data used in this paper are time series, Augmented Dickey-Fuller (ADF) unit root test is needed to determine the stationarity of data before the ARDL model is applied. If the data is non-stationary, it needs to be processed first. The methods used in the paper are introduced briefly in the following.
After the ADF test, we can get the result if the variables are stationary. As long as variables are integrated of the same order or they are integrated of the same order after processing, they can be applied with ARDL model to figure out if there is persistent equilibrium relationship between variables (Pesaran & Shin, 1998) and what kind of relationship between them. Advantages of ARDL model include it is able to estimate small samples very well and it is also able to estimate endogenous variables. The ARDL model is as follows: is integrated of order 0, I(0), which means they are all stationary sequence, then ( ) 1 2 ARDL , , , , k p q q q ⋅⋅⋅ with k variables is as follows: 2) If ARDL , , , , k p q q q ⋅⋅⋅ with k variables is as follows: α represents the intercept term; t represents a trend over time if there is any; i Φ is explained as the coefficients of t i Y − , which is the lagged term of t Y ; p represents how many lagged terms do t Y have; j jl β represents the coefficients of , j j t l X − ∆ ; j q represent how many lag terms do

Estimated Results and Discussion
The previous chapter briefly introduces the econometric knowledge involved in this study. This chapter is divided into two parts, the first shows the estimated results obtained from the ARDL model while the second discusses the meanings Advances in Applied Sociology of these results and the reasons for such results.

Results of ADF Test
According to the theory introduced in chapter 3, the ADF test is the first step to do the research. The results are shown in Table 1 and Table 2. Table 1 shows that these three variables together are stationary, that is I(0), no matter by using ADF-Fisher Chi-square or ADF-Choi Z-stat method. Because we can see that the probability of either ADF-Fisher Chi-square or ADF-Choi Z-stat method is 0.00 which is smaller than 0.05, the null hypothesis is rejected. Table 2 shows that these three variables respectively are stationary. The probabilities of variable "TER_EUROPE", "TER_UK" and "TOR" are 0.0010, 0.0001 and 0.0038 respectively. They are all much smaller than 0.05, as a result, the null hypothesis is rejected. These results indicate that there may be a long-term equilibrium relationship between variables, but whether it exists or not needs to be verified by ARDL model.

Results of ARDL Model
From the results of the ADF test, we know that all these three variables are integrated of the same order. Therefore the variables satisfy the precondition of ARDL modeling. Whether there is a long-term relationship between terrorism and tourism in the United Kingdom can be tested by ARDL model, as well as the specific relationship between tourism and terrorist attacks in the UK, tourism and terrorist attacks in the European continent.
The results of the ARDL model are shown in Table 3. The results reject the null hypothesis which means there is a long-term relationship between variables since the probability of F-statistic is 0.00 which is smaller than 0.05. Terrorist attacks in the United Kingdom do have a long-term relationship with tourism in the United Kingdom, as well as terrorist attacks in the European continent.
All variables' optimal lag numbers in the ARDL model are determined by   Eviews automatedly. As shown in Table 3, the optimal lag numbers are 4, 1 and 0 respectively, that is ARDL (4, 1, 0), which means variable "TOR" has a forth lag term, variable "TER_UK" has no lag term, and variable "TER_EUROPE" has a first lag term. The coefficients of variable "TOR", "TER_UK" and "TER_EUROPE" estimated by ARDL model are shown in Table 3. The equation with these coefficients is shown as follows:

Discussion
Reviewing the research questions of this paper, the first question is whether ter-  The impact of terrorist attacks in the European continent on tourism in the United Kingdom is not only positive in the contemporaneous period, but also in the previous period. The variable " _ t TER EUROPE " is also a dummy variable, it only has two values: 0 and 1. According to the results in the Equation (7), When minds. This is mainly because of the geographically separate of the United Kingdom and the European continent. What's more, the United Kingdom is neither a Eurozone nor a Schengen country, though the United Kingdom is one of the European Union (EU) countries. As a result, once terrorist attacks occur in the European continent, overseas travelers will prefer traveling to the United Kingdom, especially those from EU countries. They will tend to travel outside the Schengen countries but close to them geographically and rich in tourism resources. The best choice for them is the United Kingdom.
Secondly, there are various types of tourists, such as conference tourists, leisure tourists, family visit tourists, and so on. Different tourists have different degrees of response to terrorist attacks. Some tourists (such as conference tourists) have a large demand elasticity in tourism destination while others (such as tourists for visiting relatives) have a small demand elasticity. Once terrorist attacks happen, those having a large demand elasticity in tourism destination will choose another destination for replacing. For example, those conference organizers will move the venues to the United Kingdom which is relatively safer but not far. As a result, the number of overseas tourists arrivals to the United Kingdom will increase.  Therefore, the ARDL model is more direct and convenient to determine whether there is a certain relationship in variables, compared with simple multiple regression. This method is equivalent to controlling all other variables, as long as the data of variables are stable, or stable after processing.

Limitations and Recommendations for Future Research
There still have many areas that can be further improved. Firstly, the definition of the terrorism attacks is not accurate enough, I am not sure if the terrorism attacks included in this paper is correct. Secondly, the result can only apply to the United Kingdom according to the specific data and the specific location of the United Kingdom.
Terrorist attacks are divided into two categories in this research, one is the terrorist attacks in the United Kingdom, the other is the terrorist attacks in the European continent. This research studies the impacts of these two kinds of terrorist attacks on tourism in the United Kingdom. My original assumption was that the terrorist attacks' influencing is less if the place is further. However, terrorist attacks' impacts on British tourism is not positively correlated with distance according to the estimated results. What's more, the decrease of overseas tourism arrivals in the United Kingdom in 2001 is due to the September 11 attack (see Figure 1). Therefore, if any further research is needed, the origin of terrorist attacks can be expanded.
Secondly, the impact of terrorist attacks on the European continent on British tourism is positive according to the estimated results. Such a result is likely when there's a significant increase in the number of several certain categories of tourist arrivals. As analyzed before, a certain kind of category may be conference tourism. However, this research does not verify if there is any different impact among different tourists. If further research is needed, new ARDL models can be built by using data of the different type of tourists to determine if there are different impacts and which categories are the most affected.
Thirdly, a terrorist attack is one kind of crisis. The crisis also includes natural disasters such as earthquakes, political events, financial crisis and so on Athanasopoulos & Hyndman (2008)  lia. According to their results, they conducted some suggestions on how to recover quickly after being affected. Therefore, terrorism can be expanded to the crisis to further study the impact of the crisis on British tourism.
Fourthly, as can be seen from Figure 1, the number of overseas tourist arrivals to the United Kingdom is continuous increasing apart from several specific years, 1991, 2001, 2008 and 2009. Terrorist attacks occur from time to time besides these specific years, but the number of overseas tourism arrivals to the United Kingdom has not been greatly affected. Is it possible that the reason for this phenomenon is that the UK government has taken timely measures to prevent the decrease, such as releasing the visa difficulty? If further research is to be carried out, I think it is necessary to disentangle the impact of policy.
In general, this research has concluded that terrorist attacks in the United Kingdom have a negative impact on British tourism, while terrorist attacks in the European continent have a positive impact on it by using ARDL model.

Conflicts of Interest
The author declares no conflicts of interest regarding the publication of this paper.