The impact of terrorism and corruption on tourism in Turkey: A regression analysis

1. Introduction

The importance of tourism to Turkey’s economy is unquestionable: the total contribution to Gross Domestic Product was 12.1% and supported 7.7% of total employment in 2018 [1]. But tourism needs a healthy economic environment to operate [2] and so this sector can be adversely affected by terrorism, regardless of whether or not it is targeted to tourists. In Turkey, during the last two decades the terrorist activities (domestic and international) have dramatically increased. Between 2011 and 2017 Turkey was targeted 87 times, 956 people were killed and 4,717 wounded. However, terrorism is not the only threat to tourism, affecting the choice of the tourism destination. In countries with high levels of corruption the degree of uncertainty increases and it acts as a tax, increasing the costs (economic and social) of traveling to these destinations and, consequently, shifting the demand. Corruption is widespread in Turkey’s public and private sectors. The 2018 Corruption Perception score for Turkey, published by Transparency International was 41 out of 100 and from 1995 till 2018 the average score was 39.4 denoting a high level of corruption.

This paper investigates the extent to which terrorism and corruption affect tourism demand in Turkey for the period 1995–2015. The choice of this country is due to the fact that historical statistics show that Turkey has been experiencing a high number of terrorist attacks within the OECD countries [3] and it is known as having a high level of corruption. Besides that, Turkey was ranked as the world’s eighth most popular tourism spot, just behind United Kingdom, in 2017, and although the improvement observed in the number of international tourist arrivals to Turkey in 2017, in the previous year there was a 23% decline. In the UNESCO “World Cultural Heritage List” there are 17 heritage sites from Turkey, out of 1073 properties around the world. Two of these sites are “mixed”, Goreme and Pamukkale, and 15 of them are “cultural” heritage sites.

This article is organized as follows: firstly, we review the literature on the impact of terrorism and corruption on tourism; next it is described the model to estimate, methodology and data. After presenting the results, we will discuss and reflect on points raised up by this study.

2. Theoretical framework and prior research

Turkey is a popular country among tourists in the Eurasia. Tourists prefer destinations that offer prestigious products as UNESCO heritage sites [4, 5] and safety, and tourism destination choices can be affected by instability factors as terrorism, corruption and political instability. As Baker [6, p. 58] referred “being safe on vacation is an expected requirement for any visitor in a tourist destination or city”. A high risk perception of a certain destination will discourage travellers and travel agencies will also reduce promoting tours in these countries, shifting offer.

Tourism is one of the sectors that suffers the most from terrorism and also impact negatively related services. Even if the attacks are not target to tourists, they have adverse effects, reducing the number of tourists, international receipts with the resulting repercussion on the balance of payments, employment and economic growth. In the last years several terrorist cells have committed bloody attacks against tourist destinations in different parts of the world, affecting seriously the image of these destinations [7]. Tourist destinations are attractive for terrorists for several reasons. Murdering innocent visitors on holidays is a shock all over the world. It will guarantee international media coverage, mainly in the home countries of the victims, making the conflict visible worldwide [8, 9, 10, 11, 12]. Affecting the image of a destination, intimidating potential tourists, undermines the positive dynamism of tourism in the economy and so the economic power of the nation [8]. Places attended by tourists provide anonymity to terrorists, allowing them to easily infiltrate [13]. It is difficult for the police and other security forces to identify potential attacks in public and private spaces with a high-crowd density. All this discourages tourism in destinations affected by terrorism.

More recently corruption has been considered a determinant of tourism. Corruption can have a detrimental effect in the development of a nation and so “sand the wheels of growth”, since it generates distortions and inefficiency. Corruption can reduce tourism competitiveness [14] and can also affect negatively the tourism industry, once the social and cultural image can be deteriorated by the practices of bribery. In countries with a high level of corruption the cost of travelling is higher as travellers may have to pay bribes or permits to some tourist destinations. Another current of though advocates that corruption can “grease the wheels” as it can facilitate business development, increasing efficiency and growth [4].

Several publications have emerged in recent years demonstrating the relation between terrorism, corruption and tourism but according to the authors’ knowledge it can’t be found much literature discussing the impact of terrorism and corruption on tourism in the case of Turkey.

Current research on the impact of terrorism on tourism suggests a negative relationship [15, 16, 17]. Different techniques have been used in the research of this relationship for time series models, as cross-sectional gravity equations, general equilibrium models, structural equations, and demand-supply models. Feridun [15] has investigated the causal impact of terrorist attacks on the tourism industry in Turkey based on the Autoregressive Distributed Lag (ARDL) bounds testing procedure between 1986 and 2006. The study has shown the existence of a negative causal effect of terrorism on tourism. Yaya [18] has found that, between 1985–2006, the impact of terrorism on tourism in Turkey was negative, but the magnitude of the reduction of foreign tourist inflow is small. Moreover, the duration of the impact is observed approximately within one year. Other authors found the adverse effect of terrorism on tourism in Turkey [19, 20]. Afonso-Rodríguez [20] suggest a negative impact of terrorism on real GDP through a reduction of the contribution of tourism demand on economic growth. Bassil et al. [21] analysed the impact of domestic and transnational terrorism on tourism demand to Lebanon, Turkey and Israel and found that terrorism in a country affects tourism demand not only in the country but has spill-over effects on the others. Isaac and Velden [22] suggest that German travellers have a low safety perception of Turkey as tourist destinations and they perceived Turkey and Egypt as unsafe countries. In a study about the causal relationship between terrorism and tourism, Nikšić Radić et al. [12] investigate the hypothesis that tourism attract terrorism and not only that terrorism affects tourism, and for Turkey they confirm the tourism-led-terrorism hypothesis. Through the estimation of a gravity model for the bilateral flows between OECD countries Harb and Bassil [23] also suggest the existence of a curvilinear relationship between terrorism and tourism, whereby tourist arrivals start to decline after a threshold.

The existing literature of corruption and tourism demand reveals some mixed conclusions. While Lau and Hazari [24], using panel ordinary least squares, conclude that an increase in corruption will not have adverse effects on tourism development, several published research studies suggest an inverse relationship [14, 25, 26].

Yap and Saha [4] researched the impact of political instability, terrorism and corruption on tourism demand, controlled by the presence of cultural and natural heritage. They found a negative effect of corruption on tourist arrivals, but a positive one when heritage variables are included in the estimation, which suggests that for countries with an important cultural and natural heritage, the corruption is not significant for the traveller decision. Demir and Gozgor [27] analysed the impact of absolute and relative corruption on the number of tourist arrivals and found a negative impact of relative corruption in Turkey between 1996–2014. Likewise, Das and Dirienzo [14] showed that a decrease in corruption affects positively tourism competitiveness in 119 countries. However, the marginal gain in tourism competitiveness due to a reduction in corruption is greater in developing countries compared to developed countries. Poprawe [25] finds that the effect of corruption on total tourist inflows is negative in 100 countries over the period 1995–2010. The paper also finds that one-point decrease in corruption leads to 2% higher tourist inflows.

While most of the research focus on linear relationships, Saha and Yap [28] and Lv and Xu [29] analyse the impact of corruption on tourism in a non-linear framework, considering the level of impact, and found that corruption increases inbound tourism at a lower level, but after a threshold it has a detrimental effect on tourism demand.

3. Model, methodology and data

According to Song et al. [30], tourist arrivals is the most commonly used measure of international tourism demand, followed by tourist expenditure and tourists’ overnight stays. In this study tourism demand is measured by international tourist arrivals to Turkey, using data covering the period from 1995 to 2015.

The data has been collected from different sources for parameters such as the total number of attacks, mortalities, including number of injured in all terror attacks, control of corruption, political stability and absence of violence/terrorism, heritage, inflation and foreign tourist arrivals. The hypotheses considered by our study are the following:

Hypothesis 1: Terrorism adversely affects the foreign tourist arrivals in Turkey.

Hypothesis 2: Corruption adversely affects the foreign tourist arrivals in Turkey.

Single equation demand models appear in the functional form shown in Eq. (1),

ITA t = f ⁢ ( ITA t - 1 , TERR t , COR t , HERIT t , INF t , PS t )(1)

The basic variables for empiric analysis are: ITA = International tourist arrivals; TERR = Terrorism; COR = Control of corruption; HERIT = cultural and natural heritage; INF = Inflation; PS = Political stability.

The model is chronological and the structure estimated by MRLM using the Enter method. Some of the independent variables in Eq. (1) were excluded: political stability and inflation.

The empirical research on the impact of terrorism on tourism is based on the following regression Eq. (2):

ln ⁡ ( ITA t ) = β 0 + β 1 ⁢ ln ⁡ ( ITA t - 1 ) + β 2 ⁢ ln ⁡ ( TERR t ) + β 3 ⁢ ( COR t ) + β 4 ⁢ ( HERIT t ) + ε t(2)

The number of International Tourist Arrivals to Turkey was obtained from the World Development Indicators of the World Bank. For the independent variables it was used the International Tourist Arrivals lagged during a period [31], once tourists share their experiences, reducing the uncertainty for other visitors. Terrorism (TERR) was measured by two variables: the number of attacks (ATK) and the number of kills and wounds (NKW) from the Global Terrorism Database produced by the National Consortium for the Study of Terrorism and Responses to Terrorism [32]. The corruption measure used in this study was the estimates of Control of Corruption, one of the six components of the World’s Bank Worldwide Governance Indicators. Control of Corruption captures the “perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as “capture” of the state by elites and private interests [33, p. 4]. The estimates range from - 2.5 to 2.5, with higher scores corresponding to better outcomes (less corruption). Heritage refers to the number of heritage sites, either cultural or natural, from the World Heritage List created by UNESCO. This study uses annual data from the period from 1995 to 2015.

Analysis involved two related phases, using SPSS 23. The first examined variable means, standard deviations and bivariate correlations among the variables related to international tourist arrivals, terrorism, corruption and heritage were calculated. The second phase employed OLS regression on the data collected.

The following Table 1 presents a brief description of the variables used in the regression equation.

Correlations among the variables related to international tourist arrivals, terrorism, corruption and heritage, appear in Table 2.

The variable Log of number of attacks is positively correlated with Log of number of kills and the effect is statistically significant at the 1% level with the coefficient of 0.774. That variable is also negatively correlated with cultural and natural heritage, and the effect is statistically significant at the 5% level with the coefficient of - 0.568.

The variable Log of number of kills and wounds is negatively correlated with Control of corruption, and the effect is statistically significant at the 1% level with the coefficient of - 0.449.

As a robustness test, we run OLS regression on the data collected with the variables related to terrorism, corruption and heritage.

The diagnostic statistics show that all the coefficient of all independent variables are significant at 1% level and have the expected sign. In the attempt to investigate the performance of the specified model some basic diagnostic statistics were performed. The Variance Inflation Factors are used to test the estimated model for the presence of collinearity. The variance inflation factor estimates how much the variance of a coefficient is “inflated” because of linear dependence with other predictors. A general rule states that, if the VIF of a variable exceeds 10, that variable is considered to be highly collinear. Table 2 shows that the VIF values of the independent variables are less than 10, which suggests that multicollinearity is not a serious problem in the present study. The Durbin-Watson statistic test examines the null hypothesis that the residuals from an ordinary least-squares regression are not auto correlated against the alternative that the residuals follow an AR1 process. The values of Durbin-Watson (2.76 and 2.053 in regression model for Terror (ATK) and Terror (NKW), respectively) indicate that the residuals from OLS regression are not auto correlated.

Based on the regression results, and after transforming the estimated coefficients of the independent variables, in order to interpret them, the estimated equation for the model with Log of number of attacks (3) and Log of number of kills and wounds (4), respectively, can be presented as follows:

ln ⁡ ( ITA t ) = 940.05 + 0.553 ⁢ ln ⁡ ( ITA t - 1 ) - 0.168 ⁢ ln ⁡ ( ATK t ) + 1.38 ⁢ ( COR t ) + 0.13 ⁢ ( HERIT t )(3) ln ⁡ ( ITA t ) = 437.34 + 0.623 ⁢ ln ⁡ ( ITA t - 1 ) - 0.081 ⁢ ln ⁡ ( NKW t ) + 1.29 ⁢ ( COR t ) + 0.07 ⁢ ( HERIT t )(4)

The estimated coefficients and all performed diagnostic statistics showed specified models that fit the classical assumptions and models that pass all the usually used tests. The model, as measured by the adjusted R-squared, shows that approximately 97.2% of the variations in international tourist arrivals are explained by variations in the independent variables included in the analysis. All of the coefficients on the independent variable have the expected signs.

According to the elasticity as interpreted from the log-log model, a one percent increase in the number of attacks will decrease foreign tourist arrivals by 0.168% and an increase of one percent in the number of kills and wounds decreases inbound tourism by approximately 0.08%. As expected, there is a negative sign on the Terrorism variable showing a detrimental effect in inbound tourism.

The coefficients of corruption are statistically significant in all regressions and with a positive sign. The results suggest a negative impact of corruption on tourism: an increase by 0.1 in the control of corruption causes an increase on international arrivals between 12.9% and 13.8% ( ( e β - 1 ) × 100 ) , supporting the “sanding the wheels” hypothesis.

The results also suggest that the cultural and natural heritage has a positive sign. This is expected as it is presumed tourists would choose countries with points of interest and that have universal value. A coefficient of 0.7 and 0.13 implies that tourist arrivals increase by 7% to 13% if the Turkish heritage recognized by UNESCO increases one.

Terrorism and corruption are two of the major challenges of the 21st century and in countries with high levels of corruption usually there are problems of security and terrorism, impacting negatively the tourism industry. Corruption is widespread in Turkey’s economy and there has been a significant number of attacks, including to tourists. Turkey government should become more pro-active in controlling corruption and strengthen tourists’ security.

The policy implications of this paper are also valuable. Our research finds evidence that Turkey is a least favourable destination for international tourists if terrorism and corruption increase.

A favourable macroeconomic policy, governance mechanisms and low level of violence are desirable for the growth of a tourist based economy.

The results of this research can contribute to alert the tourist authorities, encourage the performance of campaigns and define strategies in order to retake the past positive trajectory of tourism in Turkey.