Armed conflict,military expenditure and international tourism

Abstract

This article uses a gravity model to explore whether military spending has any moderating effect on the link between armed conflict and international tourist flows. The data set consists of a panel of 188 countries over the period 1995 to 2015. We show that the moderating effect of military spending depends on the levels of relative military spending as well as geographical location. Specifically, in the presence of armed conflict, ‘moderate’ level of relative military spending helps to promote the international tourism attractiveness of destination countries, whereas ‘high’ level of relative military spending cannot reverse the negative impact of armed conflict, it rather fuels the problem. In general, countries in regions such as Southeast Asia that allocate ‘moderate’ amount of resources for security attract more international tourists relative to countries in regions, such as the Middle East and North Africa, that spend a larger share of GDP on security.

Keywords

armed conflict geographical location gravity model international tourism military expenditure

Introduction

Armed conflicts tend to dampen international attractiveness of a country, while tourists are more likely to visit most peaceful countries. A country or region that is highly attractive to international tourists risks losing its attractiveness once engulfed in armed conflict. For example, the Middle East has set a burning evidence of the potential negative impact of violent conflict on international tourism. While there was as high as 8.4% growth in international tourist arrivals in Asia and the Pacific, there was more than 4% decrease in international tourist arrivals in the Middle East in 2016. Similarly, in the aftermath of security incidents coupled with negative travel advisories from some countries, the number of tourist arrivals in Egypt fell by 42% in 2016 (World Tourism Organization, 2017a).

In the context of the demand side of tourism, conflict or war tends to damage the international reputation of a country. Empirical evidence suggests that conflict and threats have adverse impacts on tourism demand. For instance, conflict and war tend to deter potential tourists from visiting a country (Pizam and Mansfeld, 2006). Similarly, tourists tend to avoid conflict-prone areas and are more likely to substitute risky destinations for safe ones (Araña and León, 2008). Eilat and Einav (2004) show that the political risk of a destination country is a vital determinant of tourism. Therefore, a higher level of political risk tends to depress tourism demand.

In the context of the supply side of tourism, conflicts and threats tend to affect the supply of tourism services as well. In the presence of conflicts, a significant number of service providers and operators in the tourism sector may suspend business activities. For example, a number of studies find that military conflict obstructs the growth of tourism industry due to the absence of peace and security (Hyndman, 2015; Saha and Yap, 2014). This suggests that armed conflict potentially hinders the growth of the tourism industry.

Notwithstanding the potential negative impact of armed conflict on tourism demand and/or tourism supply, a country could regain international attractiveness after a period of conflict. For instance, there is evidence of post-conflict significant growth in the tourism industry (Lee, 2006; Pratt and Liu, 2016). In the presence of armed conflict, a country could allocate adequate resources for security in order to restore peace and promote international tourism. Hajibaba et al. (2016) show that the provision of security staff, and change of accommodation accompanied with an upgrade, among others, can prevent tourists from cancelling bookings in destinations experiencing crises. In general, military spending accounts for a significant share of security spending. Military spending is, therefore, a robust proxy for security spending. The question now arises whether countries that have been facing negative growth in international tourism due to the violent conflict can further revive it by spending more on military in order to re-establish peace and improve their images for international tourism.

This article, therefore, investigates whether relative military spending has a moderating effect on the link between armed conflict and international tourism using gravity panel data. The data set consists of a panel of 188 countries for the period 1995 to 2015. In the context of this study, we define conflict as having 25 or more battle-related deaths resulting from the use of armed force between two parties, of which at least one of the parties is a government (Pettersson and Wallensteen, 2015). Our results show that, in conflict-prone countries, military spending potentially has a positive impact on international tourism, as military spending is considered as a form of spending for the maintenance of peace and security. Thus, in the presence of armed conflict, an increase in military spending could signal to the outside world that every effort is being made to re-establish peace and security. However, the moderating effect of military spending is likely to be non-linear and vary with the levels of military spending. Striking the right balance between military expenditure and other productive expenditures could promote international tourism on the one hand. On the other hand, excessive military spending could ‘crowd out’ private investment, such as tourism investment thereby potentially lowering the international attractiveness of a destination country.

The rest of the paper is structured as follows. The next section presents a review of literature related to the relationship between armed conflict, military expenditure and international tourism. The third section discusses the methodology and describes the empirical specification. The fourth section presents the overview of data, description of the variables and summary statistics. The fifth section discusses the results, while the final section concludes.

The relationship between armed conflict, military expenditure and international tourism

A growing literature has explored the impact of political instability, terrorism and conflict on international tourism. Neumayer (2004) finds that human rights violations, conflict and other politically motivated violent events negatively affect tourist arrivals. Similar findings were reported by Eilat and Einav (2004) showing that political risk is a vital determinant of tourism. Other studies empirically assessed the impacts of terrorism on tourism, both domestic and regional, and have found a direct negative effect (Araña and León, 2008; Drakos and Kutan, 2001; Enders and Sandler, 1991; Enders et al., 1992; Llorca-Vivero, 2008; Raza and Jawaid, 2013).

Drakos and Kutan (2001) in a study covering Greece, Israel and Turkey, and using Italy as a ‘control variable’, find that the intensity of terrorist activities has significant own and spillover effects on the market shares in the tourism sectors of the countries covered in the study. Raza and Jawaid (2013) investigate the impact of terrorism on the tourism sector in Pakistan. The results indicate that terrorist attacks have a significant negative impact on inbound tourism in both the long and short runs. Similarly, Buigut and Amendah (2016) show that increase in number of fatalities from terrorism incidents in Kenya significantly dampens international tourist arrivals.

On the contrary, Liu and Pratt (2017) show that the impact of terrorism on tourism is limited in the long run and partial at best in the short run concluding that tourism is resilient to terrorism. As highlighted by Liu and Pratt (2017), one possibility for this weak relationship could be that the frequency of terrorist attacks matters more than the severity of attacks for tourism. Saha and Yap (2014) analyse the effects of political instability and terrorism on tourism development and find that political instability has a more severe impact on tourism industry than a one-off terrorist attack. Issa and Altinay (2006) further highlight the need for using crisis management procedures as a possible response to boosting tourism in the face of political instability. Above-mentioned studies, therefore, indicate that safety and security measures are required to attract more tourists in the regions where terrorist attacks are more frequent.

A related strand of the literature also indicates that military expenditure can dampen the adverse effect of armed conflict and contributes to domestic and foreign investment as well as economic growth. While, armed conflict significantly reduces foreign investments (Kim, 2016; Li and Vashchilko, 2010; Nigh, 1985), however, if the host country can offer high security as could be achieved through the allocation of an adequate level of military spending, then foreign investment increases (Maizels and Nissanke, 1987). Several studies found that military expenditure stimulates business confidence in the presence of regional conflict, which, in turn, create an enabling environment for investment and economic growth to flourish (Barro and Sala-i-Martin, 2004; Benoit, 1978; Dunne et al., 2005; Whynes, 1979). Other studies suggest that in the presence of conflict, the effect of military expenditure on investment and economic growth is non-linear (Aizenman and Glick, 2006; Aziz and Asadullah, 2017; Aziz and Khalid, 2019; Cothren, 2002; Frederiksen and Looney, 1982; Landau, 1996). As discussed above, several studies have investigated the link between conflict and tourism. Although the interplay between military spending and tourism has been explored by Nassani et al. (2017), however, the conditional effect of military expenditure or the moderating role of military spending on the underlying link between armed conflict and international tourism has not been investigated yet.

In the presence of conflict, the moderating effect of military expenditure on international tourism can be better understood by considering the relationship between military spending and macroeconomic performance. Military expenditure without threat may have a ‘substitution effect’ (guns for butter) or ‘complimentary effect’ (patriotic effect). Landau (1996) finds a conditional effect of military expenditure such as the ‘Pentagon effect’ (resource use and expansion of government due to the expansion of defence) and the ‘patriotic effect’ (expenditure due to threats). The earlier lowers economic growth, while the latter fosters it. Shieh et al. (2002) show that maximum economic growth can be achieved in the presence of an optimal threshold of defence expenditure. There is also some evidence of ‘peace dividends’ resulting from freeing resources from military expenditure (see, Deger and Smith, 1983; Dunne et al., 2002; Smith, 1977, 1978, 1980).

Similarly, the military expenditure may have substitution (a negative) or complementary (a positive) effect on international tourism inflows. If an increase in military expenditure reduces investment in the tourism industry, military expenditure will depress international tourist inflows. However, if higher military expenditure ensures a safer environment for the tourists and if an increase in military expenditure does not reduce investment in the tourism sector, military expenditure will have a positive impact on international tourist arrivals. For the developing countries, in general, the latter is the more likely case than the earlier. In general, developing countries do not cut investment in tourism in order to invest more on the military. For instance, Nassani et al. (2017) find a positive role of military expenditure on international tourism growth, which indicates that there should be protective measures for a safe and secure place for the tourists that ultimately increases the international tourism income.

Information about military spending across countries is increasingly available to the public. For instance, the Stockholm International Peace and Research Institute (SIPRI) military expenditure project gathers, analyses and publicizes information concerning military expenditures across countries and over time (SIPRI, 2019) which is then further disseminated via press releases (see, SIPRI, 2018) and picked up by other media outlets. Based on a recent review by SIPRI, it was reported that Venezuela’s annual military expenditure increased by 26% over the period 2005–2015 which contributed in strengthening the power of the military in the ongoing political crisis in the country (Silva and Tian, 2019). Similarly, based on a review by SIPRI, it was reported in the Independent (a British online newspaper), that Uganda is pushing to double its military spending for the period 2019/2020 (Matsiko, 2019).

Further, information concerning military spending, especially in the wake of armed conflict is often publicized in the media. For instance, several developing countries that rely heavily on tourism revenues tend to publicize their military spending, especially following an armed conflict or terrorist attacks. Kenya is one such example. Following the 2013 Westgate Mall attack and 2015 Garissa University attack, the government of Kenya allocated KSh112.5 billion ($1.2 billion) to the National Intelligence, which was widely publicized in the media (Oxford Business Group, 2015). Similarly, in the face of increasing threat of terrorism, the Philippine government has widely publicized its budgetary allocation for the military and police in 2019, citing the importance of ensuring security for both foreign and domestic tourists in various destinations in the country (Jiao, 2018).

Government authorities are keen to disseminate this information as it signals to the international community that efforts are being made to resolve the conflict. Besides, military expenditure might be seen as acting as a deterrent on warring parties, thus creating the incentives for warring parties to publicize their military expenditures. For instance, Collier (2006) highlights that governments spend more on their military in order to deter any internal rebellions or external threat and in the wake of a recent history of internal or external conflict, military spending is seen as signal of the government’s commitment to peace (Albalate et al., 2012). Further, in most circumstances, increased military spending in a conflicting country leads to counter-insurgency operations that are often reported widely in the media. For instance, the effectiveness of counter-insurgency operations in Pakistan in rooting out terrorism received a lot of attention from the international media outlets (see, ‘Pakistan’s army is getting serious about defeating domestic terrorism’, 2018).

Increased military spending also translates into efforts to militarize hotel security and key tourist locations. Industry experts and security analysts whose recommendations and commentaries are published in magazines (Weaver, 2011) as well as on travel websites that are the main source of information for the tourists view these efforts favourably. Similarly, hotels and travel agencies providing tours and holiday packages often publicize the security and safety measures taken by the government to ensure tourists’ safety. The tourism landscape is increasingly changing as

many tourist bubbles are well-protected zones of comfort and consumption from which tourists can gaze at ‘other’ places. The prevalence of heightened security practices may feed into a process whereby visibly militarized approaches to hazard management – and consumable images of such approaches – become well established, customary, and conventional. (Weaver, 2011: 683)

The prevalence of social media further promotes the effectiveness of the media in terms of disseminating information about military spending. In the modern era, travel blogs and vlogs, as well as social media among others, play a significant role in disseminating information about the safety and security of tourist attractions in different locations. For instance, in the case of Pakistan, several blogs highlighted how secure it was to visit the country and the measures that the government have taken to ensure the safety of tourists such as assigning bodyguards to the foreigners and police motorcade in some circumstances that would only be possible if the country allocated significant funding for security.

It is notable that previous studies estimate a direct relationship between military expenditure and tourism flows. The findings from these studies did not explain if military spending could help to mitigate the adverse effect of armed conflict on international tourism nor how the impact of armed conflict augmented by military spending varies by the level of military expenditure across countries. The story, therefore, remains incomplete – if the moderating role of military spending on the nexus between armed conflict and international tourism is unexplored. We attempt to fill this gap in previous literature.

Methodology

Theoretical rationale

The gravity models are recognized as a workhorse of international trade analysis. Unsurprisingly, gravity models have been used widely for analysing the determinants of bilateral trade flows across countries or to explain international trade patterns (Khadaroo and Seetanah, 2008; Khalid, 2017; Kinuthia, 2017; Narayan and Nguyen, 2016). By extension, gravity models can be used to explore the determinants as well as the patterns of international tourism. This stems from the fact that international tourism is one of the main categories of international trade in services. As a result, an augmented gravity model is an appropriate method for investigating the patterns of international tourism (Khadaroo and Seetanah, 2008; Khalid et al., 2019; Santana-Gallego et al., 2016).

We specify an augmented gravity model in which the determinants of international tourism flows include measures of military spending, armed conflict and classical gravity determinants as follows

\begin{array}{l} T F_{o d t} = f (P O P U_{o t}, P O P U_{d t}, G D P P C_{o t}, G D P P C_{d t}, D I S T_{o d}, C O N T_{o d}, C O M O F F L_{o d}, \\ C O M N O F F L_{o d} F C O L_{o}, F C O L_{d}, L A N D_{o d}, I S L A_{o d}, C O N F_{d}, M I L S P_{d}, \\ C O N F_{d} \times M I L S P_{d}, C T Y F E_{0}, C T Y F E_{d}) \end{array}

where $o d t$ denotes origin, destination and time period, $P O P U$ is population, $G D P P C$ is GDP per capital measured at purchasing power parity (PPP), $D I S T$ is distance, $C O N T$ is contiguity, $C O M O F F L$ is common official language dummy, $C O M N O F F L$ is common unofficial language dummy, $F C O L$ is former colony dummy, and $L A N D$ is landlocked dummy and ISLA is island dummy. CONF is a measure of both internal and external conflict, $M I L S P$ is the ratio of military expenditure to GDP, and $C T Y F E$ is country specific fixed effects.

A particular focus of the present study relates to the interplay between armed conflict and international tourism attractiveness. This includes the influence of military spending in the underlying relationship. We, therefore, capture the moderating effect of military expenditure on armed conflict, if any, on the international tourism attractiveness with the use of an interaction term, $C O N F_{d} \times M I L S P_{d}$ , as specified in equation (1).

We use population to capture the effect of economic size as opposed to GDP. In the trade literature, GDP is a commonly used measure to capture economic size. However, population is a more appropriate measure as tourism is a form of international trade in services as opposed to products (Gil-Pareja et al., 2007). We also control for real GDP per capita measured at PPP as tourists from advanced economies are more likely to travel across international borders for touristic purposes or face a lower level of an obstacle in terms of migrating across international borders (Gil-Pareja et al., 2007). Further, controlling for real GDP per capita at PPP helps to dampen the influence of price effects in the underlying link.

Empirical specification

In line with the above theoretical rationale, the reduced-form baseline augmented gravity model is specified as follows

\begin{matrix} Ln T F_{o d t} = β_{0} + β_{1} Ln P O P U_{o t} + β_{2} Ln P O P U_{d t} + β_{3} Ln G D P P C_{o t} + β_{4} Ln G D P P C_{d t} \\ + β_{5} Ln D I S T_{o d t} + β_{6} C O N T_{o d} + β_{7} C O M O F F L_{o d} + β_{8} C O M N O F F L_{o d} \\ + β_{9} F C O L_{0} + β_{10} F C O L_{d} + β_{11} L A N D_{o d} + β_{12} I S L A_{0 d} + β_{13} C O N F_{d} \\ + β_{14} M I L S P_{d} + λ_{t} + Γ_{0} + Γ_{d} + ε_{o d t} \end{matrix}

where Ln denotes natural logarithm, $β_{0}$ is a constant that captures the unobserved quality of destination in a given year to inhabitants from the originating country, $λ_{t}$ is year effects, $Γ_{0}$ is country of origin fixed effects, $Γ_{d}$ is country of destination fixed effects and $ε$ is an error term that is distributed i.i.d across country pairs over time. $β^{'} s$ are parameters to be estimated.

An alternative baseline specification estimated using Fixed Effects Estimator is specified as follows

Ln T F_{o d t} = α_{0} + α_{1} Ln P O P U_{o t} + α_{2} Ln P O P U_{d t} + α_{3} Ln G D P P C_{o t} + α_{4} Ln G D P P C_{d t} + α_{5} C O N F_{d} + α_{6} M I L S P_{d} + λ_{t} + Ω_{o d} + ε_{o d t}

where $Ω_{o d}$ is country pair fixed effects and $α^{'} s$ are parameters to be estimated.

To capture the moderating effect of the levels of military spending on the relationship between armed conflict and international tourism, we interact armed conflict with different measures of the levels of military expenditures of the destination countries. To ascertain the levels of relative military spending, the distribution of the average of the ratio of military spending to GDP of destination countries is divided into four quartiles. The first quartile consists of destination countries with low level of relative military spending. The second quartile includes destination countries with moderate level of relative military spending. The third quartile comprises of destination countries with the moderately high level of relative military spending, and the fourth quartile consists of destination countries with high level of relative military spending.

To capture the moderating effect of the different levels of military spending on the underlying relationship, equation (2) is extended as follows

\begin{matrix} Ln T F_{o d t} = a_{0} + a_{1} Ln P O P U_{o t} + a_{2} Ln P O P U_{d t} + a_{3} Ln G D P P C_{o t} + a_{4} Ln G D P P C_{d t} \\ + a_{5} C O N T_{o d} + a_{6} Ln D I S T_{o d t} + a_{7} C O M O F F L_{o d} + a_{8} C O M N O F F L_{o d} + a_{9} F C O L_{0} \\ + a_{10} F C O L_{d} + a_{11} L A N D_{o d} + a_{12} I S L A_{0 d} + a_{13} C O N F_{d} + a_{14} L M I L_{d} + a_{15} C O N F_{d} \\ \times L M I L_{d} + λ_{t} + Γ_{0} + Γ_{d} + ε_{o d t} \end{matrix}

where $L M I L$ denotes the level of military spending relative to GDP and $a^{'} s$ are parameters to be estimated.

Equation (4) will be estimated separately by the levels of relative military spending of the destination countries. This approach will allow us to gauge the moderating effect of relative military expenditures on the link between armed conflict and international tourism according to the levels of relative military spending.

In addition, we use the distribution of the average relative military spending by regions of destination countries to classify the sample into different levels of relative military spending. Similar to the initial classification, the regions were classified into four quartiles. Regions classified as low relative military spending include Central Asia (CA), Oceania (OC) and Sub-Saharan Africa (SSA). Regions classified as moderate relative military spending include Latin American and the Caribbean (LATCA) and Southeast Asia (SEA). Regions classified as moderately high relative military spending are East Asia, Europe and South Asia. Regions classified as high relative military spending include the Middle East and North Africa.

We expect the estimates obtained when the levels of relative military spending are defined by the countries of destination to be congruent with the ones obtained when the levels of relative military spending are defined by regions of destination. Further, we perform robustness checks by using the internal conflict measure in lieu of overall armed conflict. Armed conflict is a broader measure of conflict that includes both internal and external conflicts. We expect the estimates obtained using the two measures of conflict to be similar in terms of signs and magnitudes.

In this type of analysis, estimates may be susceptible to potential endogeneity issue. Endogeneity might arise if a larger inflow of tourists reduces the likelihood of conflict between countries. However, Pratt and Liu (2016) show that tourism does not foster peace in the destination countries ruling out the likelihood of any feedback from increased tourist flows to conflict. Notwithstanding, to ensure that our results are not particularly vulnerable to endogeneity concerns, we expect the estimates obtained using the fixed estimator for the base model to be consistent with the ones obtained using ordinary least square (OLS) estimator. Fixed effects estimator addresses potential endogeneity arising from permanent country-specific effects. Further, we use lagged conflict, lagged levels of relative military spending and lagged interactions of conflict and relative military spending in lieu of current conflict, current levels of relative military spending and current period interaction term of conflict and levels of relative military spending. In general, we expect the estimates obtained from the static specification to be consistent with the ones obtained from the dynamic specification. Assuming that is the case, that would help to show that the results are not particularly vulnerable to endogeneity concern.

Overview of data and description of variables

The empirical model is estimated using data collected from different sources. The data for relative military spending was made available by the SIPRI Military Database, Sweden. The data on armed conflicts come from the UCDP/PRIO Armed Conflict Dataset, published by the Centre for the Study of Civil War at the International Peace Research Institute, Oslo (PRIO), and the Uppsala Conflict Data Program (UCDP) at the Department of Peace and Conflict Research, Uppsala University. In general, media reports on military expenditure refers simply to the defence budget of the country in question. However, many countries have significant military expenditure in other budget lines. The superiority of SIPRI’s military expense data set is that it captures all possible military-related expenses. The main categories of spending as captured in SIPRI’s military expense data set include personnel expenses, consisting of salaries and benefits of military and civilian defence personnel; operations and maintenance expenditure, consisting of running costs such as costs relating to fuel, repair and maintenance, travel, general supplies, rent, utilities and payments for services. Others include procurement costs consisting of expenditure for military equipment; expenses for research and development (R&D) including expenditure for developing new weapon systems and technologies; costs for constructions including expenditure for military bases and other military infrastructure; and also the military aid to other countries (usually a small category) (SIPRI, 2019).

The information on variables such as Gross Domestic Product, GDP per capita at PPP, and land area for 217 countries was collected from the World Bank Development Indicators (World Bank, 2017). Information pertaining to the flow of visitors distinguished by the country of origin and destination for 222 countries over the period 1995 to 2015 was collected from the UNWTO database (World Tourism Organization, 2017b). The data on contiguity, colonial link and language for 224 countries were obtained from the Gravity database from Head et al. (2010).

We merged the different data sets obtaining a panel data consisting of 188 countries. The final data set after dropping observations with missing data is an unbalanced panel of 61,949 observations. The missing data resulted from some countries not having data at the beginning of the period for some relevant variables or the paucity of data from developing countries.

Dependent variable

In line with most previous studies, we use tourist arrivals to capture tourism demand (Dogru et al., 2017; Gil-Pareja et al., 2007; Khadaroo and Seetanah, 2008). We capture the flow of tourists from the originating countries to the destination countries worldwide using the data compiled by UNWTO. UNWTO is suitably equipped to compile tourism data, as part of its responsibility is to ensure that different countries follow a uniform definition regime.

To ensure that an appropriate measure of tourist arrivals is available and in order to avoid conflicting definitions of a tourist, UNWTO refers to tourism as the act of travelling to and staying in destinations other than the usual abode for a period not exceeding 1 year. The purpose of the trip could be for leisure, business or other activities that are not compensated for in the place visited (United Nations, 2010).

Tourist arrivals are a robust measure of tourist flows as it is relatively easier to ascertain the number of individuals entering a country. Expenditures of tourists are sometimes used as a measure of tourist flows. Tourist expenditures are not easily accessible, as they have to be estimated. In addition, tourism receipts data as published in the balance of payments tend to limited due to the problem of inaccuracy (Sinclair, 1998). Neumayer (2004) shows that tourist arrivals and tourism receipts are highly correlated. Thus, tourist arrivals are an appropriate measure of international tourism. We, therefore, use tourist arrivals as the dependent variable in the empirical analysis.

Explanatory variables

The main explanatory variables used in the empirical analysis are a measure of relative military spending, a dummy for armed conflict and an interaction term that captures the joint effect of relative military spending and armed conflict. Relative military spending, $M I L S P$ , is captured using the ratio of military spending to GDP. The relative military spending variable is derived from military spending to GDP ratio. Given that the effect of military spending potentially varies by the level of military spending, we generate four dummy variables that capture relative military spending using the distribution of the average of the ratio of military spending to GDP of destination countries based on quartiles. Low level of relative military spending is captured by the variable ‘LOWMIL’ that takes the value 1 if a destination country’s average military spending to GDP ratio falls in the first quartile and 0 otherwise. Moderate level of relative military spending is denoted by ‘MIDMIL’ taking the value 1 if a destination country’s average military spending to GDP ratio falls in the second quartile and 0 otherwise. ‘MODHIMIL’ takes the value 1 if a destination country has an average military spending to GDP ratio in the third quartile and 0 otherwise, and ‘HIGHMIL’ equals to 1 if a destination country has an average military spending to GDP ratio that lies in the fourth quartile and 0 otherwise.

According to UCDP, an armed conflict is defined as

a contested incompatibility that concerns government or territory or both, where the use of armed force between two parties results in at least 25 battle-related deaths in a calendar year. Of these two parties, at least one has to be the government of a state. (Pettersson and Wallensteen, 2015: 536)

Thus, our conflict variable,

C O N F

, is set to 1 if there were 25 or more battle-related deaths resulting from the use of armed force between two parties, of which at least one of the parties is a government of a country in a year and zero otherwise.

Further, the definition of internal conflict also comes from the UCDP/PRIO Armed Conflict Dataset that classifies any armed conflict which took place inside the geographical territory of a country between the government of a state and internal opposition groups, without intervention from other states as an internal conflict (Pettersson and Wallensteen, 2015). For instance, a conflict involving the government and the dissident or opposition groups within a country. A conflict is defined as external if the parties to the conflict are two sovereign governments. To avoid selection bias, both conflicting and non-conflicting countries have been included in the model.

Control variables

Following the literature on the determinants of tourist flows, the estimation model includes a set of control variables that potentially affect international tourism flows (Gil-Pareja et al., 2007; Khadaroo and Seetanah, 2008; Martins et al., 2017). We control for the level of economic development of the origin and the destination countries by including real GDP per capita at PPP. It is likely that the flow of tourists to a country is correlated with the level of economic development of the originating or destination country (Shafiullah et al., 2019). Some tourists may prefer to visit advanced economies or developing economies while others may be indifferent. $G D P P C_{o}$ denotes origin country’s GDP per capita at PPP, while $G D P P C_{d}$ is destination country’s GDP per capita at PPP. Further, the population of the originating and destination are controlled for as the size of the destination and origin countries is more likely to affect international tourist flows. $P O P U_{o}$ represents the population of the originating country, while $P O P U_{d}$ is the population of the destination country.

We also control for the distance, $D I S T_{o d},$ between the country of origin and destination. Distance is used as a proxy to capture the cost of travel between the origin and the destination countries. As pointed out by Khadaroo and Seetanah (2008), air travel is the major mode of transportation for international tourism; however, the prevalence of price discrimination in the sale of air tickets makes it difficult for one to estimate the cost of travel between pairs of countries.

Cultural and language proximity plays a key role in international tourism. We control for the influence of official language on international tourism by including a dummy variable, $C O M O F F$ , that is set to 1 if the country of origin and destination shares a common official language, and 0 otherwise. As shown by Okafor et al. (2018), common unofficial language helps to promote international tourism. We therefore control for common unofficial language by using a dummy, $C O M N O F F$ , which takes the value 1 if the origin and destination countries share a common unofficial language that is spoken by at least 9% of the population and 0 otherwise.

Tourist preferences are more likely to influence international tourism flows. We capture tourist preferences with contiguity and colonial linkage dummies. Contiguity, $C O N T_{o d},$ takes the value 1 if the destination and origin countries share a border and 0 otherwise. $F C O L_{0}$ is set to 1 if the origin country is a former colony of the destination country and 0 otherwise, whereas $F C O L_{d}$ takes the value 1 if the destination country is a former colony of the origin country.

We also include an island dummy, $I S L A_{0 d},$ which is set to 1 if the origin country and destination country are an island and 0 otherwise, and a landlocked dummy, $L A N D_{o d},$ which is set to 1 if the origin and destination country are landlocked and 0 otherwise. Island nations tend to attract more tourists than landlocked countries. Tourists have the option of travelling by sea if a nation is an island which is not a possibility if a country is landlocked. However, the expected link between island or landlocked variable and international tourism may not hold if most tourists travel by air.

Summary statistics

Table 1 presents the summary statistics of the data used in the empirical analysis. Preliminary evidence from the unconditional means suggests that countries that spend moderately high and high levels on military tend to attract more tourists compared with countries that spend relatively low or moderate levels on security. It also appears that high levels of military spending is correlated with the high incidence of internal conflicts or armed conflicts broadly measured. This suggests that in the presence of conflicts, countries tend to allocate more resources for security and defence.

Table 1.

Summary statistics.

Variable	LOWMIL	MODMIL	MODHIMIL	HIGHMIL	All
Ln tourist flows	5.732 (3.171)	6.815 (3.184)	7.949 (2.863)	7.171 (2.898)	6.921 (3.137)
Conflict	0.034 (0.182)	0.274 (0.446)	0.324 (0.468)	0.361 (0.480)	0.248 (0.432)
Internal conflict	0.034 (0.182)	0.273 (0.446)	0.327 (0.469)	0.327 (0.469)	0.240 (0.427)
Ratio of military spending to GDP	0.755 (0.386)	1.385 (0.296)	2.267 (0.630)	3.515 (0.926)	1.967 (1.192)
Ln population (origin)	16.193 (1.814)	16.308 (1.852)	16.116 (1.878)	16.329 (1.765)	16.235 (1.831)
Ln population (destination)	16.157 (1.696)	16.618 (1.201)	18.649 (1.870)	17.089 (1.290)	17.140 (1.812)
Ln GDP per capita, PPP (origin)	9.258 (1.246)	9.454 (1.165)	9.257 (1.233)	9.351 (1.233)	9.329 (1.222)
Ln GDP per capita, PPP (destination)	8.443 (0.915)	8.734 (0.751)	8.724 (0.743)	8.853 (0.496)	8.687 (0.759)
Ln distance	8.658 (0.789)	8.633 (0.831)	8.744 (0.753)	8.392 (0.815)	8.610 (0.807)
Contiguity	0.036 (0.187)	0.032 (0.177)	0.067 (0.250)	0.047 (0.211)	0.046 (0.209)
Common official language	0.228 (0.419)	0.187 (0.390)	0.181 (0.385)	0.193 (0.394)	0.197 (0.398)
Common unofficial language	0.181 (0.385)	0.181 (0.385)	0.223 (0.416)	0.205 (0.404)	0.198 (0.398)
Former colony (destination)	0.018 (0.133)	0.016 (0.125)	0.012 (0.110)	0.013 (0.112)	0.015 (0.121)
Former colony (origin)	0.000 (0.000)	0.001 (0.030)	0.002 (0.049)	0.017 (0.129)	0.005 (0.070)
Landlocked	0.010 (0.097)	0.045 (0.207)	0.017 (0.128)	0.023 (0.151)	0.024 (0.152)
Island	0.059 (0.236)	0.013 (0.112)	0.000 (0.000)	0.011 (0.106)	0.021 (0.143)
Observations	15,554	15,495	15,980	14,920	61,949

Note: The figures without parentheses are means, while figures in parentheses are standard deviations. Ln denotes logarithms, LOWMIL is low military spending, MODMIL is moderate military spending, MODHIMIL is moderately high military spending and HIGHMIL is high military spending. PPP: purchasing power parity.

Further, destination countries that spend moderately high or high on military tend to be more populated than their counterparts, whereas military expenditure in the originating countries tends to increase with higher population levels. Originating countries that spend moderately high or too much on military appear to have a higher GDP per capita than those that spend low or moderately high on the military. GDP per capita at PPP in the destination countries tends to be higher for those that spend too much on military followed by those that spend moderately, moderately high, or those that devote smaller resources for security and defence.

In general, the influence of key determinants of international tourism such as distance, language, colonial linkage, landlock and island may differ according to the levels of relative military spending. For instance, countries that spend a larger share of GDP on military spending tend to be closer to each other, whereas countries that spend moderately high tend to be farther apart from each other.

Countries that spend a smaller share of GDP on the military are more likely to share a common official language than others, whereas those that spend moderately high on the military are more likely to share a common unofficial language than their counterparts. Countries with low military expenditure are more likely to be an island nation, whereas countries that spend moderately on the military are more likely to be landlocked than others.

Figure 1 shows the link between relative military spending and international tourist flows according to different levels of military spending. The horizontal axis measures the military spending as a share of GDP, whereas the vertical axis measures the natural logarithm of tourist arrivals. Preliminary evidence suggests that the interplay between relative military spending and international tourism is non-linear. At lower and moderate levels of military spending, military expenditure as a share of GDP is positively correlated with the number of tourist arrivals. As the level of military increases more than the moderate level, the link between the variables turns negative. At high levels of military spending, there seems to be no strong correlation between the two variables. This figure suggests that the moderating effect of military spending, if any, may vary with the levels of military spending. Military spending is an integral part of security spending. Countries that allocate a larger share of GDP for security and defence invariably spend more on the military than others.

Figure 1.

Relationship between tourist arrivals and relative military spending by level of military spending of destination countries. Note: Ln denotes logarithms, LOWMIL is low military spending, MODMIL is moderate military spending, MODHIMIL is moderately high military spending and HIGHMIL is high military spending.

Table 2 reports the test for multicollinearity. The ‘rule of thumb’ in the econometrics literature is that a tolerance level less than 0.1 or a variance inflation factor (VIF) larger than 10 signals multicollinearity problem. As shown in Table 2, there is no strong evidence suggesting the presence of multicollinearity problem.

Table 2.

Tests of multicollinearity: VIFs and tolerance-dependent variable: Ln tourist flow.

Variable	VIF	Tolerance
Common official language	2.96	0.338
Common unofficial language	2.87	0.349
Ln population (destination)	1.37	0.729
Ln distance	1.33	0.752
Conflict	1.27	0.787
Contiguity	1.26	0.793
Ratio of military spending to GDP	1.19	0.843
Ln GDP per capita, PPP (destination)	1.08	0.923
Ln GDP per capita, PPP (origin)	1.06	0.943
Ln population (origin)	1.06	0.944
Former colony (destination)	1.05	0.951
Landlocked	1.05	0.953
Island	1.05	0.954
Former colony (origin)	1.04	0.962
Mean VIF	1.4

Note: The ‘rule of thumb’ in the econometric literature is that a variance inflation factors greater than 10 or a tolerance level less than 0.1 is a sign of a severe multicollinearity problem. VIF: variance inflation factor; PPP: purchasing power parity.

Discussion of results

The parameter estimates obtained using OLS and the fixed effects estimators are reported in Table 3. The first three columns report OLS estimates where we controlled for year, origin and destination fixed effects. The last column reports fixed effects estimates where we controlled for year and country-pair fixed effects. In general, the parameter estimates obtained using different model specifications and estimation methods are consistent in terms of direction and significance. The parameter estimates and R ² reported in columns 1 to 3 are stable across different specifications. Obtaining similar R ² values with gravity model that is modified with the inclusion or exclusion of one or two variables is common in trade and tourism literature (Okafor et al., 2018; Ruiz and Vilarrubia, 2007). In general, the inclusion of country of origin fixed effects, country of destination fixed effects, year fixed effects or country-pair fixed effects tend to absorb a significant share of the variations in the data set. As a result, the R ² values for closely related models tend to be similar.

Table 3.

Link between international tourism, conflict and relative military spending.

	(1)	(2)	(3)	(4)
Variable	OLS	OLS	OLS	FE
Ln population (origin)	0.176** (0.074)	0.177** (0.075)	0.181** (0.075)	0.354*** (0.105)
Ln population (destination)	0.056 (0.136)	0.096 (0.136)	0.075 (0.136)	−0.099 (0.188)
Ln GDP per capita, PPP (origin)	0.753*** (0.051)	0.755*** (0.051)	0.756*** (0.051)	0.797*** (0.061)
Ln GDP per capita, PPP (destination)	1.220*** (0.054)	1.202*** (0.054)	1.200*** (0.054)	1.001*** (0.074)
Ln distance	−1.754*** (0.009)	−1.754*** (0.009)	−1.754*** (0.009)
Contiguity	1.801*** (0.036)	1.801*** (0.036)	1.801*** (0.036)
Common official language	1.092*** (0.025)	1.093*** (0.025)	1.092*** (0.025)
Common unofficial language	0.162*** (0.025)	0.162*** (0.025)	0.162*** (0.025)
Former colony (destination)	0.667*** (0.043)	0.667*** (0.043)	0.668*** (0.043)
Former colony (origin)	0.178** (0.076)	0.177** (0.076)	0.177** (0.076)
Landlocked	0.478*** (0.038)	0.477*** (0.038)	0.477*** (0.038)
Island	−0.113*** (0.040)	−0.114*** (0.040)	−0.113*** (0.040)
Conflict	−0.064*** (0.020)		−0.066*** (0.020)	−0.048*** (0.015)
Ratio of military spending to GDP		0.030** (0.013)	0.032** (0.013)	0.037*** (0.012)
Observations	61,949	61,949	61,949	61,949
R ² (within)	0.859	0.859	0.859	0.301

Note: OLS denotes ordinary least square and FE is fixed effects estimator. The dependent variable is log of tourist flows; Ln denotes natural logarithms. The numbers in parentheses in columns are robust standard errors. All regressions include constant but they are not reported. PPP: purchasing power parity.

*Significance at the 10% levels.

**Significance at the 5% levels.

***Significance at the 1% level.

All the control variables have expected sign except for the landlocked and island variables. The impacts of landlocked and island variables on international tourism are potentially dependent on the stage of development of different countries and the predominant mode of transportation across countries. Okafor et al. (2018) find that the impact of the landlocked variable is income and region dependent. For instance, the landlocked variable has a negative impact on international tourism in wealthy countries, whereas its impact is positive for low-income countries. Similarly, if most international tourists travel by air as opposed to travelling by sea, then island may not provide a boost for international tourism.

As reported in Table 3, the GDP per capita at PPP in the origin and destination has a significant positive impact on international tourism. Citizens from rich countries travel more compared to citizens from poor countries. Wealthy countries also have better tourist infrastructure and can afford to devote more resources to promote tourism compared to low-income countries. Population in the originating countries has a significant positive impact on international tourism, whereas the impact of the population in the destination is statistically insignificant. Countries with the large population demand more tourist services relative to countries with a smaller population.

As shown in Table 3, distance has a significant negative impact on international tourism. Contiguity, common official and unofficial languages and colonial linkages help to promote international tourism. These results are in line with the findings of previous research (for detailed discussion, see Okafor et al., 2018).

The variables of interest in this study, namely conflict and the ratio of Military Spending to GDP have the expected signs. Armed conflict broadly measured has a substantial negative impact on international tourism. A country that experiences any form of internal and/or external armed conflict attracts a lower flow of international tourists as compared to a country without armed conflict. For instance, as reported in columns 1, 3 and 4 of Table 3, a country with an episode of armed conflict receives approximately between 4.88% and 6.76% fewer tourists as compared to a country, which has no armed conflict. This finding is in line with the findings of previous studies (see, Hyndman, 2015; Saha and Yap, 2014). This finding clearly underscores the importance of peace and security in promoting international tourism.

On the contrary, the ratio of military expenditure to GDP has a positive and significant effect on tourist flows. The parameter estimates in columns 2 to 4 of Table 3 suggest that a 10% increase in relative military spending would lead to between a 3% and 4% increase in international tourist flows. In general, this finding suggests that countries that spend relatively more on military tend to attract higher tourist flows. A possible explanation for this relationship could be that relative increase in security spending creates an enabling environment for the host countries to better manage and prevent conflicts.

International tourists are attracted to countries that are perceived to be safe as well as having the capacity to manage existing or new conflicts. Some countries allocate a considerable amount of money towards the maintenance of security, and part of that allocation goes towards military expenditure as well. For instance, following the 2013 Westgate Mall attack and 2015 Garissa University attack, tourism industry which is one of the mainstays of Kenya’s economy was badly hit. There was a decrease in tourist numbers of 11.1% year-on-year in 2014, and a further decrease was projected for 2015 as official figures indicate that there was a 16.9% drop in tourist numbers for the first 9 months in 2015. To strengthen security and boost confidence, Keyna’s government allocated KSh112.5 billion ($1.2 billion) to the National Intelligence and defence (Oxford Business Group, 2015).

In the presence of armed conflict, an increase in security spending potentially help to dampen the negative impact of armed conflict on international tourism. In order to capture the moderating effect of relative military spending, we interacted armed conflict with different measures of the levels of military spending of destination countries. Table 4 presents the parameter estimates.¹ In general, the estimates confirm that military expenditure has a moderating impact on armed conflict; however, this effect varies according to the levels of relative military spending. At lower levels of military expenditure, there is no significant moderation effect while at higher levels the moderation effect is negative. Further, moderate level of military spending interacted with armed conflict has a substantial positive effect on international tourist flows.

Table 4.

Link between international tourism, conflict and relative military spending by levels of relative military spending of destination countries.

Variable	LOWMIL	MODMIL	MODHIMIL	HIGHMIL
Conflict	−0.06*** (0.02)	−0.10*** (0.03)	−0.06** (0.03)	−0.05** (0.02)
LOWMIL	0.28 (0.33)
Conflict × LOWMIL	−0.03 (0.07)
MODMIL		−0.29 (0.33)
Conflict × MODMIL		0.08** (0.04)
MODHIMIL			4.81*** (0.80)
Conflict × MODHIMIL			−0.02 (0.04)
HIGHMIL				1.92*** (0.12)
Conflict × HIGHMIL				−0.12** (0.05)
Year effects	Yes	Yes	Yes	Yes
Country effects (origin)	Yes	Yes	Yes	Yes
Country effects (destination)	Yes	Yes	Yes	Yes
Other controls	Yes	Yes	Yes	Yes
Observations	61,949	61,949	61,949	61,949
R ²	0.86	0.86	0.86	0.86

Note: The dependent variable is log of tourist flows. The numbers in parentheses in columns are robust standard errors. All regressions include constant but they are not reported. LOWMIL is low military spending, MODMIL is moderate military spending, MODHIMIL is moderately high military spending and HIGHMIL is high military spending.

*Significance at the 10% levels.

**Significance at the 5% levels.

***Significance at the 1% level.

As shown in Table 4 columns 1 to 4, armed conflict has a significant negative impact on international tourist flows irrespective of the levels of relative military spending of the destination countries. The parameter estimates from column 1 suggest that low level of military spending or its interaction with armed conflict has no statistically significant effect on international tourist flows. Perhaps, countries that spend too little on security are not equipped to manage or prevent armed conflicts. In general, international tourists are not attracted to countries with poor security credentials or countries that are notorious for frequent serious security lapses and failures.

As shown in column 2, a moderate level of military spending has a substantial dampening effect on the negative impact of armed conflict. While the moderate level of military spending has no significant impact on international tourism, the joint effect of moderate military spending and armed conflict is positive and significant. The parameter estimate suggests that in the presence of armed conflict, countries that spend moderately on the military attract about 8.3% more international tourists relative to other countries with different levels of relative military spending. This finding underscores the importance of allocating adequate resources for security, especially during the period of armed conflict or serious security challenges. In general, adequate resources in this instance translates into better management and prevention of conflicts. This, in turn, helps to strengthen both local and international confidence as well as promote international tourism.

It could also be the case that moderate level of military spending in the presence of conflict sends a signal to the international community that government places a high priority on security. The allocation of adequate resources for the maintenance of law and order without sacrificing expenditures on other productive avenues helps to boost tourism and investment in general. As highlighted earlier, in the aftermath of terrorist attacks, Kenya’s government increased security spending. As a result, there is a growing sense of optimism in the tourism sector after years of anxiety. In general, the increase in security spending has led to improved security. The improvement in security created an enabling environment for the tourism industry to rebound. The strengthening of confidence in security led to an increase in hotel occupancy in the coastal region (Mwakio, 2017).

As shown in columns 3 and 4, the parameter estimates for moderately high and high levels of relative military spending is positive and statistically significant. This suggests the positive link between military spending and international tourism holds when the level of military spending is moderately high or high. As discussed earlier, this could be due to the fact that increased military spending sends a signal to the tourists that country is safe and secure and measures are taken by the government to ensure the safety and security.

The effect of armed conflict interacted with moderately high levels of military spending is insignificant, whereas the effect is negative when interacted with high level of military spending. The effect of armed conflict conditional on the high level of military spending is −0.17, which is almost three times as high compared to relative to low and moderately high military expenditure and approximately eight times more compared to moderate military spending. This suggests that in the presence of armed conflict, countries that spend the most on military attract approximately 11.3% fewer international tourists compared to other countries with different levels of military spending. This finding suggests that excessive military spending amidst armed conflict discourages tourists from visiting a destination. Subsequently, in the presence of armed conflict, high military spending might signal that a country is preparing for an escalation of the existing armed conflict. Further, excessive military spending during the period of armed conflict diverts resources from the private sector, including the tourism industry, and thus tends to depress tourism investment.

Table 5 presents parameter estimates for the link between tourism, conflict and military spending by levels of military spending based on the region of destination. As expected, armed conflict has a significant negative effect on international tourism regardless of the levels of military spending across different regions. Overall, the findings are consistent with those reported in Table 4 in terms of the moderating effect of military spending on armed conflict according to levels of military spending.

Table 5.

Link between international tourism, conflict and relative military spending by levels of relative military spending based on region of destination.

Variable	LOWMIL_{CA,OC & SSA}	MODMIL_{LATCA & SEA}	MODHIMIL_{EA, EUR & SA}	HIGHMIL_MENA
Conflict	−0.06*** (0.02)	−0.10*** (0.03)	−0.05** (0.02)	−0.05** (0.02)
LOWMIL	5.07*** (0.25)
Conflict × LOWMIL	−0.02 (0.06)
MODMIL		0.30 (0.33)
Conflict × MODMIL		0.10*** (0.04)
MODHIMIL			−1.22*** (0.25)
Conflict × MODHIMIL			−0.04 (0.04)
HIGHMIL				−1.02*** (0.35)
Conflict × HIGHMIL				−0.12** (0.06)
Year effects	Yes	Yes	Yes	Yes
Country effects (origin)	Yes	Yes	Yes	Yes
Country effects (destination)	Yes	Yes	Yes	Yes
Other controls	Yes	Yes	Yes	Yes
Observations	61,949	61,949	61,949	61,949
R ²	0.86	0.86	0.86	0.86

Note: The dependent variable is log of tourist flows. The numbers in parentheses in columns are robust standard errors. All regressions include constant but they are not reported. LOWMIL_{CA,OC & SSA} is low military spending for groups in Central Asia (CA), Oceania (OC) and Sub-Saharan Africa (SSA) regions; MODMIL_{LATCA & SEA} is moderate military spending for groups in Latin American and Caribbean (LATCA) and Southeast Asia (SEA); MODHIMIL_{EA, EUR & SA} is moderately high military spending for groups in East Asia (EA), Europe (EUR) and South Asia (SA); and HIGHMIL_MENA is high military spending for groups in Middle East and North Africa (MENA).

*Significance at the 10% levels.

**Significance at the 5% levels.

***Significance at the 1% level.

As shown in column 1 of Table 5, low level of military spending has a significant positive impact on international tourism for countries in regions such as CA, OC and SSA. The positive effect of low levels of military spending for these countries might emanate from the perception of tourists that countries in CA and SSA are politically unstable and prone to conflict. As a result, even relatively small expenditure on military sends a positive signal to the tourists about the safety and security of a country in these regions. On the contrary, there is no strong evidence suggesting that low military spending has any moderating effect on armed conflict for countries in regions such as CA, OC and SSA. In general, countries in these regions do not allocate a lot of money for defence and security.

Interestingly, moderate levels of military spending for countries in LATCA and SEA completely offset the negative effect of conflict on international tourism. This indicates that in the presence of armed conflict countries that allocate moderate levels of military spending will receive almost 10.5% higher tourists than their counterparts that allocate different levels of military spending. This finding highlights the importance of striking a balance between military spending and other productive private expenditures for attracting international tourists. Apart from security considerations, tourists are more likely to travel to countries with adequate tourism infrastructure.

As shown in columns 4 and 5 of Table 5, for countries in regions such as East Asia (EA), Europe (EUR), South Asia (SA), Middle East and North Africa (MENA) that spend moderately high or high on the security, military spending has a significant negative effect on international tourism. This finding suggests that the effect of military spending on international tourism is region dependent. Countries that spend a larger share of GDP on military tend to divert resources away from the private sector as government competes with the private sector for resources.

Further, there is no strong evidence that moderately-high military spending for countries in EA, EUR and SA has any significant dampening effect on the negative impact of armed conflict on international tourism. On the contrary, countries in the MENA that spend a larger fraction of GDP on defence, the joint effect of relative military spending with armed conflict is negative and statistically significant. Specifically, in the presence of armed conflict for countries in the MENA region, high levels of military expenditure tend to depress the number of tourist arrivals by approximately 11.3% compared to countries that devote different levels of relative military spending to defence and security. This finding reinforces the notion that high military expenditure in the presence of conflict sends a negative signal to the international community. Perhaps, it signals that conflict is spiralling out of control.

We perform two checks in order to test the sensitivity of the results. First, we use a measure of internal conflict in lieu of overall conflict to test if our estimates are driven by a particular type of conflict. The parameter estimates are qualitatively similar to the previous estimates. The estimates are not reported, but they are available from the authors upon request.

Second, we check if the estimates are particularly vulnerable to endogeneity concern as discussed earlier by using lagged conflict and lagged ratio of relative military spending and the interaction of lagged conflict and lagged ratio of relative military spending in lieu of their respective current variables. We rerun the base model as well as regressions in which the sample was classified by the levels of relative military spending of the destination countries. As expected the parameter estimates are qualitatively similar to previous estimates. The parameter estimates are not reported, but they are available from the authors upon request.

Conclusion

This study uses an augmented gravity framework to explore the effect of armed conflict on international tourist flows while considering the role of military spending in the underlying relationship. The empirical analysis is performed with the use of panel data on bilateral tourism for 188 countries over the period of 1995–2015. In general, we show that armed conflict has a significant negative impact on the number of international tourist arrivals regardless of the stage of development, geographic location or the level of relative military spending. Armed conflicts tend to dampen the international tourism attractiveness of destination countries.

Our findings also show that security spending as captured by relative military spending helps to promote international tourism. However, the impact of military spending depends on the level of relative military spending and geographical location. This is consistent with the notion that military spending has both substitution and complementary effect on international tourism.

The moderating effect of military spending on the link between armed conflict and international tourism depends on the levels of relative military spending as well as geographical location. For instance, in the presence of armed conflict, moderate level of relative military helps to promote international tourism, whereas a high level of relative military spending tends to hinder international tourism. In general, countries in the regions such as LATCA and SEA that allocate a moderate amount of resources for security are more attractive to tourists compared to the countries in the regions, such as MENA that devote a larger share of GDP for security and defence.

The findings of this article have important policy implications. First, in the presence of armed conflict, policymakers should ensure that adequate resources are allocated for security and defence. The allocated resources should be utilized efficiently to manage and prevent future armed conflicts in order to create an enabling environment for tourism to thrive. Second, policymakers should not rely heavily on military strength as the only tool for conflict resolution. In the presence of armed conflict, diplomacy coupled with the moderate level of security spending contributes towards timely resolution of conflict. In this instance, moderate level of military spending helps to mitigate the negative impact of armed conflict on international tourism. On the contrary, excessive military spending and neglect of diplomacy could lead to an escalation of armed conflict, crowding out of tourism investment and declining international tourism attractiveness of destination countries.

Supplemental Material

Supplemental Material, Appendix - Armed conflict, military expenditure and international tourism

Supplemental Material, Appendix for Armed conflict, military expenditure and international tourism by Usman Khalid, Luke Emeka Okafor and Nusrate Aziz in Tourism Economics

Footnotes

Declaration of conflicting interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Usman Khalid

Note

Supplemental material

Supplemental material for this article is available online.

References

Aizenman

Glick

(2006) Military expenditure, threats, and growth. The Journal of International Trade & Economic Development 15(2): 129–155.

Albalate

Bel

Elias

(2012) Institutional determinants of military spending. Journal of Comparative Economics 40(2): 279–290.

Araña

León

(2008) The impact of terrorism on tourism demand. Annals of Tourism Research 35(2): 299–315.

Aziz

Asadullah

(2017) Military spending, armed conflict and economic growth in developing countries in the post-Cold War era. Journal of Economic Studies 44(1): 47–68.

Aziz

Khalid

(2019) Armed conflict, military expenses and FDI inflow to developing countries. Defence and Peace Economics 30(2): 238–251.

Barro

Sala-i-Martin

(2004) Economic Growth, 2nd ed. Cambridge: The MIT Press.

Benoit

(1978) Growth and defense in developing countries. Economic Development and Cultural Change 26(2): 271–280.

Buigut

Amendah

(2016) Effect of terrorism on demand for tourism in Kenya. Tourism Economics 22(5): 928–938.

Collier

(2006) War and military spending in developing countries and their consequences for development. The Economics of Peace and Security Journal 1(1): 10–13.

10.

Cothren

(2002) A model of military spending and economic growth. Public Choice 110(1/2): 121–142.

11.

Deger

Smith

(1983) Military expenditure and growth in less developed countries. Journal of conflict resolution 27(2): 335–353.

12.

Dogru

Sirakaya-Turk

Crouch

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

13.

Drakos

Kutan

(2001) Regional effects of terrorism on tourism: evidence from three Mediterranean countries. Journal of Conflict Resolution 47(5): 621–641.

14.

Dunne

Smith

Willenbockel

(2005) Models of military expenditure and growth: a critical review. Defence and Peace Economics 16(6): 449–461.

15.

Dunne

Nikolaidou

Smith

(2002) Military spending, investment and economic growth in small industrialising economies. South African Journal of Economics 70(5): 789–790.

16.

Eilat

Einav

(2004) Determinants of international tourism: a three-dimensional panel data analysis. Applied Economics 36(12): 1315–1327.

17.

Enders

Sandler

(1991) Causality between transnational terrorism and tourism: the case of Spain. Terrorism 14(1): 49–58.

18.

Enders

Sandler

Parise

(1992) An econometric analysis of the impact of terrorism on tourism. Kyklos 45(4): 531–554.

19.

Frederiksen

Looney

(1982) Defense expenditures and economic growth in development countries: some further empirical evidence. Journal of Economic Development 7(1): 113–126.

20.

Gil-Pareja

Llorca-Vivero

Mart’inez-Serrano

(2007) The effect of EMU on tourism. Review of International Economics 15(2): 302–312.

21.

Hajibaba

Boztuğ

Dolnicar

(2016) Preventing tourists from canceling in times of crises. Annals of Tourism Research 60: 48–62.

22.

Head

Mayer

Ries

(2010) The erosion of colonial trade linkages after independence. Journal of International Economics 81(1): 1–14.

23.

Hyndman

(2015) The securitisation of Sri Lankan tourism in the absence of peace. Stability: Int. Stability: International Journal of Security and Development 4(1): 1–16.

24.

Issa

Altinay

(2006) Impacts of political instability on tourism planning and development: the case of Lebanon. Tourism Economics, 12(3): 361–381.

25.

Jiao

(2018) Duterte plays catch-up on defense spending with cash for troops. Available at: https://www.bloomberg.com/news/articles/2018-08-15/duterte-plays-catch-up-on-defense-spending-with-cash-for-troops (accessed 18 April 2019).

26.

Khadaroo

Seetanah

(2008) The role of transport infrastructure in international tourism development: a gravity model approach. Tourism Management 29(5): 831–840.

27.

Khalid

(2017) The effect of trade and political institutions on economic institutions. The Journal of International Trade & Economic Development 26(1): 89–110.

28.

Khalid

Okafor

Shafiullah

(2019) The effects of economic and financial crises on international tourist flows: a cross-country analysis. Journal of Travel Research. DOI: 10.1177/0047287519834360

29.

Kim

(2016) The effects of inter-state conflicts on foreign investment flows to the developing world: enduring vs ephemeral risk of conflicts. International Political Science Review 37(4): 422–437.

30.

Kinuthia

(2017) Export spillovers: comparative evidence from Kenya and Malaysia. Journal of African Economies 26(1): 24–51.

31.

Landau

(1996) Is one of the ‘peace dividends’ negative? Military expenditure and economic growth in the wealthy OECD countries. The Quarterly Review of Economics and Finance 36(2): 183–195.

32.

Lee

Y-S

(2006) The Korean war and tourism: legacy of the war on the development of the tourism industry in South Korea. International Journal of Tourism Research 8(3): 157–170.

33.

Vashchilko

(2010) Dyadic military conflict, security alliances, and bilateral FDI flows. Journal of International Business Studies 41(5): 765–782.

34.

Liu

Pratt

(2017) Tourism’s vulnerability and resilience to terrorism. Tourism Management 60: 404–417.

35.

Llorca-Vivero

(2008) Terrorism and international tourism: new evidence. Defence and Peace Economics 19(2): 169–188.

36.

Maizels

Nissanke

(1987) The causes of military expenditure in developing countries. In: Deger

West

(eds) Defence Security and Development, Great Britain: Frances Pinter. Washington: Frances Pinter, pp. 129–139.

37.

Martins

Gan

Ferreira-Lopes

(2017) An empirical analysis of the influence of macroeconomic determinants on World tourism demand. Tourism Management 61: 248–260.

38.

Matsiko

(2019) Museveni doubles military spending. Available at: https://www.independent.co.ug/museveni-doubles-military-spending/ (accessed 18 April 2019).

39.

Mwakio

(2017) Security Pushes Up Numbers in Coast Hotels. Kenya: Standard Digital.

40.

Narayan

Nguyen

(2016) Does the trade gravity model depend on trading partners? Some evidence from Vietnam and her 54 trading partners. International Review of Economics & Finance 41(Supplement C): 220–237.

41.

Nassani

Zaman

Aldakhil

, et al. (2017) War economy and pleasure: assessing the effects of military expenditure on tourism growth. Quality & Quantity 51(4): 1733–1754.

42.

Neumayer

(2004) The impact of political violence on tourism: dynamic cross-national estimation. Journal of Conflict Resolution 48(2): 259–281.

43.

Nigh

(1985) The effect of political events on United States direct foreign investment: a pooled time-series cross-sectional analysis. Journal of International Business Studies 16(1): 1–17.

44.

Okafor

Khalid

Then

(2018) Common unofficial language, development and international tourism. Tourism Management 67: 127–138.

45.

Oxford Business Group (2015) The Report: Kenya 2016. The UK Oxford Business Group.

46.

Pakistan’s army is getting serious about defeating domestic terrorism (2018) The Economist. Available at: https://www.economist.com/asia/2018/03/01/pakistans-army-is-getting-serious-about-defeating-domestic-terrorism

47.

Pettersson

Wallensteen

(2015) Armed conflicts, 1946–2014. Journal of peace research 52(4): 536–550.

48.

Pizam

Mansfeld

(2006) Chapter 1 – toward a theory of tourism security. In: Pizam

Mansfeld

(eds) Tourism, Security and Safety. Boston: Butterworth-Heinemann, pp. 1–27.

49.

Pratt

Liu

(2016) Does tourism really lead to peace? A global view. International Journal of Tourism Research 18(1): 82–90.

50.

Raza

Jawaid

(2013) Terrorism and tourism: a conjunction and ramification in Pakistan. Economic Modelling 33: 65–70.

51.

Ruiz

Vilarrubia

(2007) The wise use of dummies in gravity models: export potentials in the Euromed region. Working Paper Series, Number 0720, BANCO DE ESPAÑA, Madrid .

52.

Saha

Yap

(2014) The moderation effects of political instability and terrorism on tourism development: a cross-country panel analysis. Journal of Travel Research 54(4): 509–521.

53.

Santana-Gallego

Ledesma-Rodríguez

Pérez-Rodríguez

(2016) International trade and tourism flows: an extension of the gravity model. Economic Modelling 52(Part B): 1026–1033.

54.

Shafiullah

Okafor

Khalid

(2019) Determinants of international tourism demand: evidence from Australian states and territories. Tourism Economics 25(2): 274–296.

55.

Shieh

Lai

Chang

(2002) The impact of military burden on long-run growth and welfare. Journal of Development Economics 68(2): 443–454.

56.

Silva

Tian

(2019) The crucial role of the military in the Venezuelan crisis. Available at: http://www.ipsnews.net/2019/04/crucial-role-military-venezuelan-crisis/ (accessed 18 April 2019).

57.

Sinclair

(1998) Tourism and economic development: a survey. The Journal of Development Studies 34(5): 1–51.

58.

SIPRI (2018) Global military spending remains high at $1.7 trillion [Press release]. Available at: https://www.sipri.org/media/press-release/2018/global-military-spending-remains-high-17-trillion.

59.

SIPRI (2019) Arms transfers and military spending. Stockholm International Peace and Research Institute (SIPRI), Sweden. Available at: https://www.sipri.org/research/armament-and-disarmament/arms-transfers-and-military-spending. (accessed 15 February 2019)

60.

Smith

(1977) Military expenditure and capitalism. Cambridge Journal of Economics 1(1): 61–76.

61.

Smith

(1978) Military expenditure and capitalism: a reply. Cambridge Journal of Economics 2(3): 299–304.

62.

Smith

(1980) Military expenditure and investment in OECD countries, 1954–1973. Journal of comparative economics 4(1): 19–32.

63.

United Nations (2010) International Recommendations for Tourism Statistics 2008 (Series M No. 83/Rev.1). Department of Economic and Social Affairs, Statistics Division, New York: United Nations Publications.

64.

Weaver

(2011) Tourism and the military: pleasure and the war economy. Annals of Tourism Research 38(2): 672–689.

65.

Whynes

(1979) The Economics of Third World Military Expenditure. Berlin; Austin: Springer; University of Texas Press.

66.

World Bank (2017) World Bank Development Indicator. Washington, USA: The World Bank.

67.

World Tourism Organization (2017a) UNWTO Tourism Highlights: 2017 Edition, UNWTO, Madrid. Available at: https://doi.org/10.18111/9789284419029

68.

World Tourism Organization (2017b) UNWTO’s Tourism Statistics, CD-ROM. Madrid, Spain: UNWTO.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.37 MB