Trade in the face of terror: Examining Turkey's export dynamics

Ünlü, Evrim Zeynep Article Trade in the face of terror: Examining Turkey's export dynamics Peace Economics, Peace Science and Public Policy (PEPS) Provided in Cooperation with: De Gruyter Brill Suggested Citation: Ünlü, Evrim Zeynep (2025) : Trade in the face of terror: Examining Turkey's export dynamics, Peace Economics, Peace Science and Public Policy (PEPS), ISSN 1554-8597, De Gruyter, Berlin, Vol. 31, Iss. 1, pp. 91-122, https://doi.org/10.1515/peps-2024-0023 This Version is available at: https://hdl.handle.net/10419/333339 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Evrim Zeynep Ünlü* Trade in the Face of Terror: Examining Turkey’s Export Dynamics https://doi.org/10.1515/peps-2024-0023 Received May 8, 2024; accepted November 20, 2024; published online December 13, 2024 Abstract: In the preceding 11 years, Turkey has witnessed 925 terrorist incidents, resulting in the unexpected passing of 1,439 people. Concurrently, the neighbouring nations, Syria, and Iraq, have also experienced a multitude of terrorist activities. This study’s main objective is to explore how these events have affected Turkey’s exports and intra-industry trade. Theinvestigation is centred onTurkey’s trade relationships with 71 countries, representing its primary trading partners, employing a one-way trade model. Utilizing the Poisson pseudo maximum likelihood (PPML) estimator, the study spans 2012 to 2022. The findings underscore the adverse impact of domestic terrorist incidents, casualties in other countries, and the global terrorism index on Turkey’s exports. Surprisingly, the investigation finds a positive connection between Turkey’s exports and terrorist incidents in both neighbouring and non-bordering countries. Furthermore, the empirical analysis sheds light on Turkey’s engagement in intra-industry trade concerning manufacturing industry and GDP similarities. However, the scenario changes for information and communication technology (ICT) exports, revealing an inter-type of industry trade. Keywords: empirical studies of trade; international conflicts; political economy JEL Classification: F14; F51; F50 1 Introduction Terrorism, as defined by Enders and Sandler (2000), is the intentional use of deadly violence or threats to further partisan goals, instilling fear or intimidation among a sizable portion of the population. A key element of this definition is the political motive behind terrorist actions, distinguishing them from ordinary criminal activities that lack such motives. Additionally, terrorism is characterized using extreme violence, often intended to attract media attention. *Corresponding author: Evrim Zeynep Ünlü, Independent Researcher, Istanbul, Türkiye, E-mail: [email protected]. https://orcid.org/0000-0002-2045-5702 Peace Econ. Peace Sci. Pub. Pol. 2025; 31(1): 91–122 Open Access. © 2024 the author(s), published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 International License. Terrorism has long been recognized as a significant disruptor of economic stability and growth. Globally, terrorist activities have been shown to undermine investor confidence, reduce foreign direct investment, and disrupt trade flows, leading to broader economic instability (Abadie and Gardeazabal 2003; Gaibulloev and Sandler 2009, 2019). The economic consequences of terrorism extend far beyond immediate destruction. Terrorism fosters a climate of uncertainty, leading to behavioural changes among economic agents. This can result in reduced investment, decreased consumer spending, and disruptions to international trade. Additionally, the increased costs associated with terrorism, such as higher shipping costs and insurance premiums, can be passed on to consumers, further impacting economic activity (Bandyopadhyay 2017). The relationship between terrorism and international trade has been of particular interest in recent years, with studies indicating that terrorism can significantly reduce trade flows, disrupt supply chains, and lead to the reallocation of trade routes (Bandyopadhyay 2016). These effects are often compounded in regions with high levels of geopolitical instability, where terrorism worsens existing economic vulnerabilities (Bandyopadhyay, Sandler, and Younas 2014). Terrorism not only imposes direct human costs but also generates significant economic disruptions, particularly through its impact on international trade. Research has shown that terrorism can create spillover effects, where a terrorist event in one country adversely affects the bilateral trade of neighbouring nations (Pham and Doucouliagos 2017). These spillover effects are long-lasting, often persisting for up to five years after the event and can arise even from incidents with a small death toll. The mechanisms through which terrorism impacts trade include increased trade costs due to stricter security measures, heightened regulatory burdens, and psychological distress among economic agents (Pham and Doucouliagos 2017). Given Turkey’s geopolitical position, understanding these spillover effects is vital. The potential increase in trade costs and resulting frictions are particularly relevant for Turkey, which shares borders with several countries that have experienced significant terrorist activity. This study aims to build on the existing literature by examining how terrorism, both within Turkey and in neighbouring countries, affects Turkey’s trade dynamics, with a specific focus on intra-industry trade. Although the economic impacts of terrorism are well-documented, research specifically examining its effect on intra-industry trade, especially in developing economies like Turkey, remains limited. Turkey’s unique position, across both Europe and the Middle East, and its experience with both domestic and transnational terrorism, make it a critical case study for understanding these dynamics (Bilgel and Karahasan 2019). 92 E. Z. Ünlü While some studies examined the effects of terrorist incidents on different areas, for instance, on the Turkish stock market (Aksoy and Demiralay 2019; Gok, Demirdogen, and Topuz 2020), on foreign direct investment (FDI) (Çetin, Keser, and Ay 2019; Ibrahim and Ari 2021) and on domestic bank loans (Ilarslan and Yildiz 2022). To the best of my knowledge, no study has explored the relationship between trade, intra-industry trade, and terrorism, especially regarding Turkey. Intra-industry trade allows workers and businesses to improve their skills and expertise in specific products, encouraging innovation and leveraging economies of scale. Consequently, a close analysis of how terrorism impacts trade, particularly intra-industry trade, is essential for understanding the unique challenges that could affect Turkey. Intra-industry trade allows workers and businesses to improve their skills and expertise in specific products, encouraging innovation and leveraging economies of scale. Consequently, a close analysis of how terrorism impacts trade, particularly intra-industry trade, is essential for understanding the unique challenges that could affect Turkey. Despite the significance of intra-industry trade, very few studies have employed gravity models to examine this type of trade, such as (Leitão, Faustino, and Yoshida 2010; Wakasugi 2007; Yoshida, Leitão, and Faustino 2009). However, these studies did not focus on the impact of terrorism, particularly in developing economies like Turkey. This gap in the literature further underscores the importance of the present study. To achieve the research goals, six unique one-way gravity models were implemented, employing terrorism data sourced from the Institute for Economics and Peace (IEP) (2023). These models were designed to address the following inquiries: (i) What impact do the global terror index (GTI), the frequency of terrorism incidents, and the number of fatalities among Turkey’s trading partners have on Turkey’s exports? (ii) In light of terrorist attacks, how do Turkey’s exports be impacted by the economic similarities with its trade partners? (iii) What effect does Turkey’s manufacturing industry similarities with those of the countries it trades with have on Turkey’s exports in the case of terrorist incidents? (iv) How does terrorism affect Turkey’s intra-industry trade, introducing a novel variable-relative endowment (RELEN) of information and communication technology (ICT) service exports compared to population size to capture the intra-industry trade context? This study has two major contributions to the literature. First, it considers three distinct terrorism variables: the global terror index (GTI), the number of incidents (INC), and the number of fatalities (FTL) in each incident. Second, the study addresses the terror-trade relationship with a one-way gravity model tailored to Turkey and its top 71 export destination countries. The study is arranged as follows: Section 2 provides comprehensive definitions relevant to the research context. Section 3 includes an in-depth review of the Trade and Terror: Turkey’s Export Dynamics 93 literature, current research, and relevant hypotheses about the relationship between international trade and terrorism. Section 4 outlines the empirical strategy and explain data sources. Section 5 is devoted to the model estimation process and the presentation of the estimation results. Section 6 discusses the findings in the context of existing literature, while Section 7 summarizes the key insights and suggests avenues for future research. 2 Theoretical Framework The gravity model draws its theoretical basis from Newton’s law of universal gravitation. This model proposes that the volume of trade between two countries is directly related to their economic sizes and inversely related to the distance between them. This idea originates from Tinbergen (1962), who applied Newton’s gravity theory; the model mathematically expresses the total trade (TT) between countries ‘i’ and ‘j’as a function of their gross domestic products (GDP), distance, and an error term. This can be expressed mathematically, TTij =β0⎛ ⎝GDPβ1 iGDPβ2 j Dβ3 ij ⎞ ⎠εij (1) While the theoretical foundation of the gravity model is widely accepted, modern applications have incorporated several adjustments to address economic and econometric challenges, particularly those raised by Anderson and Wincoop (2003) in relation to Multilateral Resistance Terms (MRT). MRT refers to the concept that trade between any two countries depends not only on bilateral factors (such as distance or tariffs) but also on their trade relationships with all other countries. This introduces the need for time-varying importer and exporter fixed effects to control for the unobserved multilateral resistance affecting trade flows. In standard gravity models, failure to account for these MRTs can lead to biased estimates. To address the econometric challenges presented by Silva and Tenreyro (2006), specifically, the issue of heteroskedasticity and zero trade flows –the Poisson pseudo maximum likelihood (PPML) estimator has become a widely adopted solution. Unlike the traditional log-linearization of the gravity equation, which can produce inconsistent estimates when there are zero trade flows, the PPML estimator handles the multiplicative nature of the trade model in its original form, mitigating biases caused by zero trade observations and heteroskedasticity. The author employed the Poisson pseudo maximum likelihood high dimensional fixed effects (PPMLHDFE) command, as developed by Correia, Guimarães, and Zylkin (2020), for its ability to handle 94 E. Z. Ünlü high-dimensional fixed effects efficiently. This Stata command builds upon the Poisson pseudo maximum likelihood (PPML) estimator and enhances computational performance, particularly when dealing with large datasets and numerous dummy variables. The primary motivation for employing PPML estimator in this study is its ability to effectively address both the zero-trade flow problem and the numerous dummy variable problem, while also accounting for multilateral resistance via importer fixed effects. The decision to exclude exporter fixed effects in this case is justified by the fact that Turkey is the sole exporter in the model, thus negating the need for such effects, as there is no variation in Turkey’s characteristics across observations. Given the presence of large dependent variablevalues (i.e. trade data reportedin millions), the PPML estimator is preferred over alternatives like the Pooled Ordinary Least Squares (POLS) or Fixed Effects (FE) estimators. POLS, for instance, fails to account for heteroskedasticity, while FE approaches may introduce collinearity issues when multiple dummy variables are included. The model specification is expressed as follows: TTij =exp[β0+β1ln(GDPi)+β2ln(GDPj)−β3ln(Distanceij)+μij]εij (2) In this formulation, the inclusion of high-dimensional fixedeffects–such as importer fixed effects –addresses multilateral resistance and controls for time-varying factors that affect trade flows. Importantly, the model’s ability to handle the zero-trade issue and mitigate potential biases stemming from heteroskedasticity makes PPML the appropriate choice. 3 Literature Review Sandler and Enders (2008, 17–47) underscored the diverse economic ins and outs of terrorism and identified how targeted nations bear financial burdens. These implications span trade restrictions, infrastructure destruction, the diversion of Foreign Direct Investment (FDI), and changes in public investment toward security. Noteworthy research has studied the impact of terrorism on various economic variables, including its impact on FDI (Arif, Rawat, and Khan 2020; Dimitrova, Triki, and Valentino 2022; Lanouar and Shahzad 2021; Polyxeni and Theodore 2019). A comprehensive understanding of these connections is essential for assessing the broader impact on the gross domestic product (GDP) and overall economic growth. Simultaneously, terrorist attacks impose numerous costs on nations, causing tangible damages to critical infrastructure and property, as well as intangible losses like human lives, labour, and increased insurance risk premiums. The consequences Trade and Terror: Turkey’s Export Dynamics 95 extend to reduced business profitability, a decline in tourist visits, and an increased demand for enhanced security measures. These multifaceted economic implications highlight the necessity of comprehending the interconnectedness of terrorism and various economic sides. For example, Bilgel and Karahasan (2017) investigated the impact of terrorism on Turkish exports. They revealed that between 1988 and 2001 when the study was conducted, income per person in the country’s eastern regions declined by 6.6 %. While numerous studies have explored the relationships between GDP, imports, exports, FDI, and terrorism, none comprehensively understand the dynamics surrounding trade in the same industry or the impacts of terrorism. Haq, Ullah, and Iqbal (2018) studied the effects of terror incidents on bilateral trade, utilizing a gravity model across 50 developing and developed nations with intense terror incidents from 1990 to 2013. Their research revealed that terrorism exerts an unfavourable effect on imports and exports, particularly when developing nations trade with wealthier counterparts. Similarly, Baronchelli, Caruso, and Ricciuti (2022) used a gravity model to investigate the impact of arms embargoes on the trade of Small Arms and Light Weapons (SALW) from 1990 to 2017. Their findings showed that while total embargoes reduced SALW trade, EU embargoes were more effective than UN embargoes, which had no measurable impact. This study demonstrates how external restrictions, akin to the pressures terrorism exerts on trade flows, can be effectively analysed using gravity models. Bandyopadhyay, Sandler, and Younas (2016) extended this investigation to include the analysis of terror incidents’effects on bilateral trade. Their findings indicated that compared to domestic terrorism, terrorist activities abroad have a more pronounced negative impact on overall manufactured trade, including exports, imports, and other categories. Interestingly, this negative effect is more perceivable on imports than on exports. Chaudhury and Sinha (2019) examined the effects of terrorist activities on GDP and international commerce by using a panel vector autoregressive (VAR) model to examine 44 industrialized and developing nations between 1991 and 2015. Their findings demonstrated that while terror attacks have a more significant negative impact on foreign trade in underdeveloped countries, their influence is less noticeable in rich countries. Julius, Azu, and Muhammad (2019) contributed to the discourse by examining the implications of terror incidents on commerce within the South African Development Community (SADC) countries, utilizing the Poisson pseudo maximum likelihood (PPML) estimator. Their findings suggested that terrorism has an extremely low effect on bilateral trade, with lowand middle-income countries experiencing 96 E. Z. Ünlü positive effects. In contrast, high-income countries observe negative impacts from terror activities. In a departure from the direct impact of terrorism on trade, Okafor and Piesse (2018) shifted the focus towards understanding the factors contributing to terrorism. Their examination spanned from 2005 to 2014, involving 38 countries from various regions with the highest ranks on the Fragile States Index (FSI). Their results indicated that factors such as the proportion of refugees, youth unemployment, and countries with unstable economies all positively influence terrorism. Additionally, foreign direct investment (FDI) was identified as having an adverse effect on terrorism. Pham and Nguyen (2024) investigated the impact of terrorist incidents in one country on a neighbouring country. According to the results of panel data analysis with more than 160 countries, trade with countries neighbouring the country with terrorist incidents decreased, and this effect was mainly in the Sub-Saharan Africa region. Meierrieks and Schneider (2021) examined how terrorism affected global economic policy for 170 nations. They indicate that governments react to terrorism by adopting more restrictive foreign economic policies to prevent the formation and funding of terrorism, restrict capital flight, stabilize state finances, and demonstrate political determination. More importantly, they argue that the impact of terrorism is especially critical for smaller nations. Some scholars studied the GDP-terror relationship with different econometric methodologies. Enders, Hoover, and Sandler (2016), in their study, established a Lorenz curve for terrorism to demonstrate the higher concentration of both domestic and international terrorist strikes in middle-income nations, pointing to a nonlinear link between income and terrorism. Their study is significant in the sense that they assume the relationship to be nonlinear. They also employed diverse data from The International Terrorism: Attributes of Terrorist Events (ITERATE) and the Global Terrorism Database (GTD). They also shed light on poverty and terrorism literature and answered questions about whether poverty stimulates terrorism. Several regional studies have revealed the link between economic indicators and terrorism. Sezgin (2019) devised a novel index for Turkey spanning the years 1970–2016, showing the growth in trade volume relative to the growth of terror incidents. Despite a lack of empirical evidence, the study challenges the prevailing notion that terrorist attacks negatively impact international trade. Notably, Sezgin (2019) investigation failed to identify any substantial evidence supporting the negative influence of terrorism on Turkey’s trade. Ruiz Estrada, Park, and Khan (2018) investigated how Turkey’s economic performance was impacted by terrorism. Employing a terrorist attack vulnerability evaluation model covering terror incidents, GDP distribution, and income Trade and Terror: Turkey’s Export Dynamics 97 distribution, they argued that economic decline escalated between 1990 and 2016. Additionally, Bilgel and Karahasan (2019) focused on the PKK terror group’sinfluence on the Turkish economy, utilizing the synthetic control method for the period 1955–2008. Their findings suggested that Turkey’s GDP per capita could have been 13.8 % higher if not for the exposure to PKK terrorism. While existing studies have extensively explored the connections between terrorism and broader economic indicators, such as GDP, imports, exports, and foreign direct investment, there is a gap in the literature concerning the nuanced effect of terrorism on intra-industry trade. Limited scholarly attention has been given to understanding how terrorism influences the intra-industry trade patterns of nations. This study aims to address the significant gap in the understanding of the link between terror and intra-industry trade by exploring previously undiscovered areas. This study is the first to focus on this issue, offering a distinctive viewpoint on the financial effects of terrorism. By analysing the data through the lens of the proxy relative endowment (RELEN) calculated from ICT service exports per capita and introducing novel variables such as GDP similarity and manufacturing output similarity, the aim is to enrich the understanding of the interaction between terrorism and the broader economic structure, albeitat a broader level thanspecific industries. This approach contributes to the existing trade-terror relationship literature and sheds light on the potential implications for policymaking. This study seeks to broaden the discourse on terrorism’s economic impact, particularly in intra-industry trade dynamics, offering fresh insights extending beyond the current scope of scholarly investigation. Helpman and Krugman (1987) proposed using differences in GDP per capita among countries and economic similarity as proxies for analysing intra-industry trade. Subsequent studies have adopted a similar approach; employing these variablesasproxiestounderstandintra-industry dynamics. Some examples include (Antonucci and Manzocchi 2006; Karacan and Korkmaz 2022; Ülengin et al. 2015). The literature review provides information on complex relationships between terrorism, economic indicators, and trade dynamics. While existing studies have made significant additions in understanding the macroeconomic impacts of terrorism, a notable gap remains in exploring the trade in the same industry specifically. This work aims to close this gap by focusing on Turkey’sexportsto71 countries, employing one-way gravity model and contributing novel insights into the interplay between terrorism and trade, particularly at the intra-industry level. 98 E. Z. Ünlü Cities with the highest number of incidents include Hakkari, Şırnak, Diyarbakır, Mardin, Van, and Tunceli. Notably, major urban centres like Istanbul and Ankara witnessed 59 and 22 incidents, respectively. Conversely, the Mediterranean and Aegean coastal areas generally reported fewer occurrences. This spatial pattern aligns with the presence of established terrorist organizations near the Iraq-Syria border. As noted by the Republic of Türkiye Ministry of Foreign Affairs (2022a), the PKK terrorist group, maintains training camps in the Qandil Mountains, enabling them to infiltrate Turkey and launch attacks (Republic of Türkiye Ministry of Foreign Affairs 2022b). We see a decline in PKK-related incidents following counter-terrorism operations by the Turkish Armed Forces. The number of incidents attributed to the PKK reportedly decreased from 147 in 2016 to 10 in 2020. 3 This figure presents the spatial distribution of terrorism-related fatalities across Turkey from 2012 to 2020. The map visualizes cumulative fatalities in various regions, highlighting cities with varying levels of impact. The Eastern Anatolia Region experienced the most significant concentration of fatalities, with cities like Hakkari, Şırnak, Diyarbakır, Mardin, and Gaziantep reporting particularly high numbers. Additionally, Istanbul and Ankara, major urban centres, recorded a total of 191 and 248 fatalities, respectively. These incidents targeted not only security personnel but also civilians, underscoring the indiscriminate nature of terrorist attacks. A particularly devastating event was the Ankara train station bombings of October 2015, which claimed the lives of 109 civilians. This attack remains the deadliest terrorist incident in Turkey’s history. In 2016, Fethullah Gülen Terrorist Organization (FETO) had a coup attempt. Two thousand people were injured, and 246 people died during the coup attempt. The Turkish Parliament and the Presidential Palace were two of the several government structures that were bombarded. This coup attempt was thwarted, and the Turkish people upheld democracy and the rule of law: the Prime Minister, the Government, the Members of the Grand National Assembly, and the President, as stated by the Republic of Türkiye Ministry of Foreign Affairs (2022b). After conducting graphical analysis to explore the relationship between terror incidents and fatalities, attention turns to the descriptive statistics presented in Table 2. This table provides a summary of the data used to explore the relationship between trade and terrorism. Table 2 provides a summary of terrorism incidents, associated fatalities and other variables among Turkey’s trading partners. The data reveals a wide range of 3For chronological calculation the following data are used: “START (National Consortium for the Study of Terrorism and Responses to Terrorism). (2022). Global Terrorism Database, 1970–2020 [data file]. https://www.start.umd.edu/gtd”. Trade and Terror: Turkey’s Export Dynamics 105 incidents, from zero to 2,974 (in Iraq, 2015). Turkey itself experienced incidents ranging from 10 to 247. Fatalities also varied significantly, with Iraq reporting a maximum of 4,807 (in 2015) and Turkey’s highest being 591. Regarding exports, Turkey’s largest trading partners included Germany (US$20.90 billion), the United States, the United Kingdom, Iraq, and Italy. In contrast, Bahrain, Mauritania, and Venezuela received relatively low exports from Turkey. Following the presentation of Table 2, which shows the descriptive statistics of the data used in this study, Table 3 presents the estimation results. Six distinct models were conducted to analyse the relationship between terrorism, Turkey’s exports, and intra-industry trade. The PPML estimator was employed in both models, and the correctness of the model specifications was assessed using the Reset Test. The estimation results for the six econometric models are presented in Table 3. Models 1 and 2 correspond to Equations (6) and (7), respectively, focusing on the impact of terrorism on Turkey’s exports using the “Global Terror Index”(GTI) as the primary variable. Models 3 and 4, derived from Equations (8) and (9), incorporate “terror incidents”(INC) as a key measure. Finally,Models 5 and 6, basedon Equations (10) and (11), introduce “fatalities”(FTL) as an additional explanatory variable, with Model 6 assessing the combined effects of GTI and FTL. The results underscore the significant influence of terrorism on Turkey’s export performance, whether the threat is domestic or originates from the trading partner. Models 1, 2, and 6 highlight the Global Terror Index (GTI) as a critical explanatory factor. In these models, the GTI of partner countries exerts a positive and significant Table :Descriptive statistics. Variable N Mean Std Dev Min Max Export ijt  ,, ,, ,. ,, Global Terror Index (GTI) of partner  . .   Global Terror Index (GTI) of Turkey  . . . . Incidents (INC) in partner  . .  , Incidents (INC) in Turkey  . .   Fatalities (FTL) in partner  . .  , Fatalities (FTL) in Turkey  . .   Sum of GDPs  ,,a,,a,a,,a GDP similarity  . . . . Manufacturing similarity  . . . . Relative endowment (ICT per capita)  . . . . Population of trading partner  ,, ,, ,, ,,, Population of Turkey  ,, ,, ,, ,, Distance ij  ,. ,. . ,. aRepresented in thousands. 106 E. Z. Ünlü Table :Gravity model estimation results using PPML estimator. Variables Model () Model () Model () Model () Model () Model () Global Terror Index (GTI) of partner .c (.) .c (.) .b (.) Global Terror Index (GTI) of Turkey −.c (−.) −.c (−.) −. (−.) Incidents (INC) in partner . (.) .a (.) Incidents (INC) in Turkey −.b (−.) −.b (−.) Fatalities (FTL) in partner −. (−.) . (.) Fatalities (FTL) in Turkey −.b (−.) . (.) Sum of GDPs .c (.) .c (.) .c (.) .c (.) .c (.) .c (.) GDP similarity .c (.) .c (.) .c (.) Manufacturing similarity .c (.) .c (.) .c (.) Relative endowment (ICT per capita) .c (.) . (.) .c (.) .b (.) .c (.) .b (.) Population of trading partner −.c (−.) −. (−.) −.c (−.) −.c (−.) −.c (−.) −.c (−.) Population of Turkey . (.) . (.) .b (.) .c (.) .a (.) . (.) Distance ij −.c (−.) −.c (−.) −.c (−.) −.b (−.) −.c (−.) −.c (−.) Border ij .b (.) .b (.) −.c (−.) −.c (−.) −.c (−.) −.c (−.) Colony ij .c (.) . (.) .b (.) . (.) .c (.) .c (.) Customs union .c (.) .c (.) .c (.) .c (.) .c (.) .c (.) Border*Incidents .c (.) .c (.) Border*Fatalities .c (.) GlobalTerrorIndex*Fatalities −.a (−.) Border*GlobalTerrorIndex .c (.) Constant −. (−.) −. (−.) −. (−.) −.c (−.) −. (−.) −.a (−.) Trade and Terror: Turkey’s Export Dynamics 107 effect on Turkey’s exports. For example, the coefficients in Model 1 (0.265) and Model 2 (0.352) are significant at the 1 % level, suggesting that heightened terrorism risks in partner countries are associated with increased trade volumes for Turkey. In contrast, Turkey’s own GTI consistently shows a negative impact on exports, with coefficients of −0.863 (Model 1) and −1.097 (Model 2), both significant at the 1 % level, indicating that elevated terrorism risks within Turkey reduce its export performance. In Model 6, the interaction between GTI and border proximity exhibits a significant positive effect (coefficient of 3.540), suggesting that geographical proximity to high-risk areas can paradoxically enhance Turkey’s exports. Models 3 and 4 examine the influence of terrorism incidents (INC) on Turkey’s exports. The findings reveal that an increase in the number of incidents in partner countries has a modestly positive impact on exports, with a coefficient of 0.0349 in Model 4, significant at the 10 % level. Conversely, terrorism incidents within Turkey have a consistently negative effect on exports, with coefficients of −0.040 (Model 3) and −0.047 (Model 4), both significant at the 5 % level. These results indicate that while rising terrorism incidents in partner countries may marginally boost trade, increased incidents within Turkey lead to reduced export performance. Furthermore, the interaction between border proximity and incidents is significant and positive in both models, with coefficients of 0.501 (Model 3) and 0.598 (Model 4), both significant at the 1 % level, suggesting that incidents near borders may encourage cross-border trade, possibly due to security-related commerce. Models 5 and 6 focus on the effect of fatalities (FTL) on Turkey’s exports. The analysis shows that fatalities in partner countries do not have a significant impact on Turkey’s exports, as evidenced by the insignificant coefficients in both models. However, fatalities within Turkey exhibit a significant negative impact on exports in Model 5, with a coefficient of −0.041. Interestingly, this relationship becomes insignificant in Model 6. The interaction between border proximity and fatalities in Model 5 is positive and significant (coefficient of 0.329), implying that fatalities near borders may stimulate cross-border trade. Table :(continued) Variables Model () Model () Model () Model () Model () Model () N      Pseudo R. . . . . . Fixed effects Yes Yes Yes Yes Yes Yes Reset test [.][.][.][.][.][.] (), [] denote t-statistics and p-values respectively. (a), (b), (c) indicate significance at  %, % and % respectively. PPMLHDFE command is used for the estimates. 108 E. Z. Ünlü Economic indicators such as the sum of GDPs and GDP similarity emerge as consistent and significant drivers of trade across all models. The sum of GDPs yields coefficients ranging from 0.484 in Model 1 to 1.221 in Model 4, while GDP similarity has coefficients ranging from 1.310 in Model 5 to 1.508 in Model 1, aligning with economic theory that suggests larger and economically similar countries tend to trade more with each other. Manufacturing similarity is also significant in Models 2, 4, and 6, with coefficients between 0.235 and 0.493, indicating that greater similarity in manufacturing structures fosters increased trade. Additionally, the relative endowment of ICT per capita is significant in most models, with coefficients reflecting that countries with better ICT capabilities engage more in trade. The coefficient sign of ICT service export per capita is positive in all six models indicating inter-type of trade. Population dynamics show mixed effects: while the trading partner’s population consistently has a negative impact on exports, with coefficients ranging from −0.046 to −0.281, the significance of Turkey’s population varies across models. Both economic similarities (SIMGDP) and manufacturing industry similarities (SIMMANUF) coefficients are positive, indicating intra-industry trade. This implies that Turkey and its partner countries are engaging in intra-industry trade, corroborating the findings of Karacan and Korkmaz (2022). Geographical distance between trading partners remains a significant constraint on trade across all models, with coefficients ranging from −0.848 to −0.986. Border proximity and colonial ties generally enhance trade, although their effects vary between models. The presence of a customs union consistently shows a significant and positive impact on trade, with coefficients between 0.280 and 0.657, highlighting its role in facilitating cross-border commerce. All models include high-dimensional fixed effects and pass the RESET test for functional form misspecification, indicating that the models are well-specified. The pseudo-R-squared values range from 0.7282 to 0.9073, suggesting a good fit for the trade data. The empirical findings of this study align with existing literature, which consistently highlights the negative impact of terrorism on trade. Doucouliagos (2017) reported that an additional terrorist attack in a bordering country could reduce bilateral trade by approximately 0.013 %, leading to a loss of roughly $6.4 million USD in total trade value. Additionally, fatalities resulting from terrorist events have a diminishing effect on bilateral trade. Similarly, Shah, Hasnat, and Sarath (2020) found that a 1 % increase in terror incidents led to a 0.091 % reduction in Pakistan’s exports. Comparable results were obtained by Egger and Gassebner (2015), who demonstrated that terrorist activities, both historical and ongoing, in exporting and importing countries typically result in a decline in trade. However, the extent of this impact varies depending on the timing of the attacks and the characteristics of the countries involved. Furthermore, their analysis underscores the Trade and Terror: Turkey’s Export Dynamics 109 potential for time aggregation bias, which could affect the precision of the findings. Bardwell and Iqbal (2021) further noted that 74 % of terror-related fatalities occurred in Syria, Iraq, Afghanistan, Nigeria, and Pakistan, leading to a 22 % decrease in Afghanistan’s GDP in 2018 and a 27 % decrease in Iraq’s GDP in 2014. While Turkey is a developing nation, it faces challenges related to terrorism, a pervasive issue in many regions (Luca 2021). suggests that developed economies are often better equipped to mitigate the economic consequences of terrorism due to their stronger security capabilities and faster response mechanisms. Conversely, underdeveloped nations like Syria, Iraq, and Afghanistan have borne a disproportionate economic burden, with terrorism accounting for over 50 % of GDP in some cases. The findings of this study are consistent with those reported by (Çetin, Keser, and Ay 2019), who observed that terrorism exerts a negative influence on key macroeconomic indicators, including GDP, exports, and foreign investment, particularly in the Middle East. Like the estimation results in this study, they found that the economic impact of terrorism is especially pronounced in Turkey, where it significantly increases public expenditure. Their use of spatial lag estimation further supports the idea that the effects of terrorism are not confined to the countries directly experiencing the attacks but have broader regional implications due to international interactions. This aligns with the findings that highlight the substantial impact of terrorism on Turkey’s exports, particularly when considering border proximity and regional instability. In line with the broader literature on the impact of violence, including terrorism and conflict, the findings of this study echo the conclusions drawn by Dinçer and Yüksel (2019), who emphasize that conflict risks lead to significant economic and social disruptions in affected countries. Like how terrorism negatively impacts GDP, exports, and public expenditure, conflicts also result in lower economic growth and heightened unemployment, particularly in the Middle East. The overlap between the economic consequences of terrorism and conflict is evident in the destabilizing effects observed in countries like Syria, Iraq, and Turkey. Furthermore, the social and political repercussions of violence, such as increased corruption and reduced government effectiveness, align with the observed negative impacts of terrorism on macroeconomic stability. 6 Robustness Checks 6.1 Estimation with Year and Importer Fixed Effects To test the robustness of the findings presented in Table 3, I re-estimate the models using the PPML estimator while incorporating year and importer fixed effects. The 110 E. Z. Ünlü results of these estimations are displayed in Table 4. This approach allows us to control for any unobserved time-variant or country-specific factors that may influence Turkey’s exports, providing a more comprehensive understanding of the relationships between terrorism, economic indicators, and trade performance. Models 1 and 2 in Table 4 include the Global Terror Index (GTI) as the primary explanatory variable, while Models 3 and 4 focus on terror incidents (INC), and Models 5 and 6 incorporate fatalities (FTL) as key measures. These results are compared with the baseline models (without fixed effects) to assess the consistency and robustness of the estimates. Unlike the previous estimations, where the GTI of partner countries had a positive and significant effect on Turkey’s exports, in the robustness models, the effect is either insignificant or marginally significant. For instance, in Models 1 and 2, Table :Gravity model estimation results using PPML estimator with “year”and “importer fixed effects”. Variables Model () Model () Model () Model () Model () Model () Global Terror Index (GTI) of partner −. (−.) −. (−.) .a (.) Incidents (INC) in partner .b (.) .a (.) Fatalities (FTL) in partner −. (−.) −. (−.) Sum of GDPs .c (.) .c (.) .c (.) .c (.) .c (.) .c (.) GDP similarity .b (.) .b (.) .b (.) Manufacturing similarity .b (.) .b (.) .a (.) Relative endowment (ICT per capita) . (.) . (.) . (.) .c (.) .a (.) . (.) Population of trading partner −.c (−.) −.c (−.) −.b (−.) . (.). (.)−. (−.) Customs union −. (−.) −.a (−.) −.a (−.) −.c (−.) −.c (−.) −. (−.) Constant .c (.) .c (.) . (.) −.c (−.) −.c (−.) −. (−.) N      Pseudo R. . . . . . Fixed effects Importer year Importer year Importer year Importer Importer Importer year Reset test [.][.][.][.][.][.] (), [] denote z-statistics and p-values respectively. (a), (b), (c) indicate significance at  %, % and % respectively. PPMLHDFE command is used for the estimates. Trade and Terror: Turkey’s Export Dynamics 111 the coefficients are (−0.050) and (−0.054), respectively, with both being statistically insignificant. However, Model 6 shows a positive and marginally significant effect (0.057, significant at the 10 % level). This contrasts with the results in Table 3, where the GTI of partner countries consistently exhibited a positive and highly significant effect. The results for terror incidents (INC) in partner countries in Models 3 and 4 remain consistent with those in Table 3. The coefficient for incidents is positive and significant in both models, with the effect being stronger and more statistically significant in Model 3 (0.025, significant at the 5 % level). This aligns with the previous finding that increasing terrorism incidents in partner countries can paradoxically boost Turkey’s exports. Fatalities (FTL) in partner countries continue to show no significant impact on Turkey’s exports across Models 5 and 6. This finding remains consistent with the estimation results presented in Table 3, indicating that fatalities in trading partners do not exert a direct influence on trade performance. Economic factors such as the sum of GDPs and GDP similarity remain highly significant across all models, consistent with the previous estimation. The sum of GDPs continues to exhibit strong positive effects on Turkey’s exports, with coefficients ranging from 0.603 (Model 1) to 1.624 (Model 5), all significant at the 1 % level. GDP similarity also remains a positive and significant driver of trade, though the coefficients are slightly lower than in the previous models, ranging from 0.830 (Model 3) to 0.952 (Model 1). Like the previous estimated models, manufacturing similarity continues to exert a positive and significant influence on exports. The coefficients are smaller in magnitude, ranging from 0.215 to 0.277, but remain significant in Models 2, 4, and 6, indicating the importance of intra-industry trade. The population of trading partners exhibits a strong negative effect in Models 1. However, in Models 4, 5, and 6, the significance of the population variable diminishes, with coefficients becoming either insignificant or changing signs. This divergence could be attributed to the inclusion of fixed effects, which control for unobserved country-specific characteristics that may be correlated with population. The customs union variable, which was previously positive and significant in Table 3, now exhibits negative or insignificant coefficients across most models in Table 4. The inclusion of fixed effects appears to alter the impact of customs unions on Turkey’s exports, suggesting that the positive association observed earlier may be driven by unobserved factors captured by the importer or year effects. The RESET test results across all models indicate that the models are correctly specified in terms of functional form. Only importer fixed effects are included in Model 4 and Model 5 because including year effects resulted in a specification error, as identified by the RESET test. This ensures that the model remains well-specified 112 E. Z. Ünlü and avoids overfitting by focusing on the significant importer-specific variations without introducing unnecessary year effects. When comparing the results from Table 4 (with year and importer fixed effects) to those in Table 3 (without fixed effects), several differences emerge: –The positive impact of the Global Terror Index of partner countries observed in Table 3 becomes weaker or insignificant in Table 4, suggesting that controlling for unobserved importer and year-specific factors alters the relationship between terrorism and trade. –The effects of terror incidents in partner countries remain consistent across both tables, reinforcing the finding that an increase in incidents may boost exports. –Economic variables such as the sum of GDPs and GDP similarity remain robust and significant across both models, highlighting their strong influence on trade performance. –The influence of population and customs unions varies between the two sets of results, with the fixed-effects models providing more nuanced insights into these relationships. Including year and importer fixed effects provides a more refined understanding of the factors affecting Turkey’s exports, particularly in terms of the role of terrorism and economic indicators. While some results remain consistent with the no-effects models, others, such as the impact of the Global Terror Index and customs unions, suggest that unobserved factors play a critical role in shaping these relationships. 6.2 Random Coefficients Model The application of random coefficients models is particularly suited for capturing the inherent complexity and heterogeneity in gravity models. As noted by Baltagi, Egger, and Pfaffermayr (2014), random effects models effectively handle the double or even triple indexing of country pairs, addressing the unobserved heterogeneity that arises in bilateral trade relationships. In gravity models, unobserved factors such as cultural affinity, historical ties, or participation in trade agreements often vary across countries and are not directly measurable. By incorporating random effects, the model introduces flexibility, allowing these unobserved factors to be treated as random, rather than fixed, across country pairs. This enables the model to account for variations in the way different countries interact in trade, without the restrictive assumptions imposed by fixed-effects models. Mathematically, consider a standard gravity model of trade where bilateral trade flows Y ij between country i and country j are influenced by observed factors X ij Trade and Terror: Turkey’s Export Dynamics 113 (such as GDP, distance, and tariffs). A basic fixed-effects model assumes that the unobserved heterogeneity across country pairs is constant: Yij =Xijβ+αi+αj+εij (12) where α i and α j represent country-specificfixed effects. However, this specification may not capture the full complexity of trade relationships, especially when the unobserved factors are not constant across trading partners. By contrast, a random coefficients model assumes that the impact of some explanatory variables varies across countries: Yij =Xij(β+uij)+αi+αj+εij (13) where u i represents the random deviation from the average effect of X ij , allowing for heterogeneity in the coefficients across country pairs. This random variation captures the idea that, for instance, the effect of distance or tariffs on trade may differ depending on the specific trade relationship. Egger and Prusa (2014) further justify the use of random coefficients models in the presence of measurement error or heterogeneous responses to variables like tariffs and distance. In trade models, these variables are often subject to varying levelsof accuracy or influence depending on thecountries involved. Forexample, the impact of the Global Terrorism Index (GTI) or GDP on trade may not be uniform across all countries; some trading relationships may be more sensitive to terrorism or economic size than others. A random coefficients model captures this variation by allowing for different sensitivities across trading partners, thus reflecting the nuanced nature of international trade dynamics. To formalize this, consider the random coefficient for a variable X ij in the context of international trade. The model can be represented as: Yij =Xijβ+Zij(γ+uij)+εij (14) here, Z ij represents the set of variables for which the coefficients γare allowed to vary across countries, with u ij ∼N(0,σ2) representing the random coefficients. This specification introduces flexibility, as it accounts for the country-specific or pairspecific deviations from the average effect, providing a more accurate representation of the heterogeneous nature of trade flows. In conclusion, the adoption of a random coefficients model allows for a more realistic and flexible approach to modelling international trade, particularly when dealing with complex phenomena such as terrorism’s impact on trade. Standard fixed-effects models impose uniformity on the relationships between variables, which may be overly restrictive in a global context where economic, political, and social conditions vary widely. By allowing key variables such as the GTI and GDP to have different effects across countries, random coefficients models offer a more 114 E. Z. Ünlü Dinçer, Hasan, and Serhat Yüksel. 2019. “Analyzing the Possibility of Violent Conflict in the Middle East Economies Using Determinants of Global Conflict Risk Index with an Integrated Fuzzy Multicriteria Decision Making Model.”In The Impact of Global Terrorism on Economic and Political Development, edited by Ramesh Chandra Das, 155–66. Leeds: Emerald Publishing Limited. Doucouliagos, Hristos. 2017. “An Injury to One Is an Injury to All: Terrorism’s Spillover Effects on Bilateral Trade.”SSRN Scholarly Paper. Rochester, NY. https://doi.org/10.2139/ssrn.2998968. Egger, Peter, and Martin Gassebner. 2015. “International Terrorism as a Trade Impediment?”Oxford Economic Papers 67 (1): 42–62. Egger, Peter, and Jan Prusa. 2014. “The Determinants of Trade Costs: A Random Coefficients Approach.” CESifo Working Paper Series, CESifo Working Paper Series. https://ideas.repec.org//p/ces/ceswps/_ 4859.html. Enders, Walter, Gary A. Hoover, and Todd Sandler. 2016. “The Changing Nonlinear Relationship between Income and Terrorism.”Journal of Conflict Resolution 60 (2): 195–225. Enders, Walter, and Todd Sandler. 2000. “Is Transnational Terrorism Becoming More Threatening? A Time-Series Investigation.”Journal of Conflict Resolution 44 (3): 307–32. Gaibulloev, Khusrav, and Todd Sandler. 2009. “The Impact of Terrorism and Conflicts on Growth in Asia.” Economics & Politics 21 (3): 359–83. Gaibulloev, Khusrav, and Todd Sandler. 2019. “What We Have Learned about Terrorism since 9/11.”Journal of Economic Literature 57 (2): 275–328. Gok, Ibrahim Yasar, Yavuz Demirdogen, and Sefa Topuz. 2020. “The Impacts of Terrorism on Turkish Equity Market: An Investigation Using Intraday Data.”Physica A: Statistical Mechanics and Its Applications 540 (February): 123484. Haq, Zahoor Ul, Zia Ullah, and Javed Iqbal. 2018. “Terrorist Incidents and Trade.”Global Social Sciences Review 3 (2): 55–70. Helpman, Elhanan, and Paul Krugman. 1987. Market Structure and Foreign Trade: Increasing Returns, Imperfect Competition, and the International Economy. Cambridge, Massachusetts: MIT Press. Ibrahim, Bello, and AriAri. 2021. “The Impact of Terrorism on Foreign Direct Investment: The Case of Turkey.”Üçüncü Sektör Sosyal Ekonomi 56 (3): 1781–97. Ilarslan, Kenan, and Munevvere Yildiz. 2022. “The Effects of Terrorism and Economic Indicators on Bank Loans to the Private Sector: Evidence from Developing Countries.”Emerging Markets Finance and Trade 58 (2): 329–41. IMF. 2023. “IMF Direction of Trade Statistics.”Direction of Trade Statistics. https://data.imf.org/? sk=9D6028D4-F14A-464C-A2F2-59B2CD424B85. Institute for Economics & Peace (IEP). 2023. Global Terrorism Index. https://www.visionofhumanity.org/ public-release-data/. Julius, Stacy, Nnanna P. Azu, and Maimuna Y. Muhammad. 2019. “Assessing the Impact of Terrorism in Trade Development in the SADC Region: A Gravity Model Approach.”Asian Economic and Financial Review 9 (10): 1147–59. Karacan, Semih, and Özge Korkmaz. 2022. “Turkish Exports before and after the 2001 Financial Crisis: A Panel Gravity Model.”Foreign Trade Review 57 (1): 27–40. Lanouar, Charfeddine, and Umer Shahzad. 2021. “Terrorism and Capital Flows: The Missed Impact of Terrorism in Big Cities.”Applied Economics Letters 28 (18): 1626–33. Leitão, Nuno Carlos, Horácio C. Faustino, and Yushi Yoshida. 2010. “Fragmentation, Vertical Intra-Industry Trade, and Automobile Components.”Economics Bulletin 30: 1006–15. Linder, S. B. 1961. An Essay on Trade and Transformation. New York: John Wiley & Sons. Luca, Dumitru-Mihai. 2021. “The Global Economic Impact of Terrorism.”European Journal of Interdisciplinary Studies Articles 7: 19–29. Trade and Terror: Turkey’s Export Dynamics 121 Meierrieks, Daniel, and Friedrich Schneider. 2021. “Terrorism and International Economic Policy.” European Journal of Political Economy 69 (September): 102011. Okafor, Godwin, and Jenifer Piesse. 2018. “Empirical Investigation into the Determinants of Terrorism: Evidence from Fragile States.”Defence and Peace Economics 29 (6): 697–711. Pham, Cong S., and Chris Doucouliagos. 2017. An Injury to One Is an Injury to All: Terrorism’s Spillover Effects on Bilateral Trade. https://www.iza.org/publications/dp/10859/an-injury-to-one-is-an-injuryto-all-terrorisms-spillover-effects-on-bilateral-trade. Pham, Cong S., and Hoa Nguyen. 2024. “Impact of Terror on International Trade in Financial Services: Does the Development of Financial Institutions Matter?”The World Economy n/a (n/a): 1–39. Polyxeni, Kechagia, and Metaxas Theodore. 2019. “An Empirical Investigation of FDI Inflows in Developing Economies: Terrorism as a Determinant Factor.”The Journal of Economic Asymmetries 20 (November): e00125. Republic of Türkiye Ministry of Foreign Affairs. 2022a. An Informative Note of the Ministry of Foreign Affairs on the Failed Coup Attempt/Republic of Türkiye Ministry of Foreign Affairs. https://www.mfa. gov.tr/an-informative-note-of-the-ministry-of-foreign-affairs-on-the-failed-coup-attempt.en.mfa. Republic of Türkiye Ministry of Foreign Affairs. 2022b. https://www.mfa.gov.tr/pkk.en.mfa. Ruiz Estrada, Mario Arturo, Donghyun Park, and Alam Khan. 2018. “The Impact of Terrorism on Economic Performance: The Case of Turkey.”Economic Analysis and Policy 60 (December): 78–88. Sandler, Todd, and Walter Enders. 2008. Economic Consequences of Terrorism in Developed and Developing Countries: An Overview, edited by Philip Keefer, and Norman Loayza, 1st ed., 17–47. Cambridge: Cambridge University Press. Sezgin, Şennur. 2019. “International Trade and Terror Relationship in Turkey.”Sosyoekonomi 27 (40): 55–66. Shah, Syed Hasanat, Hafsa Hasnat, and Delpachitra Sarath. 2020. “The Impact of Foreign Direct Investment on Trade in Pakistan: The Moderating Role of Terrorism.”Journal of Economic Studies 47 (5): 1137–54. Silva,J.M. C. Santos, and Silvana Tenreyro. 2006. “The Log of Gravity.”The ReviewofEconomics and Statistics 88 (4): 641–58. The World Bank. 2023. WDI. https://databank.worldbank.org/source/world-development-indicators. Tinbergen, Jan. 1962. Shaping the World Economy; Suggestions for an International Economic Policy. New York: Twentieth Century Fund. Ülengin, Füsun, Bora Çekyay, Peral ToktaşPalut, Burç Ülengin, Özgür Kabak, Özay Özaydın, and Şule Önsel Ekici. 2015. “Effects of Quotas on Turkish Foreign Trade: A Gravity Model.”Transport Policy 38 (C): 1–7. Wakasugi, Ryuhei. 2007. “Vertical Intra-Industry Trade and Economic Integration in East Asia*.”Asian Economic Papers 6 (1): 26–39. Yoshida, Yushi, Nuno Carlos Leitão, and Horácio C. Faustino. 2009. “Vertical Intra-Industry Trade and Foreign Direct Investment between Japan and European Countries.”Atlantic Economic Journal 37: 351–65. 122 E. Z. Ünlü