Exporting Under Political Risk: Payment Term Selection in Global Trade

Abstract

This paper investigates the extent to which political risk affects exporter-financed trade transactions. Using industry-level trade finance data from Turkey, we show that export transactions executed under open-account terms decrease with the political risk in the export markets. Further, we also document that the effect of political risk on trade finance is disproportionately higher for industries that export complex products. This paper contributes to the literature by examining how political risk interacts with product complexity to influence exporters’ trade finance decisions—an effect that has been previously overlooked in empirical studies.

1. Introduction

Risk is an inherent part of running a business. The level of risk amplifies, and challenges increase multifold when transactions occur across borders. There is an added element of uncertainty in the international business environment that makes it more dynamic when compared with domestic trading environment. Risk across borders can manifest in various forms that range from market fluctuations to import restrictions to even insolvencies and defaults of different degrees. In addition to paying ongoing trade costs, and facing cultural and institutional differences, exporters must be ready for different forms of risks while they penetrate foreign markets.

One of the dimensions of global business risks is political risk, which arises from governments’ policies and actions that can directly affect business operations and firms’ ability to perform their economic obligations. Wars, coups, political violence, regime changes due to uprisings, currency transfer and conversion restrictions, expropriation, and regulatory changes can all wreak havoc on exporters and investors in foreign markets. Given the rising political instability worldwide, companies remain vigilant to political risks and seek ways to mitigate them. According to recent Multilateral Investment Guarantee Agency (MIGA) reports, USD 30.2 billion worth of political risk and credit enhancement guarantees have been issued since 2020, representing a 75% increase compared to the previous five-year period. Meanwhile, total trade credit insurance premiums recorded by International Credit Insurance and Surety Association (ICISA) members reached EUR 42.1 billion between 2016 and 2021.

The importance of political risk on globalization has become increasingly recognized by international economics and business researchers. They have mainly addressed the hidden barriers and threats political risk poses and documented its deteriorating impact on trade and foreign direct investment.1 In this study, we seek to complement and extend this research by analyzing the impact of political risk on trade finance using disaggregated industry-level trade finance data from Turkey.

Trade finance is broadly defined as the methods and instruments that support exporting and importing firms throughout the trade cycle (Menichini, 2009). Banks, firms, and credit agencies provide trade finance, mainly short-term credit, to enhance global trade. Auboin (2007) notes that more than 90% of international trade is underpinned by some form of trade credit. According to survey reports, one of the main reasons why world trade experienced a slowdown in the aftermath of the financial crisis in 2009 was expensive trade credit in the global recession period. That is why G20 countries agreed to provide USD 250 billion in trade finance for two years in 2009 (Chauffour & Farole, 2009). Trade finance facilitates cross-border transactions by providing capital and liquidity to companies, mitigates country- and company-specific risks, and provides a set of payment instruments that help exporters to receive their payments in a timely manner and importers to securely receive their orders.

Trade finance serves a handful of functions, but this study’s primary focus is the payment aspect of trade finance, with emphasis on its relationship with political risk. In any form of trade, it is crucial for the seller to get paid in full and on time and for the buyer to receive the goods as specified in the contract. Given the degree of incomplete information between buyers and sellers, an appropriate method of payment must be chosen in international trade to minimize the default and non-delivery risks. An exporter can execute a cross-border transaction in two main ways: through cash in advance, where the payment takes place before the goods are shipped, and on an open account (extending trade credit), where the payment is received some time after the delivery. In international business, banks also act as intermediaries to reduce the risk of the transactions and provide other financing options such as a letter of credit and documentary collections. Niepmann and Schmidt-Eisenlohr (2017) show that letters of credit constitute about 13 percent and documentary collections about 2 percent of global trade.

Schmidt-Eisenlohr (2013) developed a theoretical model to analyze the payment choice in international transactions. In his model, two risk-neutral importing and exporting firms play a one-shot game where the exporter makes a take-it-or-leave-it offers to the importer. This proposal specifies the price, quantity, and the payment method in the exchange. In this setting, there are two problems arising: a financing problem and a commitment problem. Under the open-account option, the exporter finances the transaction using their country’s financial resources and enforcement takes place in the importer’s country. Trade finance extended by the firm in the country with lower financing costs and weaker enforcement maximizes the exporter’s profit. Consequently, open-account export transactions are more likely to happen when the enforcement is strong and the financing cost is high in the importer’s country.

Political risk affects the payment choice both via financing and commitment channels, as described in Schmidt-Eisenlohr (2013). First, political risk in the importer’s country increases the risk premium in financial markets and drives up the cost of financing, which makes it less likely for the buyer to offer cash in advance. Second, sellers will perceive higher risks in terms of following country-specific procedures and claiming their rights and thus be reluctant to extend trade credit when there is political instability and repression in export destinations. Accordingly, we hypothesize that export transactions executed under open-account terms will increase if the political risk in the export markets decreases. Using industry-level trade finance data from Turkey and applying different econometric specifications, we provide strong support for this hypothesis in this paper. While our framework builds upon the model proposed by Schmidt-Eisenlohr (2013), we contribute to the literature in two important ways. First, we introduce and empirically test interaction effects between political risk and product complexity, inspired by Nunn (2007), to uncover heterogeneity in how risk conditions affect payment method decisions across industries. Second, we utilize detailed, industry-level bilateral data from a single exporter (Turkey), which enables a higher-resolution analysis than multicountry aggregate studies (e.g., Hoefele et al., 2016). This interaction-based approach has not been explored in prior empirical work on trade finance and reveals that firms exporting more complex goods are more sensitive to political risk when extending post-shipment credit—highlighting a conditional risk channel not previously isolated. Additionally, to further uncover heterogeneity in political risk sensitivity, we interact political risk indicators with product complexity. Results (Appendix A 

Table A3. Interaction effects of political risk with product complexity.

Political Risk Indicator	Main Effect (β)	Interaction with Complexity (β)	Significance
Law and Order	0.009	0.012	***
Democratic Accountability	0.011	0.006	**
Bureaucratic Quality	0.013	0.014	***
Military in Politics	0.013	0.016	***
External Conflict	0.018	0.01	**

Notes: (1) Dependent variable: open-account usage. All models include fixed effects. Interaction terms show how political risk effects vary with product complexity (source: Nunn, 2007). (2) *** and ** denote significance at 1% and 5% levels, respectively.

) show that complex products are significantly more affected by political uncertainty.

In our formal econometric analysis, we use a wide range of indicators for political risk to identify the relative importance of these indicators for extending trade finance. We investigate the influences of law and order, democratic accountability, government stability, socioeconomic conditions, investment profile, internal and external conflict, military in politics, and the quality of bureaucracy. We also perform a battery of sensitivity tests. Our results are robust to a subsample analysis in which we exclude observations from different regions. We also checked the robustness of our results by excluding the top three exporting industries from our sample. We employed linear regressions as well as probit and ordered logit models. All estimations strongly confirm the effect of political risk on trade finance. Finally, we also analyzed the differential impact of product complexity on the relationship between political risk on trade finance. We argue that industries with complex products will be more watchful for political risks before they extend trade of the degree of customization and relationship-specific investment in their products. This hypothesis is supported in our model when we interact our political risk measures with the complexity measure developed in Nunn (2007).

The paper proceeds as follows. Section 2 provides a summary of related studies and our contribution. In Section 3, we describe the dataset. Section 4 summarizes the identification strategy, results, robustness checks, and extensions. Section 5 concludes.

2. The Related Literature and Contributions

Our paper bridges two strands of the literature. One body of research studies the impact of political risk on international trade and investment. Within this research, scholars have documented the negative impacts of political risk on internationalization. An early example on the trade side is Morrow et al. (1998), who suggest the effect of international politics on trade flows. Anderson and Marcouiller (2002), on the other hand, show that inadequate institutions constrain trade and omission of indices of institutional quality biases the estimates of gravity models of international trade. Long (2008) demonstrates that expectations of domestic or interstate conflict, in addition to violent armed conflicts, are negatively correlated with bilateral trade flows. Berkowitz et al. (2006) observe that the institutional quality of trade partners has a trade-enhancing impact, especially for complex products. Oh and Reuveny (2010) study the interaction of political risk and disasters on trade and find that the negative impact of disasters on trade flows is less severe for countries with low political risk. Moser et al. (2008) find evidence that political risk has a detrimental effect on exports. More recently, Bilgin et al. (2017) examined the exports of Turkey to Islamic countries in a gravity framework and documented the negative impact of political risk and government instability on exports. The negative impact of political risk on investment, FDI, and capital inflows has also been documented in many studies (including but not limited to Lensink et al., 2000; Busse & Hefeker, 2007; Harms & Ursprung, 2002; Harms, 2002; Solomon & Ruiz, 2012).

This study also fits into a much larger body of literature which blends trade and finance. One part of this research explores the effect of financial conditions in a country on its export flows. Financial development has been shown to be a source of comparative advantage for the industries that rely more on external finance (Beck, 2003; Manova, 2008). Similarly, Gur and Avşar (2016) demonstrate that R&D-intensive industries export more in countries where the financing cost is low. Hur et al. (2006) find that countries with better financial institutions have higher export shares and trade balance in sectors with more intangible assets. Using Chinese firm-level data, Manova et al. (2015) provide evidence that credit constraints influence not only trade but also the pattern of multinational activity. Amiti and Weinstein (2011) and Chor and Manova (2012) find that constraints in the availability of trade finance culminated in the collapse of trade in the global recession period. Auboin and Engemann (2014) identify a significantly positive effect of insured trade credit, as a proxy for trade credits, on exports. Van der Veer (2015) also points out the importance of private credit insurance on exports.

Another strand within the literature focuses on the payment aspect of trade finance. On the theoretical side, Schmidt-Eisenlohr (2013) examines the trade-off firms have between different payment terms in cross-border trade and the cross-country differences in their use. He also indirectly tests the predictions of his model using gravity regressions and shows that financial conditions and contract environments, both in the exporting and importing country, affect international trade. Love (2013) notes that a lack of reliable and comprehensive data was the main reason why the literature on the payment aspect of trade finance paled compared to other dimensions. The data were mainly gathered through bank and firm surveys, which often had insufficient coverage and did not provide bilateral information on different payment terms in exports or imports.

More recently, the literature has seen several empirical contributions as more detailed payment data on trade transactions have become available. Hoefele et al. (2016) test the predictions of Schmidt-Eisenlohr’s (2013) model utilizing the World Bank Enterprise Survey and document that international trade transactions are more likely to be paid after delivery when financing costs in the source country are high and when contract enforcement is weak. A similar result was obtained by Antras and Foley (2015), who used transaction trade data from a single firm that sells poultry products. Turkcan and Avsar (2018) show that financing cost in a country is more important in terms of offering post-shipment terms in exports for industries that rely more on external finance. Demir and Javorcik (2018) provide evidence that competitive pressures lead exporters to provide trade credit.

A few empirical works on payment methods placed the microscope on letters of credit. A theoretical model by Glady and Potin (2011) finds that exports to countries with a better financial sector are more likely to occur on a letter of credit. Using Colombian data, Ahn and Sarmiento (2013) show that adverse bank liquidity shocks led to a decline in imports through letters of credit. Olsen (2013) suggests that letters of credit can be used to overcome weak contract enforcement during crises. Niepmann and Schmidt-Eisenlohr (2017) examine letter-of-credit transactions using US banks’ data and show that the use of letters of credit causes increases in default risk and decreases in interest rates.

To sum up, this work contributes to the literature that examines the effect of political risk on international trade and mentions the trade finance channel for the first time. In addition, we also contribute to a growing body of literature on payment choice in international transactions by analyzing the effect of different dimensions of political risk on the decision to extend trade credit. The bilateral information on payment methods in export transactions at the industry level used in this study represents a significant improvement over other studies in the literature.2

3. Data

Data on the method of payments in export transactions were purchased from the Turkish Statistical Institute (TUIK). This database documents the use of different payment terms in export transactions originating from Turkey at the two-digit level of ISIC Revision 3 and covers bilateral exports in the years from 2002 to 2010 from Turkey to 96 destination countries across 24 industries. The panel is unbalanced due to missing payment term observations for certain country–year combinations. These data also report on the export markets, which allows us to examine the effect of variation in political risk on trade finance. This panel structure enables us to exploit both the cross-sectional variation (across countries and industries) and the temporal variation (over time), which is essential for estimating the evolving impact of political risk on trade finance decisions across heterogeneous bilateral relationships. A formal table of descriptive (summary) statistics (Appendix A 

Table A4. Descriptive (summary) statistics.

Variable	Observations	Mean	Std. Dev.	Min	Max
Open-Account Exports	24,886	0.532	0.422	0	1
Law and Order	24,886	3.729	1.325	0.5	6
Democratic Accountability	24,886	4.034	1.735	0	6
Bureaucratic Quality	24,886	2.193	1.116	0	4
Military in Politics	24,886	3.863	1.692	0	6
External Conflict	24,886	9.922	1.459	2.125	12
Investment Profile	24,886	8.819	2.464	1	12
Socioeconomic Conditions	24,886	5.777	2.622	0	11
Internal Conflict	24,886	9.399	1.660	2.917	12
Government Stability	24,886	8.535	1.539	3.167	11.5

Notes: While summary statistics and aggregate trends may resemble pooled data representations, our empirical strategy is rooted in panel data estimation. The estimation framework exploits both the cross-sectional and time-series dimensions of the data, allowing for a more credible identification of the relationship between political risk and trade finance terms. Unlike pooled regressions, panel models control for unobserved heterogeneity across countries and industries and improve the precision of coefficient estimates.

) and a summary of the patterns in open-account usage and product complexity are visualized in Appendix A Figure A1, Figure A2 and Figure A3.

Table 1. Shares of different payment terms in exports (in percentages).

	Open Account	Letter of Credit	Cash in Advance	Cash Against Documents
2002	52.1	17.3	3.1	24.1
2003	54.5	15.9	2.9	23.0
2004	56.5	14.7	3.4	21.9
2005	57.5	15.7	3.9	20.0
2006	59.2	14.9	5.1	18.8
2007	56.9	17.4	5.2	17.9
2008	60.1	17.2	6.2	15.6
2009	60.4	12.1	7.7	15.4
2010	61.2	13.2	7.5	15.9
Overall	57.6	15.3	5.0	19.1

shows the average use of post-shipment, pre-shipment, and letter-of-credit terms in Turkey’s exports over the years of our sample. As shown, the open-account (post-shipment) method has the lion’s share in export transactions. About 58% of exports were executed under open-account terms. This observation gives a hint of the positive impact of political stability, since Turkey’s exports are heavily concentrated in European markets. Payment methods with bank intermediation, letters of credit, and documentary collection constitute around 34% and cash-in-advance terms make around 5% of total exports. We also observe an increasing trend in the share of open-account terms, whereas a decreasing one in the shares of cash against documents and letters of credit. These trends are more visible in the crisis period. We believe that an increase in the letter-of-credit fees associated with the financial meltdown was the main reason for the decrease in bank intermediation in export transactions.

Figure 1 shows the trend of the value of exports for different payment terms. Open-account transactions reached more than USD 60 billion in our sample, and they are by far the most preferred of the payment methods. However, the increase in cash-in-advance transactions is also noteworthy. In fact, the value of cash-in-advance exports increased from USD 500 million to around USD 10 billion from 2002 to 2010. A similar trend was also observed for cash-against-document transactions. The value of exports executed under cash-against-document terms almost doubled over the years of our sample. The drastic increase in pre-shipment terms can primarily be attributed to the shift in exports to Middle Eastern countries. In fact, according to TUIK statistics, the same period witnessed an export increase of around USD 20 billion to this region.3 This finding is also parallel to our hypothesis on the relationship between political risk and trade finance since this period also coincides with anti-government protests, uprisings, and armed rebellions that spread across North Africa and the Middle East regions.

In

Table 2. Shares of different payment terms in exports for regions (in percentages, 2002–2010).

	Open Account	Letter of Credit	Cash in Advance	Cash Against Documents
Europe	63.8	7.5	4.1	14.9
Middle East	34.7	35.2	8.1	19.6
North America	48.2	23.3	5.8	16.1
Asia	32.42	35.6	8.8	16.1
Africa	38.6	33.2	7.6	21.2

, we document the export shares by payment terms for different regions. In support of our earlier discussions, Table 2 shows that companies preferred open-account terms when exporting to European and North American countries, while more pre-shipment terms were agreed in exports to regions with high political risk, such as Africa and the Middle East. In fact, more than 50% of the export transactions to those regions were executed with a form of bank intermediation. Table 2 provides suggestive evidence for our hypothesis that Turkish exporters are more prone to offering post-shipment terms to countries with less conflict and political stability.

Although our theoretical expectation is that lower political risk is associated with greater reliance on open-account terms, Table 2 shows that certain high-risk regions (including Africa and the Middle East) still exhibit substantial use of post-shipment financing. This inconsistency may be driven by contextual trade frictions. Specifically, exporters may operate under long-standing trade relationships or historical path dependencies that foster informal trust, even in environments with weak formal institutions, as shown by Anderson and Marcouiller (2002). Additionally, underdeveloped or restricted banking systems may limit access to letters of credit, effectively forcing exporters to extend credit despite the political risk, as investigated by Demir and Javorcik (2018). We discuss these mechanisms as plausible explanations for the regional heterogeneity observed in our results.

This also suggests that the relationship between political risk and payment method is conditional—not deterministic—and is influenced by institutional voids or non-contractual norms that vary across regions. Furthermore, Appendix A Figure A1 illustrates the time trend in open-account usage, showing a moderate upward pattern over the 2002–2010 period.

In

Table 3. Share of open-account exports by industries.

Industries	Overall	Exports to Countries Above Median Law-and-Order Level
Food products and beverages	0.623	0.639
Tobacco products	0.156	0.167
Textiles	0.660	0.691
Manufacture of wearing apparel; dressing and dyeing of fur	0.655	0.685
Leather products	0.530	0.579
Wood products	0.479	0.420
Paper products	0.546	0.605
Publishing and printing	0.507	0.523
Petroleum products	0.528	0.506
Chemicals	0.647	0.688
Rubber and plastics	0.664	0.717
Non-metallic mineral products	0.620	0.643
Basic metals	0.410	0.441
Manufacture of fabricated metal products	0.565	0.636
Manufacture of machinery and equipment	0.627	0.611
Manufacture of office, accounting, and computing machinery	0.374	0.387
Manufacture of electrical machinery and apparatus	0.643	0.649
Manufacture of radio, television, and communication equipment	0.437	0.464
Manufacture of medical, precision, and optical instruments, including watches	0.506	0.486
Manufacture of motor vehicles, trailers, and semi-trailers	0.586	0.627
Manufacture of other transport equipment	0.327	0.381
Manufacture of furniture	0.615	0.730

, we present the shares of exports executed under post-shipment terms for two-digit ISIC industries. In almost all industry categories, open-account terms dominate the cross-border transactions. We also observe that high-tech industries (such as machinery, electrical machinery, and chemicals) had larger shares in open-account terms compared to low-tech industries (such as paper products, publishing, and wood). The second column in Table 3 shows the same averages for the exports to the countries that are above the median law-and-order rating, one of the measures of political risk in this paper. As shown, in almost all industry classifications, exporters extended more trade credit to countries with better law-and-order ratings. This is another interesting observation from the raw data that supports our hypothesis.

We gathered the information on political risk and from the International Country Risk Guide (ICRG) provided by the Political Risk Services (PRS) Group. Howell (2011) describes the variables and the methodology of the database. We used the following political risk components in our study: (1) law and order (LO), which assesses the strength and impartiality of the legal system; (2) democratic accountability (DA), which is a measure of how responsive the government is to its people; (3) bureaucratic quality (BQ), which rates the institutional strength and quality of the bureaucracy; (4) military in politics (MP), which represents the influence of the military in politics; (5) external conflict (EC), which weighs the risk to the government from foreign action; (6) investment profile (IP), which assesses factors affecting the risk to investment; (7) socioeconomic condition (SC), which is an evaluation of the socioeconomic pressures at work in society that could constrain government action; (8) government stability (GS), which is a measure of the government’s ability to implement its policies and to stay in office; and (9) internal conflict (IC), which is a measure of political violence within country. The political risk data are available monthly, and we use the annual average of the monthly indicators in our models.

Each measure has a different scale, but a higher value indicates less political risk for all of them. As is to be expected, these measures are closely related to each other. For instance, as shown in

Table 4. Correlation matrix of political risk variables.

	Law and Order	Democratic Accountability	Bureaucratic Quality	Military in Politics	External Conflict	Investment Profile	Socioeconomic Conditions	Government Stability	Internal Conflict
LO	1.000
DA	0.3034 ***	1.000
BQ	0.6250 ***	0.5482 ***	1.000
MP	0.6180 *	0.5344 **	0.6987 *	1.000
EC	0.2371 *	0.2675	0.3459 **	0.4581 *	1.000
IP	0.6165 ***	0.4939 ***	0.6886 ***	0.6950 *	0.4368 *	1.000
SC	0.7437 ***	0.3488 **	0.7745 **	0.6915 **	0.3248 **	0.7397 *	1.000
GS	0.1771 *	0.3625 **	0.0393 ***	0.0023	0.1626 *	0.1206 **	0.1848 *	1.000
IC	0.5884 ***	0.3086 ***	0.4736 **	0.6447 **	0.5465 ***	0.5557 ***	0.5818 ***	0.2319 *	1.000

Notes: (1) ***, **, and * denote significance at 1%, 5%, and 10% levels, respectively. (2) Abbreviations are provided at the end of the paper.

above, there is a partial correlation between bureaucratic quality and law and order of 0.63, which shows that a strong legal system goes hand in hand with better institutions. There are other papers in the literature that use the same dataset such as Busse and Hefeker (2007), Bove and Nistico (2014), and Bilgin et al. (2017). Due to the high multicollinearity among the political risk indicators in Table 4, we examine the models separately to enhance interpretability. This approach allows us to isolate marginal effects and follows similar empirical treatments to those in studies such as Busse and Hefeker (2007), where institutional dimensions were introduced independently due to their conceptual overlap. In robustness checks, we also estimate a joint model and employ principal component analysis (PCA) to extract orthogonal dimensions of risk (see Appendix A 

Table A1. Principal component analysis (PCA) summary.

Component	Explained Variance (%)	Cumulative Variance (%)	Top Contributing Variables
PC1	61.3	61.3	Law and Order, Bureaucratic Quality, Investment Profile
PC2	21.7	83.0	Democratic Accountability, Internal Conflict, External Conflict

). The results remain consistent in both direction and statistical significance, confirming the robustness of our main findings.

4. Empirical Analysis

4.1. Benchmark Case

We are interested in the effect of different dimensions of political risk on trade finance. The data on the method of payments from TUIK include the export transactions for each industry and the export market combinations. To empirically evaluate our hypothesis that industries extend more trade credit when political risk in the export destination is low, we begin with the following OLS specification:

$${XO}_{ijt}={\phi }_0+{\phi }_1{PR}_{jt}+Z_{jt}+{\epsilon }_{ijt}$$

(1)

where ${XO}_{ijt}$ denotes the share of export transactions settled under post-shipment terms (open account) in industry i to export destination j.4 We also used the log of the value of exports executed under post-shipment terms as the dependent variable in an alternative specification. ${PR}_{jt}$ stands for one of the indicators for political risk and $Z_{jt}$ is a set of control variables. Following Busse and Hefeker (2007), we add political risk indicators one by one to avoid multicollinearity. A high value for the political risk measure denotes a low risk for the export destination. Thus, we expect a positive sign for these variables. Additionally, all regressions are estimated using fixed-effects panel models that control for unobserved heterogeneity across destination countries, industries, and years. This panel estimation framework allows us to isolate the within-unit (i.e., country–industry) variation in payment term choices over time while accounting for structural factors that do not vary over time. The fixed-effects specification enables us to identify the effect of political risk on payment method choice within each cross-sectional unit over time rather than across pooled observations—aligning with best practices in the empirical trade finance literature. To provide further clarity, Appendix A Figure A3 illustrates how open-account usage varies across multiple countries and time periods, supporting the panel structure of the dataset.

Schmidt-Eisenlohr (2013)’s theory of trade finance suggests that an increase in the financing cost in a country makes it harder for its importers to offer cash in advance. For this consideration, we include the net interest margin, the net interest income of the banks relative to their total earning assets, in the export destination as a control variable. We expect a positive sign for this variable. We also add the log of the total value of exports to the export destination in the previous year as another control variable. We believe that past trade creates trust between global business partners, and it also gives them an opportunity to share information with companies in the same and other sectors. Therefore, we believe that industries are more likely to extend trade credit to their partners located in countries that had more trade volume with Turkey in the past. Some industry characteristics or product features like technology intensity may also affect the method of payment in cross-border trade. Although we acknowledge the importance of institutional features such as credit information depth, judicial independence, and banking structure in shaping export financing decisions, we were unable to include these variables directly in our regression models due to limitations in data availability and comparability across countries and over time. Many of these variables are reported only intermittently or are unavailable for the full panel of 96 destination countries between 2002 and 2010. To remedy this potential bias, first we include the industry fixed effects. For the aggregate variations in Turkey such as business cycle, exchange rate, and current account shocks, we used year fixed effects.

Second, we employ extensive robustness checks using income-level and regional subsamples (see Table 7), which indirectly account for institutional variation across country groups. Third, our identification strategy aligns with the prior literature that isolates the role of political risk in trade finance, even when broader institutional variables are not directly observable (e.g., Schmidt-Eisenlohr, 2013; Anderson & Marcouiller, 2002; Demir & Javorcik, 2018). The inclusion of more detailed institutional data would be a valuable addition to the literature and represents an important direction for future research, particularly as more comprehensive and panel-consistent measures become available.

Table 5. Estimation results. Dependent variable: share of open-account exports to country i in industry j.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
LO	0.014 ***
	(0.002)
DA		0.016 ***
		(0.001)
BQ			0.018 ***
			(0.002)
MP				0.019 ***
				(0.001)
EC					0.020 ***
					(0.002)
IP						0.008 ***
						(0.001)
SC							0.004 ***
							(0.001)
GS								0.003 **
								(0.001)
IC									0.016 ***
									(0.001)
Control Variables
Log (Exports)it−1	0.037 ***	0.035 ***	0.036 ***	0.035 ***	0.039 ***	0.036 ***	0.037 ***	0.038 ***	0.039 ***
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
Net Interest Margin	0.007 ***	0.006 ***	0.008 ***	0.008 ***	0.006 ***	0.007 ***	0.007 ***	0.005 ***	0.008 ***
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
N	24,886	24,886	24,886	24,886	24,886	24,886	24,886	24,886	24,886

Notes: (1) All estimations include a suppressed constant term, and year, country–industry, and country fixed effects. (2) Robust standard errors in parentheses; ***, and ** denote significance at 1%, and 5% levels, respectively. (3) Abbreviations are provided at the end of the paper.

displays the results. All columns include a suppressed constant term for year, industry, and country fixed effects. First, all the control variables have significant coefficients with expected signs. There is a positive relationship between net interest margin and the share of open-account transactions in exports. This supports the results obtained in the earlier studies that financing cost in the importing country leads to an increase in exporter-financed trade. In addition, the coefficient estimates for the log of the value of past exports (all industries) are positive and significant in all specifications. This suggests that open-account terms are more preferred when goods are shipped to countries that imported more goods from Turkey in the past.

When it comes to the variables of interest, all nine political risk variables provide positive and significant estimates across different specifications. Turkish exporters extended more trade credit to destinations with lower risk. Consider specification 1 to gauge economic significance. If an export destination in the 25th percentile of law-and-order rating moves to the 50th percentile, such a switch will increase the share of post-shipment-financed transactions by 2.1 percentage points. Similarly, if a country in the 50th percentile of law-and-order rating moves to the 75th percentile, around a 1.4 percentage point increase is estimated in the share of exporter-financed transactions resulting from this potential decrease in political risk.

We also estimate Equation (1) with a different dependent variable, using the log of the value of exports that occurred under post-shipment terms. Below,

Table 6. Estimation results. Dependent variable: log of value of open-account exports to country i in industry j.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
LO	0.112 ***
	(0.020)
DA		0.022 *
		(0.013)
BQ			0.103 ***
			(0.024)
MP				0.040 ***
				(0.015)
EC					0.035 **
					(0.016)
IP						0.032 ***
						(0.012)
SC							0.027 **
							(0.011)
GS								0.010 ***
								(0.004)
IC									0.005 ***
									(0.001)
Control Variables
Log (Exports)it−1	1.327 ***	1.333 ***	1.322 ***	1.330 ***	1.337 ***	1.326 ***	1.327 ***	1.336 ***	1.336 ***
	(0.013)	(0.013)	(0.014)	(0.014)	(0.013)	(0.014)	(0.014)	(0.013)	(0.013)
Net Interest Margin	0.019 **	0.000	0.016 *	0.005	0.000	0.006	0.010	0.002	0.002
	(0.009)	(0.008)	(0.009)	(0.009)	(0.008)	(0.009)	(0.009)	(0.008)	(0.009)
N	24,886	24,886	24,886	24,886	24,886	24,886	24,886	24,886	24,886

Notes: (1) All estimations include a suppressed constant term, and year and country–industry fixed effects. (2) Robust standard errors in parentheses; ***, **, and * denote significance at 1%, 5%, and 10% levels, respectively. (3) Abbreviations are provided at the end of the paper.

reports the results of this exercise. Additionally, the panel estimation output and a visual representation of the panel structure are provided in the Supplementary file. In line with the earlier results, all nine political risk indicators are significant with their expected signs. With respect to the size of the estimates, specification (1) indicates that a one-point increase in the government stability index in the importing country leads to an 11.8% increase in the value of exports occurring under open-account terms.

4.2. Robustness Checks

Do our findings depend sensitively on a set of observations that come from a special set of observations? We check by selectively excluding different sets of observations. We first exclude the observations from the Middle Eastern and North African regions since the period in our dataset coincided with political uprisings in these regions. We then successively deleted observations for African, Asian, and European countries. In addition, we excluded the observations from low-income, low-to-middle-income, upper-middle-income, and high-income countries in different specifications. As shown, our findings remain salient. Political risk indicators significantly impact our regressions with a positive sign.

Some readers may argue that our results are driven by traditional exporting sectors in Turkey. To address this concern, we also exclude the observations from the top three exporting industries, textile, basic materials, and machinery, and estimate using our model for the remaining industries. As documented in the last row of

Table 7. Subsample analysis.

	LO (1)	DA (2)	BQ (3)	MP (4)	EC (5)	IP (6)	SC (7)	GS (8)	IC (9)	N
D_MENA	0.011 ***	0.010 ***	0.009 ***	0.017 ***	0.019 ***	0.006 ***	0.004 ***	0.003	0.014 ***	21,873
	(0.002)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
D_African	0.018 ***	0.020 ***	0.038 ***	0.029 ***	0.027 ***	0.009 ***	0.005 ***	0.008 ***	0.020 ***	19,412
	(0.002)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
D_Asian	0.007 ***	0.008 ***	0.015 ***	0.009 ***	0.007 ***	0.003 **	0.005 ***	0.005 ***	0.004 ***	18,699
	(0.002)	(0.001)	(0.002)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)
D_European	0.007 ***	0.010 ***	0.007 ***	0.014 ***	0.015 ***	0.003 **	0.002 ***	0.001	0.010 ***	19,573
	(0.002)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
D_LowInc	0.012 ***	0.016 ***	0.021 ***	0.022 ***	0.021 ***	0.004 ***	0.002 ***	0.007 ***	0.017 ***	21,137
	(0.002)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
D_LMInc	0.008 ***	0.016 ***	0.018 ***	0.014 ***	0.014 ***	0.005 ***	0.001 ***	0.006 ***	0.012 ***	19,711
	(0.002)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
D_UMInc	0.021 ***	0.019 ***	0.021 ***	0.024 ***	0.024 ***	0.011 ***	0.004 ***	0.004 ***	0.021 ***	18,860
	(0.002)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
D_HighInc	0.017 ***	0.007 ***	0.011 ***	0.017 ***	0.018 ***	0.010 ***	0.010 ***	0.006 ***	0.014	16,054
	(0.002)	(0.001)	(0.003)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)
D_Textile & Basic Materials & Machinery and Equipment	0.014 ***	0.015 ***	0.018 ***	0.016 ***	0.019 ***	0.007 ***	0.003 ***	0.001	0.014 ***	21,640
	(0.002)	(0.001)	(0.002)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)

Notes: (1) All estimations include a suppressed constant term, and year, country–industry, and country fixed effects. (2) Robust standard errors in parentheses; ***, and ** denote significance at 1%, and 5%, levels, respectively. (3) Abbreviations are provided at the end of the paper.

, our results are insensitive to this treatment. Overall, none of the subsample analysis shakes the confidence we have in our main finding that increases in political risk in the export market are associated with a decrease in the share of exports executed under post-shipment terms. We also exclude country groups (such as MENA, Sub-Saharan Africa, and low-income countries) to test whether results are driven by specific macro-regions or developmental levels. This strategy is common in trade finance robustness checks (e.g., Demir & Javorcik, 2018) and helps assess whether our findings are globally stable or region dependent.

We also subject our results to alternative estimations. We begin by applying the following probit model:

$$\begin{gathered} P\left({PS}_{ijt}=1\right)={\mathsf{\theta }(\phi }_0+{\phi }_1{PR}_{it}+Z_{it}+yearfixedeffects+industryfixedeffects+ \\ countryfixedeffects+{\epsilon }_{ijt}) \end{gathered}$$

(2)

In this model, our dependent variable is a dummy and is applied if the value of post-shipment-financed transactions dominates the cash-in-advance and letter-of-credit transactions combined for an industry–importer combination. We estimate the probit model using the same set of control variables and political risk indicators.

Table 8. Probit estimation results’ average marginal effects.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
LO	0.110 ***
	(0.001)
DA		0.061 ***
		(0.006)
BQ			0.122 ***
			(0.011)
MP				0.007 ***
				(0.001)
EC					0.021 ***
					(0.002)
IP						0.061 ***
						(0.001)
SC							0.059 ***
							(0.000)
GS								0.014 ***
								(0.001)
IC									0.042 **
									(0.001)
Control Variables
Log (Exports)it−1	0.152 ***	0.150 ***	0.138 ***	0.150 ***	0.158 ***	0.146 ***	0.142 ***	0.158 ***	0.157 ***
	(0.008)	(0.008)	(0.008)	(0.008)	(0.008)	(0.008)	(0.008)	(0.008)	(0.008)
Net Interest Margin	0.022 ***	0.030 ***	0.012 ***	0.027 ***	0.031 ***	0.025 ***	0.020 ***	0.034 ***	0.029 ***
	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)
N	23,804	23,804	23,804	23,804	23,804	23,804	23,804	23,804	23,804

Notes: (1) All estimations include a suppressed constant term, and year, country–industry, and country fixed effects. (2) Robust standard errors in parentheses *** and ** denote significance at 1%, and 5% levels, respectively. (3) Abbreviations are provided at the end of the paper.

demonstrates the average marginal effects of the variables estimated from the probit model. Once again, all our political risk indicators remain positive and significant. For each additional increase in the rating for law and order, industries are 11% more likely to choose post-shipment terms over pre-shipment terms. Similarly, a point increase in the democratic accountability rating increases the likelihood of industries offering trade credit by 6%.

We also performed ordered logit regressions for method of payments in industries’ exports. To do so, we considered three groupings of export financing terms: pre-shipment terms, letters of credit, and post-shipment terms.5 Our dependent variable in this case takes on a value of 1 if the majority of exports to a specific destination for an industry occurred under cash-in-advance (pre-shipment) terms. Similarly, it becomes 2 for letter-of-credit (bank intermediation) and 3 for open-account (post-shipment) terms.

Table 9. Ordered logit estimation.

	1	2	3
	Marginal Effects for Pre-Shipment	Marginal Effects for Bank Intermediation	Marginal Effects for Post-Shipment
Law and Order	−0.004 ***	−0.005 ***	0.009 ***
	(0.000)	(0.001)	(0.001)
Democratic Accountability	−0.005 ***	−0.008 ***	0.014 ***
	(0.000)	(0.000)	(0.001)
Bureaucratic Quality	−0.004 ***	−0.007 ***	0.012 ***
	(0.000)	(0.000)	(0.001)
Military in Politics	−0.006 ***	−0.009 ***	0.015 ***
	(0.000)	(0.000)	(0.001)
External Conflict	−0.002 ***	−0.004 ***	0.007 ***
	(0.000)	(0.000)	(0.001)
Investment Profile	−0.002 ***	−0.004 ***	0.006 ***
	(0.000)	(0.000)	(0.000)
Socioeconomic Conditions	−0.001 ***	−0.002 ***	0.003 ***
	(0.000)	(0.000)	(0.001)
Government Stability	−0.002 ***	−0.003 ***	0.006 ***
	(0.000)	(0.000)	(0.001)
Internal Conflict	−0.004 ***	−0.007 ***	0.012 ***
	(0.000)	(0.000)	(0.001)
N	24,886	24,886	24,886

Notes: (1) This table presents the marginal effects for three trade financing terms. (2) Each specification includes a constant term, control variables, and year, country-industry, and country fixed effects. (3) Robust standard errors in parentheses. *** denotes significance at 1% levels.

documents the ordered logit estimations. We show the marginal effects of political risk variables on the probability of pre-shipment, bank intermediation, and post-shipment terms in specifications 1, 2, and 3, respectively. Estimates suggest that political risk increases the likelihood of cash-in-advance and letter-of-credit terms whereas it decreases the likelihood of post-shipment terms. According to the marginal effects reported in Table 9, a one-unit increase in bureaucratic quality index is associated with a 1.2% increase in the likelihood of having an export transaction under open-account terms and a 0.4% decrease in the likelihood of cash-in-advance terms. Overall, both probit and ordered logit estimations confirm our hypothesis on the relationship between political risk and trade finance.

4.3. The Role of Industry Complexity

Carrying out a trade finance deal in complex, innovation-intensive manufactured products requires a significant exchange of information between trade partners. Selling these complex products often requires relationship-specific arrangements, face-to-face interactions, quality controls, and technical support as well as more dependence on financial institutions. In the periods of political instability and conflicts, importers of these industries will be more affected by the risk premiums, and it will be harder for the exporters to follow up on a payment in the case of a default. As a result, we believe that business partners in these industries will be more affected by political risks. To test whether the effect of political risk is more pronounced for complex industries, we interact our political risk variables with the industry complexity measures from Nunn’s (2007). Nunn (2007) developed a complexity measure using an input–output table from the US. In his classification, sectors that use a large share of intermediate inputs are classified as more complex.

Table 10. Estimation results: industry complexity interactions. Dependent variable: share of open-account exports to country i in industry j.

	1	2	3	4	5	6	7	8	9
Law and Order	0.009 ***
	(0.001)
LO × Complexity	0.012 ***
	(0.001)
Democratic Accountability		0.011 ***
		(0.003)
DA × Complexity		0.006 ***
		(0.001)
Bureaucratic Quality			0.013 ***
			(0.004)
BQ × Complexity			0.014 ***
			(0.001)
Military in Politics				0.013 ***
				(0.003)
MP × Complexity				0.016 ***
				(0.005)
External Conflict					0.018 ***
					(0.003)
EC × Complexity					0.010 ***
					(0.006)
Investment Profile						0.006 ***
						(0.001)
IP × Complexity						0.004 **
						(0.002)
SC							0.003 ***
							(0.001)
SC × Complexity							0.003 ***
							(0.001)
Government Stability								0.002 ***
								(0.000)
GS × Complexity								0.001
								(0.006)
Internal Conflict									0.012 ***
									(0.003)
IC × Complexity									0.001 ***
									(0.000)
N

Notes: (1) Each specification includes a constant term, control variable, and year, country–industry, and country fixed effects. (2) Robust standard errors in parentheses. ***, and ** denote significance at 1% and 5% levels, respectively. (3) Abbreviations are provided at the end of the paper.

shows the estimates when we use the share of post-shipment-financed exports as the dependent variable and interact political risk indicators with the complexity measure. In all nine specifications, both the political risk measures and the interaction terms are positive and significant. This suggests that the role of political risk in terms of extending trade finance is more important for complex industries. In an industry of low complexity (basic metals), a 1-point increase in the law-and-order rating increases the share of open-account exports by around 1.1 percentage points, but the same change causes around a 2-percentage point increase for an industry of high complexity (motor vehicles).

5. Conclusions

Serving international markets brings extra risks to companies, which they do not have in the domestic side. One of the risks in global business is the political risk which arises from governments’ policies and actions that can directly affect business operations and firms’ ability to perform their economic obligations. Political instability is on the rise worldwide. Ten years of data in the Fragile State Index indicate that the number of states in the ‘high alert’ and ‘very high alert’ categories has more than doubled in the last 15 years. Studies have shown the negative impact of political risks on foreign trade and investment. In this paper, we complement and extend this line of research by analyzing the effect of political risk on trade finance. Trade finance is the lifeblood of cross-border transactions and has become a central concern of governments, especially after the trade collapse in the aftermath of the global financial crisis. Our paper suggests that one of the channels through which political risks dampen global trade flows is the payment channel.

Political risk in the importer’s country increases the risk premium in financial markets and drives up the cost of financing, which makes it harder for the buyer to offer cash in advance. Second, sellers will perceive higher risks in terms of following up on their payments and thus be reluctant to extend trade credit when there is political instability and repression in export markets. In line with these arguments, we hypothesize that export transactions executed under open-account terms increase if the political risk in the export markets decreases. Using industry-level trade finance data from Turkey, which reports payment methods for importer–industry combinations, and applying linear as well as maximum likelihood models, we provide strong support for this hypothesis. We also show that political risk disproportionately affects industries that produce complex products in terms of extending trade finance. Our analysis is based on bilateral, industry-level export data from Turkey—a G20 emerging economy with a trade structure that is geographically concentrated in Europe and characterized by mid-complexity manufacturing sectors. While this allows for detailed institutional- and product-level insights, we caution against unqualified generalization of the results to other large exporters with different institutional frameworks or export compositions (e.g., China, Germany, or Japan). Future research should replicate this analysis using firm- or industry-level data from multiple exporter countries to test the external validity of our findings under diverse regulatory, financial, and political contexts. The use of harmonized trade finance datasets (e.g., WBES, BACI with INCOTERMS) could be particularly useful in exploring cross-national heterogeneity in response to political risk.

There are several avenues for future research in this area. For instance, it would be interesting to investigate the effect of political risk on the currency choice in export transactions. Does political risk increase the usage of vehicle currency in trade rather than the importing country’s currency? Answering this question can increase our understanding of the effect of political risk on international trade and finance. Political risk in the export destination may also impact the delivery methods and trade insurance policies in cross-border transactions as well. However, pursuing these research questions requires more detailed cross-border data on payments and insurance policies.