The effect of political risk on China’s foreign direct investment

Abstract

This study examines the impact of political risk on Chinese outward foreign direct investment (OFDI) and what motivates their preferred location. The study also analyzes the OFDI of other countries to enhance the comparison of China and other countries’ OFDI sensitivity to political risk. The study used annual panel data on 134 countries from 2003 to 2017. The results indicate that China’s OFDI tends to favor countries and regions with higher expropriation risk, and China’s OFDI exhibits strong resource-seeking motives and weak market-seeking motives. On the other hand, OFDI in countries around the world tends to favor countries and regions with lower expropriation risk and conflict risk, and OFDI in those countries exhibits market-seeking motives. The study results also show that China’s political risk preference and investment motives depend on the level of economic development and the presence of natural resources in the host country.

Keywords:

• foreign direct investment
• expropriation risk
• outward foreign direct investment
• political risk

JEL Classifications:

• D7
• F2
• F3
• F4

1. Introduction

In the last three decades, foreign direct investment (FDI) has grown dramatically as the main form of international capital transfer (Graham et al., 2014). Despite this development, FDI remains unevenly distributed, especially among developing countries (Kamal et al., 2019); why is this the case? Several explanations have been given, but the most prominent one centers on political risk (Jiménez, 2011; Jiménez et al., 2011). Political risk is the possibility that a business could suffer because political changes or instability in a country (Dunning, 2002). Political risk is known to come in various forms (Graham et. al., 2016), with each carrying its own set of concerns for foreign investors.

To this end, a growing literature (e.g., Jiménez, 2011; Jiménez et al., 2011; Gaoyan, 2020; Quer et al., 2011; Click, 2005) has analyzed the nexus between the various forms of political risk and FDI although the findings are contradictory. For example, Gaoyan (2020) employed the principal component analysis (PCA) to construct a novel political risk index (PRI) that measures multiple facets of political risk for 139 countries to examine the changes in the political risk distribution (PRD) of outward FDI (OFDI) regarding investment destinations, large projects, annual investment outflows and sectorial distributions. The study establishe a negative impact of political risk on OFDI. Similarly, Jiménez, 2011 analyzed FDI flows from southern European countries to one of two nearby developing regions: north African countries and new European Union member states in central and eastern Europe. The study found that greater levels of political risk, measured through scales of political discretion, corruption, and economic freedom, do attract higher inflows. Despite the fact that one might expect global flows to fall as a consequence of political risk, those from the countries in the sample increase, because they come from firms that are searching for a market niche where they can take advantage of their political capabilities. Quer et al. (2011), on the other hand, examined the effect of political risk and cultural distance on the location patterns of large companies and revealed that high political risk in the host country does not discourage multinational companies. However, from a more conventional point of view, the presence of Chinese companies in the host country is positively associated with Chinese outward foreign direct investment (FDI). Thus, the evidence on relationship between political risk and foreign direct investment is inconsistent and the contradictions in the results give room for further analysis on the nexus. From a logical point of view, political risk should negatively impact FDI because political instability increases uncertainty in the economic environment, which will lower the incentives for foreign investors to invest in the host country.

Distinctively, the analysis in this paper is based on a panel data of 134 countries worldwide from both developed and developing economies from 2003 to 2017. The present study explores the political risk indices and the impact on outward foreign direct investment (OFDI). Two separate empirical models are constructed in this paper to enhance the comparison of China’s and the world’s preferences for political risk in outward direct investment. The first model examines the world’s preference for political risk in overseas direct investment, and the second model ultimately examines the the impact of political risk on Chinese OFDI using the System Generalized Method of Moment (SYS-GMM) estimator, given that the past values of the stock of OFDI in a country may determine the present value of OFDI. The study found that OFDI in countries around the world tends to favor countries and regions with lower expropriation risk and conflict risk, as opposed to China’s OFDI which tends to favor countries and regions with higher expropriation risk. The OFDI in countries around the world also exhibited more market-seeking motives in terms of investment; while China’s OFDI exhibited strong resource-seeking motives and weak market demand motives.

The evidence obtained in this study contributes to the existing literature in three ways. First, by including 134 countries in our sample, the study provides empirical evidence on the political risk preference for OFDI of countries all over the world. Second, this study provides insight into the political risk preferences of China’s OFDI based on the differences in the categorized development stages of host economies; that is, developed countries, emerging market countries, other developing countries, and countries along the “One Belt, One Road”. Finally, the study provides details on the political risk preferences of China’s OFDI to all countries in the world. The rest of the study is structured as follows: Section 2 presents the research method and data; Section 3 presents analysis and discussion of the research results, and Section 4 concludes the study.

2. Literature review

Gaoyan (2020) set out to enhance understanding of the nexus between Chinese Outward Foreign Direct Investment (OFDI) and host country political risk. The study employed 15 proxy variables and applied principal component analysis (PCA) to construct a new political risk index (PRI) that measures multiple facets of political risk for 139 countries. Using this new PRI criterion, the study looked into changes in the political risk distribution (PRD) of Chinese outward FDI (OFDI) regarding investment destinations, large projects, annual investment outflows and sectorial distributions from 2006–2017. The study found that the vast majority of Chinese OFDI during this period is concentrated in moderate and low-risk countries, even at the sectorial level. Furthermore, the continuing reform of Chinese OFDI policy and strong government support have led to an unprecedented increase in Chinese OFDI, while the PRD of Chinese OFDI has maintained a gradual decline over the past decade.

Quer et al. (2011) assessed the influence of political risk and cultural distance on the location patterns of large Chinese companies. They established that high political risk in the host country does not discourage Chinese multinationals. However, from a more conventional point of view, the presence of Chinese diasporans in the host country is positively associated with Chinese outward foreign direct investment (FDI). In addition, firm size and the volume of Chinese exports to the host country have a positive influence.

Dunning (2002), in a ground-breaking paper, provided an adapatation of UNCTAD (2001) dashboard of host country determinants and muilti-national investor motives for OFDI. The paper established policy framework for FDI, economic factors and business facilitation as the host country determinants to be considered for OFDI; while market-seeking, resource-seeking, efficiency-seeking and asset-seeking motives could drive OFDI from an investor perspective.Kamal et al. (2019) designed a study to establish the motivations of Chinese FDI in 30 Asian countries for 2003–2016, using the Random effect (RE), Fixed effect (FE) and System-GMM (SGMM) methods. The study incorporated both market and natural resource (mineral richness) seeking motives of Chinese FDI in the analysis. Regarding income groups, the study confirmed the market-seeking FDI in both high and middle-income countries whereas, mineral richness is priority for Chinese FDI in the middle-income group. Thus, Chinese firms targeted middle income developing economies to acquire non-fuel natural resources. On the regional basis, the results show that in all regression models, GDP is a positive and significant predictor, characterising market-seeking FDI by Chinese firms in West, East and South East Asia. For the resource-seeking motive, among the two types of natural resources, mineral richness affects Chinese FDI positively in East & South East Asia. Thus, the market-seeking motive is common for Chinese FDI in the entire sample, whereas the resource-seeking motive varies across the income groups and regions.

Taking 2006–2016 as the time window and national distance as the perspective, Ren and Yang (2020) examined the location choice of Chinese OFDI by using an OFDI size determination model. The study ascertained that geographical distance, economic distance and informational distance will significantly promote Chinese OFDI, while institutional distance will restrain it, and cultural distance will not have a significant effect on Chinese OFDI. In addition, Chinese OFDI has strong natural resource-seeking motivation. In order to obtain natural resources, Chinese enterprises will overcome geographical distance and cultural difference to conduct OFDI in the countries far away from China.

Chen (2015) intiated an enquiry into the determinants of outward FDI by China’s provincial firms. The study found that provincial economic development, innovation and technology, and export to GDP ratio were statistically significant, while FDI inflows, import to GDP ratio and provincial market size were not statistically significant. The results suggest that the main motives for China’s provincial firms to invest abroad are mainly market-seeking and efficiency-seeking.

Stoian (2013) assessed the home country determinants of OFDI from post-communist economies. Using the Investment Development Path (IDP) with inputs from institutional theory, the study explains the effects of home country institutional factors on the level of OFDI. The hypotheses is tested using random effects estimations on a comprehensive panel dataset comprising OFDI from 20 Central and Eastern European (CEE) countries. The results affirmed the IDP’s main propositions but also established the significance of accounting for home country institutional factors when investigating the determinants of OFDI. Particularly, Stoian (2013) found that the inclusion of institutional variables increases the explanatory power of the models and that competition policy and overall institutional reforms play a crucial role in explaining OFDI from CEE countries.

Das (2013) examined home country determinants of OFDI for selected developing economies for 1996–2010, using a panel data framework. The results indicate that source country’s level of economic development, globalisation, political risk and science and technology investments contribute significantly to outward FDI from developing countries. While outward FDI might be unavoidable in the course of economic development and globalisation, developing countries need to emphasise improving political governance in order to prevent capital outflow arising out of high domestic political risk. On the flip side, science and technology investments could contribute to higher OFDI, thereby yielding complementary benefits of internationalisation in the long-run.

3. Research method and data

3.1. Measurement of study variables

3.1.1. Dependent variables

The dependent variables used in the study are the stock of OFDI in countries around the world and the stock of China’s OFDI.

3.1.1.1. The stock of OFDI in countries around the world

The stock of FDI is the overseas FDI of countries around the world. Equations 1 and 2 focuses on the extent to which political risk affects OFDI in countries around the world, and further supports empirically whether the presence of expropriation risk indirectly affects the deterrent effect of other political risks on overseas FDI. In line with Camarero et al. (2020) this study uses the stock of FDI. The stock of OFDI is chosen instead of the flow for the following reason. First, the stock has a relatively long-time span, which can enrich the sample capacity, while the flow has a short span with many missing middle years, which affects the completeness of the sample. Second, the stock data is relatively less volatile and many OFDIs are continuous and not completed in one year, while the flow data of some years have negative or zero values, which cannot be studied after taking the logarithm.

3.1.1.2. The stock of Chinese OFDI

Consistent with Camarero et al. (2020), the stock of China’s OFDI is mainly used as a dependent variable in Equation 3, reflecting the stock of China’s OFDI to countries around the world. The main purpose is to study the political risk appetite of China in overseas direct investment, in other words, the ability and level of a country’s overall political environment to attract Chinese OFDI.

3.1.2. Independent variables

3.1.2.1. Political risk variables

Political risk mainly includes expropriation risk, government stability, government efficiency, degree of democracy, and conflict. The choice of indicators and measurement of the political risk variables is motivated by prior studies such as Jeutang and Kesse (2021) and Gakpa (2020). Expropriation risk is the risk that the host government will expropriate the assets of the foreign enterprise through coercive measures, decrees, and limiting the percentage of shareholding. There is no doubt that the existence of expropriation risk can be extremely costly for investors in the home country. The strength of contract enforcement in the International Country Risk Guide (ICRG) database is used to measure the magnitude of a country’s expropriation risk, with a higher score implying stronger contract enforcement or lower expropriation risk and vice versa. Government efficiency is measured by the mean values of the quality of officials and the degree of government integrity (i.e., lack of corruption) in the ICRG as indicators, where higher scores represent a more efficient host government and better quality of governance a better domestic business environment. Government stability is measured as the stability of socio-economic conditions in ICRG. A higher score means that the host government is more stable and the socio-economic conditions are relatively more favorable. The degree of democracy reflects the impact of a country’s degree of domestic democracy on foreign business enterprises; computed as the mean of the degree of democratic accountability, ethnic conflict, religion, and military intervention in politics from the ICRG data. Conflict also represents domestic and foreign conflicts, violence, terrorist incidents, and internal and external wars in a country; computed as the mean value of internal and external conflicts in ICRG. A higher score implies a more stable country and vice versa.

3.1.2.2. Investment motivation variables

Domestic and foreign scholars mainly classify investment motives into three types: technology-seeking, market-seeking and resource-seeking. We measure market-seeking and resource seeking motives in line with Kamal et al. (2019) while the proxy used for technology-seeking motive is consistent with Driffield et al. (2021). Technology-seeking motive in this paper is the measure of a country’s investment in R&D as a share of GDP and the number of people engaged in research and technical personnel. Developed countries are currently controlling most highly sophisticated technologies, so the technology-seeking variable is added mainly in developed countries. Market-seeking motive is the measure of GDP as a direct reflection of a country’s purchasing power and consumption capacity; while the population variable directly represents the market size of a country. Resource-seeking motive computes the proportion of rents of natural resources in GDP, and the proportion of exports of crude oil and metals as the degree of natural resource endowment of the host country. Natural resources are very attractive to foreign investors, especially for traditional resource-based firms. Even in the presence of turmoil in the host country, many firms are still desperate to obtain resources.

3.1.3. Control variables

Based on the relevant empirical literature (e.g., Camarero et al., 2020; Gaoyan, 2020), the lagged period of the dependent variables, the growth rate of GDP and total bilateral trade are added to the models as control variables. OFDI usually has the “aggregation effect”, and the previous investment influences the current investment to a certain extent; GDP growth rate and the macroeconomic development speed of a country reflect, which usually has a certain promotion effect on FDI. Especially for developing countries, the increase of GDP growth rate attracts overseas investors to some degree; the total bilateral trade represents the trade effect, i.e. the pulling effect of trade on investment.

3.2. Data

Table 1 shows a summary of the variable description and the data sources. The data covers 134 countries for the period 2003 to 2017.

Table 1. Variables and data sources

Variable Types	Variable notation	Meaning of variables	Data sources
Dependent variables	Lnstock	Logarithm of OFDI stock in each country	World Bank
	Lngross	Logarithm of China’s OFDI stock	China Overseas Direct Investment Announcement
Political risk variables	Expro	Lack of expropriation risk	ICRG
	Gove	Government Efficiency	ICRG
	Govs	Government Stability	ICRG
	Pac	Degree of democracy	ICRG
	Cona	Absence of conflict	ICRG
Market-seeking variables	Lngdp	Logarithm of the host country’s GDP	World Bank
	Lnpop	Logarithm of the host country population	World Bank
	Lnpgdp	Logarithm of host country GDP per capita	World Bank
Resource-seeking variables	res1	Natural resource rents as a share of GDP by country	World Bank
	res2	Crude oil and metals exports by country as a share of total exports	World Bank
Tech-seeking variables	Rde	Investment in R&D as a share of GDP by country	World Bank
	Rd	Number of research and technical staff in each country	World Bank
Control variables	L.lnstock	Lagged period of OFDI stock in each country	World Bank
	L.lngross	Lagged period of China’s FDI stock	World Bank
	Growth	GDP growth rate of each country	World Bank
	Lntrade	Country’s total trade with other countries	UN Commodity Cargo Database

3.3. Empirical models

The study examines the impact of political risk on Chinese OFDI and what motivates their preferred investment destinations. Two separate empirical models are constructed in this paper to enhance the comparison of China’s and the world’s preferences for political risk in outward direct investment

First, the study examines the world’s preference for political risk in overseas direct investment, and ultimately, the impact of political risk on Chinese OFDI. These are done in line with empirical models used in similar studies (e.g., Jiménez, 2011; Agyeman et al., 2021; Click, 2005; Gaoyan, 2020; Jeutang & Kesse, 2021). Our first baseline model for the world’s preference for political risk in overseas direct investment is expressed as follows:

(1) $\begin{array}{rl}& lnstoc{k}_{it}={\beta }_0+{\beta }_1\times expr{o}_{it}+{\beta }_2\times gov{e}_{it}+{\beta }_3\times gov{s}_{it}\\ & +{\beta }_4\times pa{c}_{it}+{\beta }_5\times con{a}_{it}+{\beta }_6\times lngd{p}_{it}\\ & +{\beta }_7\times lnpgd{p}_{it}+{\beta }_8\times lnpo{p}_{it}+{\beta }_9\times growt{h}_{it}+Country+Year+{\epsilon }_{it}\end{array}$(1)

Where $lnstoc{k}_{it}$ denotes the stock of FDI in country i in year t,$expr{o}_{it}$, $gov{e}_{it},gov{s}_{it},pa{c}_{it}$ and $con{a}_{it}$ denote the lack of expropriation risk, government efficiency, government stability, degree of democracy, and lack of conflict in country i in year t, respectively, which mainly measure the political risk of the host country. $lngd{p}_{it},lnpgd{p}_{it},\mathrm{a}\mathrm{n}\mathrm{d}lnpo{p}_{it}$ denote the GDP, GDP per capita and population of country i in year t, respectively; $growt{h}_{it}$ denotes the GDP growth rate of country i in year t; Year and Country denote the country and year fixed effects, respectively while ${\epsilon }_{it}$ is the random error term.

The past values of the stock of OFDI in a country may determine the present value of OFDI stock in Eq. (1) may not be strictly exogenous. Thus, we employ a dynamic model, the System Generalized Method of Moment (SYS-GMM) estimator proposed by Arellano and Bover (1995) and Blundell and Bond (1998). By employing the SYS-GMM estimation method, we are able to overcome two critical econometric issues: (i) since the prior values of OFDI can determine the present values, the SYS-GMM affords us the opportunity to use the lagged values of the dependent variables to exploit the dynamic nature of the data. (ii) the explanatory variables may not be strictly exogenous, and the use of SYS-GMM can eliminate endogeneity issues while using lagged levels and lagged differences of the regressors as instruments. The dynamic form of the basic model is specified as follows:

(2) $\begin{array}{rl}& lnstoc{k}_{it}={\beta }_0+{\beta }_1\times L.lnstoc{k}_{it-1}+{\beta }_2\times expr{o}_{it}+{\beta }_3\times gov{e}_{it}+{\beta }_4\times gov{s}_{it}\\ & +{\beta }_5\times pa{c}_{it}+{\beta }_6\times con{a}_{it}+{\beta }_7\times lngd{p}_{it}\\ & +{\beta }_8\times lnpgd{p}_{it}+{\beta }_9\times lnpo{p}_{it}+{\beta }_{10}\times growt{h}_{it}+Country+Year+{\epsilon }_{it}\end{array}$(2)

We further examine the impact of political risk on China’s OFDI using the following model:

(3) $\begin{array}{rl}& lngros{s}_{it}={\beta }_0+{\beta }_1\times L.lngros{s}_{it-1}+{\beta }_2\times expr{o}_{it}+{\beta }_3\times gov{e}_{it}+{\beta }_4\times gov{s}_{it}\\ & +{\beta }_5\times pa{c}_{it}+{\beta }_6\times con{a}_{it}+{\beta }_7\times lngd{p}_{it}\\ & +{\beta }_8\times lnpgd{p}_{it}+{\beta }_9\times lnpo{p}_{it}+{\beta }_{10}\times growt{h}_{it}+{\beta }_{11}\times lntrad{e}_{it}+Country+Year+{\epsilon }_{it}\end{array}$(3)

Where $lngros{s}_{it}$ denotes the stock of OFDI from China to country i in year t, $L.lngros{s}_{it-1}$ denotes the stock of OFDI from China to country i in year t-1; $lntrad{e}_{it}$ denotes the total import and export trade from China to country i in year t; other variables have the same meaning as those in Equations (1–2).

4. Results and discussion

The empirical analysis in this section consists of four main parts: first: descriptive statistical analysis of the main variables; second: study of political risk preferences of overseas investment in countries around the world; third: full-sample analysis of political risk preferences of overseas direct investment in China; and fourth: sub-sample analysis of political risk preferences of overseas direct investment in China.

4.1. Descriptive statistics

Table 2 shows the descriptive statistics for the main variables. The stock of foreign investment attracted by countries in the world, the stock of OFDI in China, GDP, GDP per capita, population size, and total bilateral trade are expressed in natural logarithm. Political risk variables and resource variables are expressed in their raw form because they are relatively less volatile. The detailed results are shown in Table 2.

Table 2. Descriptive statistics of variables

Variable	Obs	Mean	Std.Dev.	Min	Max
Lnstock	1888	11.25	0.978	6.152	12.81
Lngross	1734	7.854	1.207	4	11.57
Expro	1943	2.349	0.778	0.500	4
Gove	1957	2.314	1.062	0.500	5
Govs	1984	5.575	2.624	0	17
Pac	1957	4.127	1.048	0.855	6
Cona	1957	9.582	1.323	3.775	11.75
Lngdp	1881	10.812	0.858	8.649	13.87
Lnpgdp	1881	3.272	0.678	2.077	5.778
Lnpop	1932	7.097	0.688	5.373	9.184
Growth	1878	0.044	0.057	−0.611	1.737
Lntrade	1886	9.417	0.975	4.841	11.783
res1	1876	0.098	0.145	0	0.641
res2	1783	0.243	0.300	0.000456	0.978
Rde	967	1.094	0.951	0.00827	4.455

4.2. Regression results

4.2.1. Analysis of political risk appetite of countries around the world for OFDI

Table 3 shows the GMM and the mixed regression results for the impact of political risk on OFDI of countries around the world. Models 1 to 3 the system GMM estimates based on different proxies for political risk. Model 1 focuses on expropriation risk in a narrow sense: the strength of contract enforcement. The expropriation risk in model 2 includes the strength of contract enforcement and transfer payment. The expropriation risk in model 3 covers a broader range, which includes the strength of contract enforcement, transfer payments, the stability of the government, and law and order. Model 4 is the estimation result of the mixed regression model similar to Model 1.

Table 3. Regression results for impact of political risk on OFDI of countries around the world

Variables	Mode 1	Model 2	Model 3	Model 4
L.lnstock	0.871***	0.870***	0.872***	0.974***
	(18.38)	(17.98)	(19.01)	(215.6)
Expro	0.018**	0.024***	0.022***	0.018***
	(2.87)	(3.01)	(2.67)	(3.67)
Gove	−0.007	−0.009	−0.007	−0.013***
	(−0.83)	(−1.05)	(−0.90)	(−2.86)
Govs	0.001	0.001	−0.003	0.003
	(0.15)	(−0.33)	(−0.83)	(1.27)
Pac	0.012	0.006	0.009	−0.001
	(1.09)	(0.57)	(1.20)	(−0.18)
Cona	0.009**	0.008	0.001	0.007**
	(2.01)	(1.74)	(0.08)	(2.27)
Lngdp	−0.764	−0.741	−0.753	−0.519***
	(−1.61)	(−1.52)	(−1.61)	(−2.64)
Lnpgdp	0.852*	0.812*	0.841*	0.544***
	(1.71)	(1.62)	(1.75)	(2.75)
Lnpop	0.841*	0.817*	0.843*	0.554***
	(1.72)	(1.63)	(1.76)	(2.80)
Growth	0.09	0.094	0.072	0.117**
	(1.47)	(0.67)	(1.38)	(2.48)
Constant	0.163	0.202*	0.094	0.078
	(1.44)	(1.84)	(0.88)	(1.43)
Country effect	Yes	Yes	Yes	Yes
Year effect	Yes	Yes	Yes	Yes
Obs	1603.000	1603.000	1603.000	1603.000
F-test	1688.774	1970.791	1971.016	12,056.69
AR (1)	0.000	0.000	0.000
AR (2)	0.371	0.377	0.371
R-squared				0.987

Note: Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1. Variable definitions and measurements are explained in Table 1.

From Table 3, the value of AR (1) is 0.000 and the value of AR (2) is 0.371 > 0.1, which signifies that the model does not suffer serial autocorrelation. The R-squared of model 4 is 0.987, indicating that the regression results are reasonable. In model 1, the lag of the dependent variables is positive at 1% significance level, and the coefficient is 0.871 higher than the parameters of the other variables, indicating that the OFDI of countries around the world is continuous. The risk of expropriation is positive at the 5% significance level, as expected, indicating that for every 1 unit reduction in the risk of expropriation in the host country, the stock of investment in other countries is increased by 0.9%, which affirms the conclusions in Gaoyan (2020). It can be seen that the three political risks of government efficiency, government stability and the degree of democracy do not pass the 10% significance level test, and even the coefficient of government efficiency is negative. Usually, improving a country’s government efficiency, a stable political environment and a good democratic system are conducive to attracting investment from overseas investors. However, through the equilibrium model of expropriation risk, it is found that the existence of expropriation risk will, to a certain extent, nullify the negative effect of other political risks on OFDI. Globally, the overseas direct investment will be more affected by expropriation risk, and even under the premise of facing higher expropriation risk in the host country, improving government efficiency will lead to the loss of foreign investment instead. Conflict risk is significantly positive at 5% significance level as expected. The mechanism of the influence is similar to that of expropriation risk, where the host country facing war is unable to provide the most basic property protection for foreign enterprises. However, the conflict variables in models 2 and 3 are not significant, mainly because the broad conflict includes diplomatic environment, trade restrictions, war, terrorism, international sanctions, and other factors. The effect of GDP on FDI is not significant, but GDP per capita and population size are positive at the 10% significance level, indicating that foreign investment tends to be in countries with higher purchasing power and better market prospects. Models 2 to 4 are also significant. From the analysis of the full sample of countries around the world, it is evident that foreign investors prefer to invest in countries with lower expropriation risk and with fewer conflicts which agrees with Gaoyan (2020), regardless of the effects of government efficiency, government stability, and the degree of democracy on OFDI which were discovered not to be significant in the face of expropriation risk.

4.2.2. Analysis of political risk preferences of China for OFDI

Table 4 shows the political preferences of China for OFDI. Models 1 and 2 are the results of the system GMM analysis using expropriation risk in a narrow sense: the strength of contract enforcement Model 2 adds resource variables to Model 1. Model 3 is the regression result of the system GMM based on expropriation risk in a broader sense: the strength of contract enforcement, transfer payments, the stability of the government, and law and order. Model 4 is the mixed-effects regression result of Model 1. Among these models, the main references are models 1 and model 2, and the other models are used as robustness checks.

Table 4. Empirical results of political risk preferences of China for OFDI

Variables	Model 1	Model 2	Model 3	Model 4
L.lngross	0.857***	0.876***	0.834***	0.840***
	(32.02)	(41.26)	(43.62)	(34.26)
Expro	−0.042***	−0.035**	−0.014	−0.034***
	(−2.32)	(−2.42)	(−1.30)	(−3.68)
Gove	0.023	0.021**	0.043*	0.063***
	(1.43)	(2.41)	(1.65)	(3.39)
Govs	0.001	0.001	0.004	0.004
	(1.03)	(1.09)	(0.66)	(1.19)
Pac	−0.023**	−0.022	−0.035	−0.028
	(−2.32)	(−1.38)	(−1.29)	(−1.33)
Cona	0.027*	0.029*	0.028*	0.027*
	(1.64)	(1.79)	(1.71)	(1.76)
Lngdp	0.948	0.905	0.754	0.915
	(1.42)	(1.12)	(0.97)	(1.54)
Lnpgdp	−0.659	−0.937	−0.768	−0.856*
	(−1.33)	(−1.33)	(−0.66)	(−1.16)
Lnpop	−0.946	−0.731	−0.579	−0.769
	(−1.25)	(−1.12)	(−0.77)	(−1.66)
Growth	0.021	0.271*	0.242	0.275*
	(0.49)	(1.63)	(1.57)	(1.42)
Lntrade	0.082	0.030	0.034	0.034***
	(1.67)	(1.05)	(0.89)	(3.52)
res1		0.327***	0.354***	0.343***
		(3.32)	(3.46)	(4.33)
Constant	0.179	0.028	−0.161	0.093
	(1.23)	(0.54)	(−0.62)	(0.26)
Country effect	Yes	Yes	Yes	Yes
Time effect	Yes	Yes	Yes	Yes
Obs	1503.000	1498.000	1498.000	1498.000
F-test	984.530	1173.861	1119.916	1558.335
AR (1)	0.000	0.000	0.000
AR (2)	0.632	0.634	0.624
R-squared				0.926

Note: Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1. Variable definitions and measurements are explained in Table 1.

In the regression results in Table 4, the AR (1) of model 1 is 0.000 and the value of AR(2) is 0.632 > 0.1, indicating that there is no second-order autocorrelation in the random error term of the system GMM. The R-squared of model 4 is 0.987, indicating that the regression results are reasonable.

The coefficient of expro here must be understood to be significantly negative at the 1% significance level which is contrary to the political risk preference of OFDI for the rest of the world. Theoretically, the presence of expropriation risk can be a serious deterrent to foreign investors. However, China presents a significant phenomenon of expropriation risk preference, especially in countries Africa and North America such as Venezuela, Zambia, Zimbabwe, South Africa, Iran, Nigeria and Congo with relatively high expropriation risk, with relatively regressive economic development and abundant natural resources. Based on this, natural resource variable (res1) is included in Model 2 for in-depth analysis. The regression results in Model 2 show that natural resource endowment is significantly positive at the 1% significance level, and the risk of expropriation decreases from −0.042 in Model 1 to −0.035 in model 2, indicating that Chinese firms’ preference for expropriation risk mainly motivated the natural resources in the host country. The risk of expropriation in model 3 does not pass the significance test, so to further investigate China’s expropriation risk preference for OFDI and the reasons, the host countries are divided into sub-samples according to different levels of economic development.

Government stability, efficiency, and democracy are not significant even after adding resource variables. The effect of the absence of conflict on China’s OFDI is significantly positive at the 10% significance level and is more robust; thus, conflict has a significant influence on China’s OFDI.

Overall, the effect of market motivation variables on China’s OFDI is not significant, and the extent of its effect must still be explored in the sub-sample study. For control variables, trade and GDP growth rates are not significant in the full-sample analysis and need to be further explored.

4.2.3. Sub-sample analysis of China’s political risk preference for OFDI

The analysis of China’s OFDI in the previous section shows China has a significant preference for expropriation risk in overseas direct investment, but the coefficient of expropriation risk in Model 3 in Table 4 is not significant; none of the market motivation variables is significant. This may not always be true that market factors do not influence China’s overseas investment. Therefore, the host countries are divided into sub-samples based on their development level: developed, emerging market and other developing countries. The main reason is that countries with relatively more developed economies have a more stable domestic political environment, and although the economic situation is not exactly proportional to the risk, it is representative to a certain extent.

Table 5 shows the system GMM results for sub-samples of developed, emerging market and other developing. The AR (2) values are all greater than 1%, indicating that the random disturbance terms do not have second-order autocorrelation, and the F-values indicate that the explanatory variables are jointly significant; therefore, the estimation results of the system GMM are relatively reasonable.

Table 5. Estimation results of the sub-sample of China’s political risk preference for OFDI

Variables	Developed Countries	Emerging Market Countries	Other developing countries
L.lngross	0.744***	0.724***	0.775***
	(24.03)	(5.05)	(21.58)
Expro	−0.132	−0.128**	−0.052**
	(−1.21)	(−2.30)	(−2.32)
Gove	−0.002	−0.035	0.032*
	(−0.20)	(−0.83)	(1.70)
Govs	0.001	−0.001	0.002
	(0.21)	(−0.29)	(0.49)
Pac	−0.039	0.000	−0.003
	(−0.91)	(−0.00)	(−0.73)
Cona	0.001	0.023	0.010
	(0.02)	(0.70)	(1.31)
Lngdp	−0.131*	0.748	0.104**
	(−1.68)	(0.13)	(2.52)
Lnpgdp	0.761***	−0.542	−0.254***
	(3.41)	(−0.35)	(−3.54)
Lnpop		−0.843	0.143
		(−0.23)	(0.74)
Growth	0.762	0.745*	0.032
	(0.64)	(2.00)	(0.65)
Lntrade	0.320***	0.132*	0.372***
	(3.24)	(1.57)	(3.23)
res1		0.420***	0.850***
		(4.30)	(18.48)
Rde	−0.037
	(−1.32)
Constant	−3.773***	0.267	−0.042
	(−3.24)	(0.52)	(−0.30)
Country effect	Yes	Yes	Yes
Time effect	Yes	Yes	Yes
Obs	246.000	260.000	942.000
F-test	719.699	2008.974	768.505
AR (1)	0.005	0.016	0.000
AR (2)	0.377	0.590	0.180

Note: Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1. Variable definitions and measurements are explained in Table 1.

China’s preference for political risk in developed countries is not significant, indicating that China’s overseas direct investment in developed countries is not sensitive to political risk. On one hand, it may be because the overall political environment in developed countries is relatively better and the overall difference in the data is smaller. On the other hand, it may be because Chinese companies believe too much in the political and market environment of developed countries. The regression results for the awareness of political risksare weak, and not significant. The market motivation variable is more significant, for instance, the GDP per capita is positive at 1% level of significance. However, the GDP variable is negative at the 10% significance level, which is not as expected. The other control variables, L.lngross is significantly positive and the total bilateral trade (lntrade) also passes the 1% significance test, indicating that bilateral trade has a driving and promoting effect on investment. The other control variables, L.lngross is significantly positive and total bilateral trade (lntrade) also passes the 1% significance test, indicating that bilateral trade has a driving and promoting effect on investment. Overall, China’s OFDI to developed countries does not show political risk preference and is not sensitive to political risk in developed countries, and the motivation of overseas direct investment tends to be market-seeking and relatively weak for technology-seeking motivation, which affirms Kamal et al. (2019) and Chen (2015).

The sub-sample regression of China’s political risk preference for emerging market countries shows that there is a significant preference for expropriation risk in China’s OFDI to emerging market countries, and the expropriation risk is significantly negative at the 5% significance level. However, other political risks, such as government efficiency, government stability, democracy and conflict are not significant. The market-seeking motives, such as GDP, GDP per capita and population size are all significant at 10%, and GDP growth is also significant at 10%, which to some extent reflects the market growth space of the host country. Thus, China’s OFDI to emerging market countries shows weak market-seeking motives. However, the natural resource endowment variable is positive and significant at 1%, indicating that China’s OFDI to emerging market countries exhibits strong resource-seeking motives although the market-seeking motives may be weak. The other control variables, L.lngross and total bilateral trade are also significant and positive, suggesting that Chinese overseas OFDI in emerging market countries is continuous and there may be a cluster effect of overseas FDI, while bilateral trade may have a pull effect on investment. Overall, China’s OFDI to emerging market countries has a strong propensity to expropriate risk preferences and exhibits strong resource-seeking motives and weak market-seeking motives, which agrees with Yang (2020).

The regression results of China’s political risk preferences and investment motives for other developing countries show that expropriation risk is significantly negative at the 5% significance level while government efficiency is significant at 10%, indicating that government efficiency increases to some extent to promote China’s OFDI, but government stability, degree of democracy, and conflicts are not significant. The results imply that China has a strong preference for expropriation risk for OFDI from other developing countries. For market motivation variables, GDP, is significantly positive at the 5% significance level, and GDP per capita is significantly negative, which may be because although some host countries have relatively high total GDP and large overall market size, their population size is also relatively large, so GDP per capita is lower. China’s OFDI to other developing countries is more inclined to countries with larger market size. The natural resource variable (res1) is significantly positive at 1% significance level, indicating a strong attraction for China’s OFDI. The other control variables also show strong trade traction, investment continuity, and investment agglomeration effect. Overall, China has a high-risk preference for expropriation for OFDI in other developing countries, is more sensitive to the abundance of natural resources in the host country, and has relatively weaker market-seeking motives. These conclusions agree with Yang (2020) and Quer et al. (2011).

4.2.4. Analysis of China’s political risk preferences for OFDI to African countries

The empirical results of China’s political risk preferences for OFDI to African countries are shown in Table 6. The difference between Models 1 and 2 is that South Africa is removed from Model 2. The reason is that compared to other African countries, South Africa is quite rich in arable land, minerals, and human resources. Also, due to the favorable location of the Indian Ocean-High Hope-Atlantic route, it makes it easy for people to ignore the political risks in the country; removing South Africa from the sample can solve endogenous selection problem to a certain extent. The AR(1) of both model 1 and model 2 are 0.001, and the AR (2) of both are greater than 0.01, indicating that the random error term does not have second-order autocorrelation, and the F-test is also significant, indicating that the estimation results of the systematic GMM are reasonable.

Table 6. Empirical results of China’s political risk preference for OFDI in African countries

Variables	Model1	Model2
L.lngross	0.643***	0.612***
	(9.34)	(8.33)
Expro	−0.056**	−0.059**
	(−2.28)	(−2.41)
Gove	0.056*	0.045*
	(1.69)	(1.79)
Govs	0.032***	0.043***
	(2.72)	(2.68)
Pac	0.061**	0.074***
	(−2.47)	(−2.73)
Cona	0.241	0.267
	(1.40)	(1.39)
Lngdp	−0.263	−0.253
	(−0.63)	(−0.74)
Lnpgdp	0.652	0.634
	(0.65)	(0.63)
Lnpop	0.887	0.865
	(1.05)	(1.38)
Growth	0.270	0.153
	(0.65)	(0.42)
Lntrade	0.181***	0.070***
	(3.71)	(3.53)
Constant	0.432	0.342
	(1.22)	(1.32)
Country effect	Yes	Yes
Time effect	Yes	Yes
Obs	453.000	358.000
F-test	1218.669	1386.210
AR (1)	0.001	0.001
AR (2)	0.451	0.560

Note: Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1. Variable definitions and measurements are explained in Table 1.

The regression results in Table 6 show that, the risk of expropriation is significantly negative at 5% significance level, and there is a strong preference for the risk of expropriation. Government effectiveness reflects the efficiency and service level of the host government. Most African countries are relatively poor and economic development is relatively slow, so the government does not invest much money in providing public services and improving efficiency, so the government attracts foreign investment to develop the economy and thus improve its efficiency and stabilize the regime. The increase in bilateral trade flows also shows a significant positive relationship with China’s investment in Africa, and after removing South Africa, the impact of bilateral trade flows on China’s “going out” becomes smaller because China is South Africa’s main trade partner. China has been South Africa’s largest trading partner for ten consecutive years.

4.3. Robustness tests

For further robustness checks, we analyze the political risk preferences of the sub-sample of Chinese OFDI in developed countries, emerging market countries, other developing countries and countries along the “One Belt, One Road” route through the mixed regression model and the alternate independent variable.

4.3.1. Mixed regression analysis

It is worth noting that further investigation was made to ascertain whether the political risk preference of Chinese OFDI was robust as this research performed a regression analysis using a mixed regression effects model with panel data. Table 7 shows the results of the mixed-regression estimates. The R-values are all greater than 0.90, and the fitting effect is relatively good. The mixed regression estimation results of China to developed countries, in which the political risk variables are all insignificant. China’s overseas direct investment in emerging market countries has obvious preference for expropriation risk and strong resource-seeking motivation, while market-seeking motivation is relatively weak. China also has strong preference for expropriation risk and resource-seeking motivation in other developing countries. For countries along the “One Belt, One Road”, China has a strong preference for expropriation risk and democracy risk, conflict avoidance, and resource-seeking motives. Overall, the estimation results of the mixed regression model are more consistent with the systematic GMM estimation, and the results are robust.

Table 7. Estimation results of the mixed regression model

Variable	Developed Countries	Emerging Market Countries	Other developing countries	Countries along the “One Belt, One Road”
L.lngross	0.741***	0.763***	0.846***	0.783***
	(23.70)	(24.03)	(32.73)	(23.99)
Expro	−0.050	−0.065**	−0.063***	−0.074***
	(−0.62)	(−2.71)	(−3.42)	(−2.63)
Gove	0.002	−0.053*	0.045	0.039
	(0.04)	(−1.93)	(2.38)	(0.52)
Govs	−0.023	−0.021	0.018	0.015
	(−0.45)	(−1.20)	(0.32)	(1.37)
Pac	−0.006	−0.007	−0.013	−0.034***
	(−0.46)	(−0.26)	(−0.78)	(−2.43)
Cona	0.001	0.051	0.016	0.047*
	(0.01)	(1.74)	(1.43)	(1.87)
Lngdp	−0.535	5.724	0.067**	−0.436
	(−1.56)	(0.45)	(2.11)	(−0.63)
Lnpgdp	0.720***	−5.350	−0.164***	0.565
	(2.96)	(−0.87)	(−4.65)	(0.49)
Lnpop		−5.471		0.598
		(−0.65)		(0.76)
Growth	0.320	0.760	0.225	0.138
	(0.39)	(1.55)	(1.16)	(0.56)
Lntrade	0.457***	0.198***	0.564***	0.291***
	(3.47)	(3.72)	(3.64)	(3.40)
Res		0.304*	0.346***	0.105**
		(1.78)	(3.74)	(2.35)
Rde	0.000
	(−0.77)
Constant	−3.521**	0.079	0.093	−0.650*
	(−2.53)	(0.23)	(0.35)	(−1.86)
Country effect	Yes	Yes	Yes	Yes
Time effect	Yes	Yes	Yes	Yes
Obs	264.000	260.000	942.000	518.000
F-test	126.631	472.935	851.472	643.709
R-squared	0.943	0.922	0.935	0.932

Note: Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1. Variable definitions and measurements are explained in Table 1.

4.3.2. Alternate independent variable

The measure of expropriation risk in this research is further expanded to include the strength of contract enforcement and delayed payment. Table 8 shows that their AR (2) are all greater than 0.1 and the F-values are more significant, so the systematic GMM for classification 3 is also more reasonable. In general, the regression results are consistent with the previous estimates, i.e., Chinese OFDI is not sensitive to political risk in developed countries and has strong market-seeking motives; regardless, it has strong risk preference for expropriation and resource-seeking motives for emerging market countries, other developing countries and countries along the “One Belt, One Road”. Therefore, the estimation results in the previous section are more robust.

Table 8. Estimation results of replacing independent variables

Variable	Developed Countries	Emerging Market Countries	Other developing countries	Countries along the “One Belt, One Road”
L.lngross	0.752***	0.753***	0.843***	0.797***
	(18.47)	(16.49)	(24.16)	(20.02)
Expro	−0.014	−0.021	−0.028*	−0.018***
	(−0.40)	(−1.43)	(−1.72)	(−2.82)
Gove	0.005	−0.008	0.003	0.012
	(0.36)	(−0.47)	(0.35)	(0.42)
Govs	−0.021	0.001	0.001	0.004
	(−1.43)	(0.30)	(0.32)	(1.21)
Pac	−0.034	−0.024	−0.016	−0.026**
	(−0.79)	(−0.35)	(−0.24)	(−2.28)
Cona	−0.012	0.021	0.029	0.049
	(−0.06)	(0.34)	(1.30)	(1.86)
Lngdp	−0.181**	1.427	0.145**	−0.456
	(−2.73)	(0.36)	(2.62)	(−1.43)
Lnpgdp	0.976***	−1.346	−0.125***	0.764
	(3.70)	(0.25)	(−3.331)	(1.43)
Lnpop		−1.530		0.643
		(−0.36)		(1.42)
Growth	0.634	0.665	0.132	0.143
	(0.83)	(1.43)	(0.71)	(0.68)
Lntrade	0.498***	0.137*	0.042	0.173**
	(3.83)	(1.84)	(0.61)	(2.32)
Res		0.436**	0.367***	0.074
		(2.42)	(3.54)	(1.65)
Rde	−0.062*
	(−2.31)
Constant	−3.321***	0.075	−0.192	−0.783***
	(−3.57)	(0.32)	(−0.56)	(−2.79)
Country effect	Yes	Yes	Yes	Yes
Time effect	Yes	Yes	Yes	Yes
Obs	246.000	246.000	942.000	518.000
F-test	774.419	1086.271	778.186	2022.145
AR (1)	0.004	0.03	0.000	0.004
AR (2)	0.345	0.710	0.162	0.786

Note: Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1. Variable definitions and measurements are explained in Table 1.

5. Conclusion

This study examined the impact of political risk on China’s OFDI and what motivates their preferred investment destinations. The study analyzed the impact of political risk on OFDI of other countries to enhance the comparison of China and the world’s preferences for political risk in overseas FDI. In terms of political risk, the study revealed that OFDI in countries around the world tends to favor countries and regions with lower expropriation risk and conflict risk, while the effects of government efficiency, government stability, and the degree of democracy on foreign direct investment are not significant. The result is consistent with the conclusion from the general equilibrium model of expropriation risk that the improvement of other political risks does not necessarily attract foreign investment in the presence of expropriation risk, or even if the host country has a high level of expropriation risk, the improvement of other political risks, leads to outflow of foreign investment. Also, the OFDI in countries around the world exhibits more market-seeking motives in terms of investment.

There are significant differences between the results of OFDI in China and those other countries in the world. For political risk, China’s OFDI tends to favor countries and regions with higher expropriation risk, and there is a significant preference for expropriation risk. This is not consistent with expectations and contradicts the conclusions drawn from the previous general equilibrium model of expropriation risk. In terms of investment motives, China’s OFDI exhibits strong resource-seeking motives and weak market demand motives.

Based on the different levels of economic development and the presence of natural resources in the host country, China may have different political risk preferences and investment motives. China’s OFDI to developed countries is insensitive to political risk. On the other hand, China’s OFDI to emerging market countries is sensitive to political risk, while government efficiency, government stability, degree of democracy, and conflict do not have significant effects on Chinese OFDI. In terms of investment motives, China exhibits strong resource-seeking motives and weak market-seeking motives in emerging market countries. For instance, for every 1% increase in the natural resources of the host country, China’s OFDI will increase by about 0.4% when other factors are held constant. Although neither GDP nor GDP per capita is significant, the growth rate of GDP is significantly positive. Emerging market countries such as India, South Africa, Brazil and Chile are developing countries, but they are in the golden stage of economic development and take-off, and their future market space and prospect are attractive to Chinese enterprises. The results also show that China’s OFDI to other developing countries is sensitive to political risk. Expropriation risk and government efficiency significantly affects China’s OFDI.

The observations made in this study has important implications. Policy makers governments of countries around the world should try to contain political risks to the barest minimum since the study results show that OFDI in countries around the world tends to favor countries and regions with lower expropriation risk and conflict risk. Contrary to expectation, China’s OFDI increases with expropriation risk and this calls for countries with experiencing expropriation risk to thoroughly examine why this risk is rather encouraging China’s OFDI into their economies. Different regression results based on countries development levels reveal that China’s OFDI in emerging countries shows resource-seeking motive than market-seeking motive. Thus, emerging countries endowed with natural resources could use formation of common markets to narrow the disparities in FDI motives between them and developed countries since persistent disparity in FDI motives could widen the disparities in economic development.

Taken together, the results in this study provide scope for further research. First, this study focused on country-level data and future research can employ firm level or industry level data to examine the impact of political risk on different firms and different sectors such as services, resource extraction, and manufacturing and most importantly how the impacts vary across sectors. Second, given the advent of the “Belt and Road” and the rejuvenation of leftists governments in some regions of the world where Chinese firms have momentous presence, future studies could also analyze how Chinese investors strategize during change of government (in a democratic regime) but with a divergent ideology, especially countries with radical leftist tendencies. Finally, although the models in this study accounts for temporary effects, no special emphasis is placed on the 2008/2009 financial crises. Thus, future studies may specifically examine the impact of the financial crises by evaluating political risk and FDI nexus before and after the crises

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors received no direct funding for this research.

Notes on contributors

Fidelis Ayangbah

Fidelis Ayangbah is a Ph.D. candidate at Changsha University of Science and Technology. His rich professional experience cut across operations and Finance. His main research interests include Strategic Management, Trade Relations, and International Finance.

Bismark Addai

Bismark Addai is a researcher at the School of Economics and Management, Changsha University of Science and Technology in China. His current research interests span financial economics, environmental economics, agricultural economics, corporate finance, and corporate governance.

Adjei Gyamfi Gyimah

Adjei Gyamfi Gyimah is an investment expert and a consultant and currently works at Plan international. He has a lot of teaching and professional experience in Finance.