Economic Policy Uncertainty and Foreign Direct Investment: A Bilateral Perspective on Push and Consistency Effects

Abstract

Against the backdrop of unprecedented global FDI volatility—with flows declining 34.7% in 2020 and a further 12% in 2022—and China experiencing its first sustained capital outflow since reform, with foreign enterprises withdrawing over USD 160 billion in the first three quarters of 2023, understanding the complex mechanisms through which EPU affects international investment has become critically important. Existing research predominantly examines unilateral EPU effects while neglecting the bilateral dynamics that characterize modern interconnected economies, creating a significant gap in explaining recent FDI pattern shifts. This study systematically examines the differential impact mechanisms of EPU on China’s FDI inflows using panel data from 20 countries spanning 2005–2023, employing FE models and GMM methods. The research reveals that policy uncertainty affects international investment through two mechanisms: first, a “push effect” whereby relatively higher EPU in home countries drives FDI flows to China (β = 0.002, p < 0.001); second, a “consistency effect” where differences in policy environments between home countries and China impede FDI flows (β = −0.004, p < 0.001), with the latter effect being stronger. Moderating effects analysis demonstrates that institutional quality and bilateral political relations exert complex non-linear moderating effects on the EPU–FDI relationship. Heterogeneity tests reveal that when China’s EPU is relatively low, the negative impact of policy uncertainty is significantly weakened. This study extends real options theory and provides empirical evidence for the dual mechanisms of the EPU–FDI relationship, emphasizing that policy coordination is more important than relative policy advantages for international investment decisions. The findings provide theoretical foundations and practical guidance for policymakers to optimize international investment environments and strengthen policy coordination.

1. Introduction

Foreign Direct Investment (FDI) has long been recognized as a key driver of global economic growth. It facilitates technology transfer, job creation, and productivity enhancement. For recipient countries, FDI brings capital, skills, and opportunities for international market access. For investing countries, it provides new business opportunities, cost advantages, and risk diversification. In developing economies such as China, FDI has played an important role in supporting industrial growth and connecting to global value chains.

However, in recent years, focusing on recent international investment trends and prospects, as shown in Figure 1, from a scale perspective, the COVID-19 pandemic in 2020 caused a sharp decline in global FDI, and post-pandemic flows have remained volatile. Specifically, in 2020, global FDI fell to its lowest level since 2005, dropping below USD 1 trillion with a decline of 34.7%. In 2022, it declined again by 12%. Shocks such as geopolitical conflicts, multiple crises, and climate change have brought enormous uncertainty and fragility to the global economy, dealing a heavy blow to the confidence of cross-border investors.

China’s FDI inflows have generally aligned with global FDI trends. Figure 2 shows that foreign investors have remained optimistic about China’s economic development prospects since the 1990s. In 2008, China attracted USD 111.2 billion in foreign investment, nearly tripling compared to 2000, ranking third globally and first among developing countries. However, since 2022, this situation has completely reversed. With China’s economic growth slowing significantly and escalating U.S.–China tensions, large numbers of foreign companies are withdrawing from the Chinese market. According to Chinese statistics, as of the end of September 2023, foreign enterprises have withdrawn a total of over USD 160 billion from China for six consecutive quarters, representing an unprecedented large-scale capital outflow in history and indicating a significant weakening of China’s attractiveness to foreign investment. This phenomenon has drawn attention from policymakers and economists.

A major factor related to this trend is the rise in EPU, as shown in Figure 3 and Figure 4. Figure 3 depicts the Global EPU Index based on the quantitative methodology proposed by Baker et al. (2016). It can be observed that the GEPU index has risen significantly since 2008 and has remained persistently at high levels, climbing again after the 2008 U.S.–China trade disputes and the outbreak of COVID-19 in 2020. In recent years, frequent policy adjustments and major events have all been reflected through increases in EPU. EPU refers to the unpredictability surrounding government policies, regulations, or fiscal decisions that affect economic activity.

Accompanied by the severe global economic situation, China’s EPU is also rising. Figure 4 depicts the trend of China’s EPU Index from 1997 to 2023. In 2008, to respond to the impact of the global financial crisis, the government launched a series of fiscal and monetary policies. In 2011, the “Twelfth Five-Year Plan” introduced a series of reform measures, reaching the highest point in twenty years at 165.74. As China undergoes reforms, faces trade tensions, and experiences regulatory changes, these uncertainties may have substantial impacts on investor confidence and decision-making.

Currently, compared to other economic factors, people’s understanding of the impact of policy environment fluctuations on economic activity is limited. Previous research has mainly focused on long-term determinants, while short-term factors have received insufficient attention. Long-term determinants refer to factors that cause changes in capital flows due to changes in the macroeconomic fundamentals of the economy itself, such as market size (Kolstad & Wiig, 2012), market openness (Nagano, 2013), natural resource endowments (Buckley et al., 2007), and infrastructure development (Asiedu, 2006). However, these traditional determinants represent relatively stable long-term characteristics that change gradually over several years, while EPU, as a more dynamic short-term factor, may experience significant fluctuations within months or quarters.

Although researchers are increasingly studying EPU and its relationship with FDI, several key questions remain unanswered. Most existing studies focus on EPU in individual countries without examining how uncertainty in home and host countries interact to shape investment flows. In an interconnected global economy, this bilateral dynamic relationship may play a crucial role.

However, existing research often examines host country and home country EPU independently (Y.-F. Chen & Funke, 2011; Pennings & Sleuwaegen, 2004), without delving into their interactions. If we view home country EPU as a “push force” for its investors’ FDI and host country EPU as a “resistance force” for foreign investors in that country, then the final investment outcome will depend on the resultant direction after these two forces offset each other. Therefore, the impact of the EPU differential between host countries and investing countries (home countries)—the relativity of economic policy environments—on FDI inflows requires further exploration.

Furthermore, based on institutional proximity theory, firms tend to invest in countries with institutional frameworks similar to their own (Habib & Zurawicki, 2002). Therefore, it is crucial to further study the impact of the absolute values of both home country and host country EPU on FDI inflows. In an era when EPU is high across countries globally, choosing an environment similar to one’s own implies greater familiarity and reduced risk. We use the term “consistency effect” rather than “similarity” or “proximity” because it emphasizes the dynamic nature of policy alignment and the costs associated with operating across different policy frameworks, rather than static institutional characteristics. The concept of “consistency” captures both the challenge of navigating policy differences and the operational friction that arises when investors must adapt to varying policy environments between home and host countries.

Although the importance of institutional factors is increasingly recognized, Chinh and Thi Minh Hue (2025) found that while “improvements in institutional quality and bilateral political relations have positive effects on FDI, the interaction between these two factors may produce opposite effects,” indicating that the moderating mechanisms of the joint effects of political stability and bilateral political relations remain insufficiently understood (Bommadevara & Sakharkar, 2021). Studying moderating variables can provide a deeper understanding of the generalizability of causal relationships under specific circumstances (MacKinnon, 2011). This study will introduce bilateral political relations and institutional quality as moderating variables, which are expected to mitigate the adverse effects of EPU on FDI. In this context, moderating variables may play a substitutive role; for example, as the levels of bilateral political relations and institutional quality improve, the negative impact of host country EPU on its FDI inflows will weaken, and may even eliminate or reverse the negative effects. In today’s era, when many countries still face high EPU, these factors can provide policymakers with additional methods to attract investment.

Therefore, we need to conduct further systematic and comprehensive analysis to explore the impact of EPU on China’s FDI inflows. Taking China as an example, this paper addresses the following specific research questions: (1) How does the difference between home country and China’s EPU affect China’s FDI inflows? (2) How does the absolute value of the difference between home country and China’s EPU affect China’s FDI inflows? (3) How do institutional quality and bilateral political relations moderate the effects of the difference between home country and China’s EPU and its absolute value on China’s FDI inflows? The specific objectives of this study are as follows: (1) To determine the impact of the difference between home country and China’s EPU on China’s FDI inflows. (2) To determine the impact of the absolute value of the difference between home country and China’s EPU on China’s FDI inflows. (3) To determine the moderating effects of institutional quality and bilateral political relations on the relationship between the difference between home country and China’s EPU (and its absolute value) and China’s FDI inflows.

This study contributes to theoretical and empirical evidence in the following aspects. First, it provides insights for addressing China’s persistent foreign capital withdrawal and explaining the decline in China’s FDI. It enriches the application of real options theory, institutional escape theory, and institutional proximity theory in the Chinese context. Second, against the backdrop of persistently high global and Chinese EPU, exploring the moderating effects of institutional quality and bilateral political relations can provide insights for mitigating this phenomenon. Third, this study can offer some enlightenment for national policymakers.

The remainder of the paper is organized as follows. Section 2 introduces the theoretical underpinning and previous studies. Section 3 describes the data sources and empirical models. Section 4 includes empirical results and robustness tests. Section 5 is findings and discussion. Section 6 presents conclusions and implications.

2. Literature Review

2.1. Theoretical Underpinning

The unique contribution of real options theory lies in providing an overall theoretical foundation upon which international business scholars can study how MNCs make investment decisions in uncertain environments and how they adjust their investment strategies in response to new information in the environment (Belderbos & Zou, 2007; J. Li & Li, 2010; Tong et al., 2008). Furthermore, real options theory emphasizes the importance of dynamic perspectives in FDI theory. For example, Buckley and Casson (1998) emphasized the need to overcome the static nature of previous models and apply real options analysis to FDI theory. Real options theory allows three important characteristics related to FDI—namely, irreversibility, flexibility, and uncertainty—to be incorporated into models and modeled. Uncertainty and volatility are respectively reflected in the magnitude and frequency of profit flows, which affect the value of investment projects.

Many empirical studies have applied real options theory to analyze the impact of EPU on investment. Dixit and Pindyck (1994) demonstrated that the combination of uncertainty and irreversibility reduces firms’ incentives to invest immediately but increases incentives to wait. Therefore, high uncertainty hinders investment, while low uncertainty increases firms’ investment incentives. Thus, the relationship between uncertainty and investment should be negative. Consistent with this theory, Leahy and Whited (1995) and Bulan (2005) used firm-specific earnings volatility to represent uncertainty and found that such volatility impedes capital investment.

From the above empirical studies, it is also clear that real options theory can provide a solid theoretical foundation for studying the relationship between EPU and FDI. However, this theory is limited to the host country perspective, neglecting the home country perspective, creating a significant theoretical gap. For example, Q. Nguyen et al. (2018), by examining outward investment data from eight Southeast Asian countries, found that when the home country policy environment is unstable, companies tend to accelerate overseas asset transfers. They prefer to invest in countries or regions with more stable policy environments to expand their outward investment. This is similar to institutional escape theory, where enterprises seek to mitigate the negative effects of domestic institutional deficiencies and regulatory constraints, thereby reducing investment costs and driving firms to seek better overseas development opportunities.

Meanwhile, these studies typically examine host country and home country EPU independently, without deeply exploring their interactions. If we view home country EPU as a “push force” for its investors’ FDI and host country EPU as a “resistance force” for foreign investors in that country, then the final investment outcome will depend on the resultant direction after these two forces offset each other. Therefore, the impact of the EPU differential between host countries and investor countries (home countries)—the relativity of economic policy environments—on FDI inflows requires further exploration.

Furthermore, based on institutional proximity theory, firms tend to invest in countries with institutional frameworks similar to their own (Habib & Zurawicki, 2002). Therefore, it is necessary to further study the impact of the absolute values of both home country and host country EPU on FDI inflows. When EPU is high across countries globally, choosing an environment similar to one’s own implies greater familiarity and reduced risk.

In summary, beyond the host country perspective emphasized by real options theory, the home country perspective should also be incorporated into research. The interaction between these two perspectives, particularly the differential and absolute values, can provide deeper insights into the driving factors of FDI decision-making. Addressing the three new dimensions of measuring EPU will help fill the theoretical gap in real options theory regarding the impact of uncertainty.

2.2. Previous Study

2.2.1. The Impact of EPU on FDI

Related research results indicate that the impact of home country and host country EPU on FDI presents three types of outcomes: negative, positive, and insignificant (Oi, 1961; Vo & Le, 2017; Zhu et al., 2019).

First, from the host country perspective, some scholars believe that high EPU in host countries reduces FDI inflows into those countries. This negative impact can be explained through three mechanisms. First, high levels of host country EPU may reduce investor confidence, as countries with high EPU are less attractive to investors, thereby hindering FDI inflows (Shahzad & Al-Swidi, 2013). Second, low EPU means a more stable economic environment and less risk. For example, U.S. multinational corporations’ investments in Mexico depend on Mexican government policies, such as tax agreements, labor market regulations, and capital controls (Zaheer & Mosakowski, 1997). Third, foreign enterprises are often more sensitive to host country policies, and uncertainty may lead them to adopt a wait-and-see attitude. As Bhattacharya et al. (2007) pointed out, in the case of disputes, host country courts often favor domestic enterprises, and foreign investors receive less protection from host country legal and political institutions (Aizenman & Spiegel, 2006; Dixit et al., 2011; Julio & Yook, 2016). Furthermore, research has found that the negative impact of EPU on FDI is more significant in developing countries (Avom et al., 2020; Canh et al., 2020), as well as in East Asian countries (Choi et al., 2021; Q. Nguyen et al., 2018).

There are also some researchers supporting growth option theory who believe that increases in host country EPU will have positive effects on IFDI (Hartman, 1972; Oi, 1961; Vo & Le, 2017). Enterprises can identify and seize investment opportunities in uncertain economic policy environments, obtaining profits through resource integration (Knight, 1921). Investors often worry about potential opportunities and returns, investing more when uncertainty increases.

In addition, some scholars believe that EPU has no significant impact on FDI inflows. Related research indicates that the host country’s economic policy environment has no significant impact on IFDI (Forbes & Warnock, 2012), a view confirmed in some studies on China, where the impact of EPU is not significant due to mitigating factors such as government subsidies or institutional buffers (Borojo et al., 2023; Yu et al., 2024).

From the home country perspective, similarly, research on the impact of home country EPU on FDI also presents the above three types of results (Hsieh et al., 2019; Jardet et al., 2023; Su et al., 2022). Some scholars believe that increases in home country EPU have adverse effects on host country FDI inflows (Canh et al., 2020; Hsieh et al., 2019; Jardet et al., 2023; Su et al., 2022). Other scholars believe that growth in home country EPU benefits FDI flows to host countries. If a developing country has political instability and erratic policy changes, enterprises often hold cash and seek suitable investment opportunities in other countries, leading to frequent capital flight (Lensink et al., 2000).

However, these existing studies often examine host country and home country EPU independently (Y.-F. Chen & Funke, 2011; Pennings & Sleuwaegen, 2004), without delving into their interactions. Therefore, few studies combine home country and host country EPU to examine their interactions. If we view home country EPU as a “push force” for its investors’ FDI and host country EPU as a “resistance force” for foreign investors in that country, then the final investment outcome will depend on the resultant direction after these two forces offset each other. Therefore, the impact of the EPU differential between host countries and investing countries (home countries)—the relativity of economic policy environments—on FDI inflows requires further exploration.

Specific to China’s context, recent empirical studies have examined EPU–FDI relationships with varying findings. Zhou et al. (2023) found that increases in China’s EPU lead to decreases in FDI inflows, with this effect being more pronounced in cities with better market fundamentals and after the 2008 Global Financial Crisis (C. P. Nguyen & Lee, 2021). The IMF (2024) analysis suggests that China’s recent downward trend in FDI inflows primarily reflects higher economic policy uncertainty, geopolitical risk, and weak future growth prospects (C. P. Nguyen & Lee, 2021). However, some studies have found that domestic policy uncertainty can drive Chinese firms’ outward FDI, as companies seek more stable investment environments abroad when facing domestic uncertainty (Chakradhar & Gupta, 2024).

From a bilateral perspective, Choi et al. (2021) examined FDI flows using OECD data and found that bilateral FDI is associated with policy uncertainty in both source and host countries (Kapuria & Singh, 2021), supporting the notion that relative uncertainty levels between countries influence international investment decisions in China’s context.

Real options theory suggests that host country EPU hinders FDI, while institutional escape theory suggests that home country EPU promotes outward investment. The balance of these forces depends on the relative uncertainty levels of each country (Gregoriou et al., 2021). For example, when host country EPU is lower than home country EPU, enterprises are more likely to engage in overseas investment (Q. Nguyen et al., 2018). Based on these perspectives, we propose:

H1. 

The difference between home country and China’s EPU has a positive impact on China’s FDI inflows.

Furthermore, institutional proximity theory emphasizes that enterprises prefer to invest in countries with similar institutional environments, as reduced institutional distance lowers integration costs and risks (Kogut & Singh, 1988). Therefore, we propose the following:

H2. 

The absolute value of the difference between home country and China’s EPU has a negative impact on China’s FDI inflows.

2.2.2. Moderators of EPU–FDI Relationship

Moderating factors, such as institutional quality and bilateral political relations, can influence the dynamic relationship between EPU and FDI:

• Institutional quality: Strong institutions attract FDI by providing stability and reducing risks (Busse & Hefeker, 2007; Feng & Mu, 2010; Warner & Zawahri, 2012; Wu & Shao, 2023). Conversely, countries with less mature institutional environments may still attract investment through lenient trade policies (Ahlquist & Prakash, 2010; Shirodkar & Mohr, 2015).
• Bilateral political relations: Positive relations between home countries and host countries promote FDI by reducing operational costs and mitigating risks (Desbordes & Vicard, 2009; Song et al., 2020). For example, diplomatic relations can buffer the host country policy uncertainty faced by enterprises (Yi et al., 2024).

The moderating effects of institutional quality and bilateral political relations have been empirically examined across various country contexts. Regarding institutional quality, research shows that its impact on FDI varies by country development level, with institutional quality having more significant positive effects on FDI in developed countries (Peres et al., 2018; Song et al., 2020), while in lower-middle income countries, excessively high rule of law and voice and accountability may actually inhibit FDI inflows (Abaidoo & Agyapong, 2022).

Concerning bilateral political relations, studies based on Sino–Japanese relations demonstrate that bilateral political relations significantly affect both FDI inflows and outflows, with leaders’ visits significantly increasing FDI, though diplomatic conflicts have relatively smaller impacts (Whitten et al., 2020). Additionally, research finds that the effectiveness of bilateral investment treaties depends on the quality of political relations between signatory countries, with treaties having stronger positive effects on FDI between countries with tense relationships (Zhang & Hao, 2018). Chinese empirical studies show that bilateral political relations significantly amplify the positive effects of host countries’ financial development on FDI inflows to China. Therefore, we propose the following:

H3. 

Good bilateral political relations and high institutional quality positively moderate the relationship between the difference and absolute value of home country and China’s EPU and FDI inflows.

3. Data and Methodology

3.1. Data Sources and Variable Definitions

To examine how the interaction between home country and China’s EPU affects China’s FDI inflows, we compiled a comprehensive panel dataset covering 20 countries (Appendix A), spanning from 2005 to 2023, totaling 380 observations.

China’s FDI inflow data comes from the International Monetary Fund (IMF), while EPU indices are sourced from the EPU database developed by Baker et al. (2016). Supplementary macroeconomic indicators come from the World Bank’s World Development Indicators (WDI) and the China Statistical Yearbook.

Table 1. Variable Description.

Variable	Variable Meaning	Variable Description	Data Source
FDI	FDI inflows to China	China’s FDI flows	IMF
EPU	China’s or other countries’ EPU	EPU Index jointly constructed by Baker et al. (2016) to measure China’s EPU, which is mainly based on articles containing at least one keyword of “China, economy, uncertainty” in the daily news of the South China Morning Post in Hong Kong, China, and related policy terms such as “currency, trade, central bank, government” in the article, and finally divides the monthly articles containing the above words by the total number of articles and standardizes them to obtain a monthly index. This article uses arithmetic averaging to convert monthly data into annual data: ${\mathrm{EPU}}_{\mathrm{t}}$ = ${\sum }_{\mathrm{m}=1}^{12}{\mathrm{EPU}}_{\mathrm{m}}$/12.	EPU website
D_EPU	Difference between home country and China’s EPU (HEPU-CEPU)	The difference variable captures the directional effect of policy uncertainty between investing countries and China.	EPU website
AD_EPU	Absolute difference value between home country and China’s EPU(|HEPU-CEPU|)	The absolute difference variable measures the magnitude of differences in policy environment stability between home countries and China, calculated as the absolute value of the HEPU minus CEPU.	EPU website
Market	Economies of scale	To measure the market size, we use China’s GDP. This variable is used to represent a country’s market potential or absorptive capacity. This variable was used by Buckley et al. (2007), and they found a significant positive relationship.	World Bank
Growth	China’s level of economic development	Higher economic growth in a country is often associated with better quality and performance of the local market, making it more attractive to foreign investors. Ramasamy et al. (2012) reveal that FDI inflows in China are disproportionately directed towards provinces with stronger economic performance. To measure a city’s or a country‘s economic growth, we utilize the GDP growth rate.	World Bank
NR	The natural resource endowment of the host country	Natural resources are the share of mineral and metal resources exported by host countries in total commodity exports. This variable represents the natural resource abundance in the host country. This variable was used by Buckley et al. (2007), and they found a positive relationship, but insignificant.	World Bank
Tec	The strategic resource endowment of the host country (Total Patents Registered)	This variable represents strategic asset seeking. The data are the number of patents registered by residents and non-residents in each country for exclusive rights for an invention of a product or process that provides a new way of doing something or offers a new technical solution to a problem. The total patent application will not be standardized by the population number, following Buckley et al. (2007) and X. Li and Li (2024).	World Bank
Inflation	Inflation rates in the host country	Inflation refers to the annual percentage of the consumer price index, which reflects the annual percentage change in the cost to the average consumer of acquiring a basket of goods and services. Kolstad and Wiig (2012) found an insignificant effect of inflation on COFDI. However, Buckley et al. (2007) found a significant but different sign from their hypothesis.	World Bank
Institutional Quality (WGI)	Institutional quality in the host country	Institutional quality refers to the effectiveness and reliability of a country’s governance framework, which encompasses the rule of law, regulatory effectiveness, and control of corruption. We employ the Worldwide Governance Indicators (WGI) developed by Kaufmann et al. (2011), which aggregate six dimensions of governance: Voice and Accountability, Political Stability and Absence of Violence, Government Effectiveness, Regulatory Quality, Rule of Law, and Control of Corruption. Each indicator ranges from −2.5 (weak) to 2.5 (strong governance performance). Previous studies have demonstrated that institutional quality plays a crucial role in attracting FDI (Busse & Hefeker, 2007; Daude & Stein, 2007; Globerman & Shapiro, 2002).	World Bank
Bilateral Political Relations (Agreement)	Bilateral political relations between the home and host country	Bilateral political relations refer to the diplomatic and political alignment between the investing country and the host country, which can significantly influence investment decisions and flows. We measure bilateral political relations using UN voting alignment, which captures the similarity of voting patterns between two countries in the United Nations General Assembly. This indicator is calculated as the correlation coefficient of voting positions on issues where both countries cast votes, ranging from −1 (complete disagreement) to +1 (perfect alignment). Higher values indicate stronger political ties and diplomatic cooperation. The extensive literature has demonstrated the importance of political relations for FDI flows. Gartzke and Li (2003) found that political affinity, measured through UN voting alignment, positively affects bilateral investment flows.	https://dataverse.harvard.edu/ (accessed on 20 August 2025).

presents the key variables used in the study along with their definitions and sources. The dependent variable is the logarithmic form of China’s FDI inflows. The key independent variables are the difference between home country and China’s EPU (HEPU—CEPU) and the absolute value of the difference between home country and China’s EPU (|HEPU—CEPU|). Control variables include economic growth, market size, inflation, resource endowments, and technology level.

This extensive set of variables enables us to control for country-specific macroeconomic and structural differences that may affect FDI inflows.

3.2. Descriptive Statistics and Multicollinearity Diagnostics

Table 2. Summary Statistic.

Var Name	Obs	Mean	SD	Min	Median	Max
lnEPU	379	4.858	0.519	3.793	4.834	6.336
lnFDI	379	9.610	2.775	1.609	10.175	13.793
D_EPU	379	−23.788	104.563	−318.990	3.516	413.451
AD_EPU	379	74.326	77.212	0.0002	39.223	413.451
lnWGI	379	−0.465	0.120	−0.622	−0.476	−0.274
Agreement	379	0.676	0.154	0.144	0.685	0.944
Market	379	2.163	0.624	0.841	2.368	2.908
Growth	379	8.089	2.874	2.340	7.779	14.150
Inflation	379	2.367	1.607	−0.728	2.000	5.925
Tec	379	13.496	0.773	12.063	13.741	14.333
Nature	379	1.390	0.270	1.122	1.322	2.150

Notes: This table presents descriptive statistics for key variables in the econometric model, including the number of observations (Obs), mean, standard deviation (SD), minimum (Min), median (Med), and maximum (Max).

presents the descriptive statistics of the main variables in this paper. The logarithmic value of FDI flowing into China (lnFDI) has a mean of 9.610 and a standard deviation of 2.775, indicating significant differences in investment from different countries to China. The mean of the EPU index (lnEPU) for various countries is 4.858 with a standard deviation of 0.519, indicating moderate variability in policy uncertainty levels among sample countries. Among the control variables, China’s economic growth rate (Growth) shows considerable variability (standard deviation of 2.874), reflecting temporal differences in China’s economic performance during the sample period. China’s technology level (Tec) presents relatively high values, with a mean of 13.496 and a standard deviation of 0.773, indicating that China’s technological development level is relatively advanced and stable. China’s inflation rate (Inflation) has a mean of 2.367 and a standard deviation of 1.607, showing moderate price fluctuations. Among the moderating variables, China’s World Governance Indicators (lnWGI) has a mean of −0.465, reflecting China’s governance quality level. China’s bilateral political relations with other countries (Agreement) has a mean of 0.676, indicating that China’s political relations with sample countries are generally positive. All variables have consistent numbers of observations (379), ensuring data completeness and reliability of empirical analysis.

Before empirical analysis, to avoid multicollinearity problems among variables, this paper uses the Variance Inflation Factor (VIF) for assessment. Generally, when a variable’s variance inflation factor (VIF) is greater than 10, it indicates multicollinearity problems.

Table 3. Multicollinearity Diagnostics.

Variable	D_EPU_VIF	AD_EPU_VIF
Market	5.320	5.420
Tec	4.200	4.300
Growth	4.080	4.070
Nature	2.260	2.180
Inflation	1.180	1.170
Independent Variables (D_EPU & AD_EPU)	1.260	1.230
Mean VIF	3.050	3.062

Notes: VIF > 5 indicates potential multicollinearity problems. VIF > 10 indicates serious multicollinearity problems.

presents the variance inflation factors (VIF) for all variables in both models.

The VIF analysis reveals that all variables demonstrate acceptable levels of multicollinearity. In the D_EPU_VIF model, the VIF values range from 1.180 (Inflation) to 5.320 (Market), with Technology showing a VIF of 4.200, Growth at 4.080, Nature at 2.260, and Independent Variables at 1.260. The mean VIF for this model is 3.050. Similarly, in the AD_EPU_VIF model, VIF values range from 1.170 (Inflation) to 5.420 (Market), with Technology at 4.300, Growth at 4.070, Nature at 2.180, and Independent Variables at 1.230. The mean VIF for this model is 3.062.

Importantly, all VIF values tested in this paper are well below the critical threshold of 10, with the highest value being 5.420 for the Market variable in the AD_EPU_VIF model and all other variables showing considerably lower values. The VIF values are also below the more conservative threshold of five, indicating no collinearity problems among variables. These results confirm that all variables in both models are free from multicollinearity issues, thereby validating the reliability of subsequent regression analyses.

3.3. The Empirical Strategy

3.3.1. Model Selection and Justification

This section lists the Fixed Effects (FE) model and Generalized Method of Moments (GMM) model, and finalizes their application through comparison with other models.

(i)

Fixed Effects (FE) Model

The FE model is highly effective in addressing unobserved heterogeneity across countries and over time. By eliminating the influence of time-invariant characteristics, it isolates the effects of variables that change and are related to FDI. This is particularly useful when unmeasured country-specific factors (such as culture or geography) may bias results.

The FE model is a robust approach for analyzing the relationship between EPU and FDI, particularly when dealing with panel data (Gujarati & Porter, 2009). The advantages of the FE model are primarily reflected in its ability to handle unobserved heterogeneity, its focus on within-entity variation, and its capacity to mitigate omitted variable bias (Imai & Kim, 2019).

In addition to controlling for individual effects that may affect FDI but cannot be directly observed (such as geography or culture), the FE model also controls for time effects that change over time but do not vary across individuals (such as financial crises), which helps isolate the impact of EPU on FDI while keeping unobserved factors constant. By controlling for fixed effects, the model reduces the risk of omitted variable bias, which is crucial when studying complex relationships such as EPU and FDI.

The model assumes that individual heterogeneity (unobserved factors) is time-invariant. However, if this assumption is violated (i.e., time-varying unobserved factors exist), the model may lead to biased estimates. Alternative approaches such as the RE model can be considered, but RE assumes no correlation between unobserved effects and regression variables, which may also lead to bias if violated. Unlike the RE model, the FE model does not assume that unobserved heterogeneity is uncorrelated with regression variables, making it more robust when such correlation exists. Compared to pooled OLS, FE accounts for unobserved heterogeneity, which is crucial in cross-country studies (Baltagi et al., 2005; Wooldridge, 2002).

(ii)

Generalized Method of Moments (GMM) Model

The GMM approach is chosen for its ability to handle dynamic relationships and endogeneity issues. In practice, FDI decisions are often influenced by past trends, and there may be bidirectional causality between EPU and FDI. GMM uses lagged variables as instrumental variables to address these challenges, producing more reliable estimates when simple models may be insufficient.

The GMM is a dynamic panel data estimation technique specifically designed for panel data models where the dependent variable is influenced by its lagged values; therefore, it is particularly suitable for analyzing the relationship between EPU and FDI (Mileva, 2008). The advantages of GMM are primarily reflected in endogeneity, dynamic relationships, heterogeneity, and measurement errors (Arellano & Bond, 1991).

The GMM model has significant advantages in endogeneity, dynamic relationships, heterogeneity, and measurement errors. First, endogeneity. EPU and FDI may have bidirectional causality, where EPU affects FDI, and FDI may also affect EPU (for example, through economic stability or policy adjustments). GMM can handle endogeneity by using lagged values of variables as instrumental variables. Second, dynamic relationships. FDI decisions are typically influenced by past FDI flows and other persistent factors. GMM incorporates lagged dependent variables, making it well-suited for capturing dynamic relationships. Third, heterogeneity. GMM accounts for unobserved heterogeneity across countries or regions, which is common in FDI research. Fourth, measurement errors. GMM mitigates the impact of measurement errors in EPU or FDI data through the use of instrumental variables. By using lagged values as instrumental variables, GMM can effectively address endogeneity, but this depends on valid instrumental variables. If instrumental variables are not strong or endogeneity exists, the results will be biased. Alternative approaches such as Instrumental Variables (IV) can also be considered, but careful selection of valid instrumental variables is required.

Unlike Ordinary Least Squares (OLS), GMM does not assume strict exogeneity and can handle endogenous regressors. Compared to static models such as pooled OLS or random effects, GMM is more suitable for dynamic panel data, which frequently occurs in FDI and EPU.

When studying the relationship between EPU and FDI, GMM and FE models are preferred because they can handle endogeneity, dynamic relationships, and unobserved heterogeneity. Using both models allows for robustness testing. If both models produce consistent results, the findings are more credible. While other models such as pooled OLS, random effects, or static models have limitations, GMM and FE provide more reliable and nuanced insights, making them the most suitable choices for such research.

3.3.2. Identification of Endogeneity Problems

The relationship between EPU and FDI faces several potential endogeneity challenges, which we comprehensively address (Antonakakis et al., 2017; Canh et al., 2020):

Reverse causality: Large-scale FDI flows from specific countries to China may influence China’s domestic policy uncertainty through lobbying, political pressure, or economic interdependence effects (Gulen & Ion, 2015). While this channel appears empirically limited given the relative size of individual country FDI flows relative to China’s economy, we cannot a priori rule out this possibility (Julio & Yook, 2016).

Omitted variable bias: Unobserved factors that simultaneously affect both EPU and FDI may bias our estimates. For example, global risk sentiment may simultaneously increase EPU indices and reduce international investment flows (Bloom, 2009; Carrière-Swallow & Céspedes, 2013). Our time fixed effects partially address global confounding factors, but country-specific time-varying omitted variables remain a concern.

Measurement error: The EPU index constructed based on newspaper reports may not perfectly capture true policy uncertainty, potentially leading to attenuation bias in our estimates (Azzimonti, 2018; Baker et al., 2016).

3.3.3. Baseline Panel Regression Model

To evaluate the impact of EPU on FDI, FE model regression is applied to account for country-specific heterogeneity. However, the FE model cannot address endogeneity issues caused by omitted variables or reverse causality. Therefore, the GMM model is employed to mitigate potential bias by using lagged values as instrumental variables, ensuring more reliable and consistent estimates. However, to address potential endogeneity issues, the GMM model is used. The benchmark model is:

$$lnFD{I}_{it}={\beta }_0+{\beta }_{1}ln \left(EP{U}_{it}-EP{U}_{ct}\right)+{\beta }_2{X}_{ct}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

$$lnFD{I}_{it}={\beta }_0+{\beta }_{1}ln\left(\left|EP{U}_{it}-EP{U}_{ct}\right|\right)+{\beta }_2{X}_{ct}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(2)

$$lnFD{I}_{it}={\beta }_{0}lnFD{I}_{it-1}+{\beta }_{1}ln\left(EP{U}_{it}-EP{U}_{ct}\right)+{\beta }_2{X}_{ct}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(3)

$$lnFD{I}_{it}={\beta }_{0}lnFD{I}_{it-1}+{\beta }_{1}ln\left(\left|EP{U}_{it}-EP{U}_{ct}\right|\right)+{\beta }_2{X}_{ct}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(4)

where subscripts i, c, and t stand for home country, China, and year, respectively. $\ln (FD{I}_{it})$ denotes the logarithmic form of FDI to China for different country i in year t. The dependent variable $lnEP{U}_{it}$ and $lnEP{U}_{ct}$, which measures the logarithmic form EPU for home country and China separately in year t. $X_{ct}$ represents the set of control variables. ${\epsilon }_{it}$ represent the error term.

Note that Equations (3) and (4) do not include explicit constant terms, as the system GMM specification with individual fixed effects (μ_i) and time fixed effects (λ_t) renders a global intercept redundant and potentially creates identification issues (Arellano & Bond, 1991; Blundell & Bond, 1998). This specification is standard practice in dynamic panel GMM estimation to ensure proper model identification.

The coefficient of interest in our study is ${\beta }_1$, which represents the marginal effect of EPU on FDI. We expect ${\beta }_1$ in Models (1) and (3) to have a positive effect, indicating that ${EPU}_{it}-{EPU}_{ct}$ has a favorable effect on China FDI inflows. And we expect ${\beta }_1$ in Model (2) and Model (4) to have a negative effect, suggesting that $\left|EP{U}_{it}-EP{U}_{ct}\right|$ has an unfavorable effect on China’s FDI inflows.

3.3.4. Moderating Effects Model

To investigate whether political stability and financial development moderate the relationship between China’s EPU and its FDI inflows, we introduce interaction terms in the model:

$$lnFD{I}_{it}={\beta }_0+{\beta }_{1}ln\left(EP{U}_{it}-EP{U}_{ct}\right)\times Institution+{\beta }_2{X}_{ct}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(5)

$$lnFD{I}_{it}={\beta }_0+{\beta }_{1}ln\left(EP{U}_{it}-EP{U}_{ct}\right)\times Relation+{\beta }_2{X}_{ct}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(6)

$$lnFD{I}_{it}={\beta }_0+{\beta }_{1}ln\left(\left|EP{U}_{it}-EP{U}_{ct}\right|\right)\times Institution+{\beta }_2{X}_{ct}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(7)

$$lnFD{I}_{it}={\beta }_0+{\beta }_{1}ln\left(\left|EP{U}_{it}-EP{U}_{ct}\right|\right)\times Relation+{\beta }_2{X}_{ct}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(8)

Here, Institution is an indicator that measures the quality of China’s governance system and institutional environment, and Relation reflects the level of bilateral political relationship between home country and China. A positive and significant interaction coefficient indicates that higher institution quality or stronger bilateral political relationships can mitigate the negative impact of EPU on FDI.

4. Empirical Results

4.1. Baseline Results

Table 4. Baseline Regression Results.

	(1)	(2)
	lnFDI	lnFDI
D_EPU	0.002 ***
	(0.001)
AD_EPU		−0.004 ***
		(0.001)
c_Market	0.297	0.078
	(0.416)	(0.408)
c_Growth	0.035	0.025
	(0.032)	(0.032)
c_Inflation	0.013	0.029
	(0.031)	(0.030)
c_Tec	−0.223	0.005
	(0.332)	(0.330)
c_natures	−0.160	−0.175
	(0.256)	(0.249)
_cons	9.666 ***	9.882 ***
	(0.047)	(0.073)
N	379	379
F	5.379	6.704
r2	0.084	0.102

Standard errors in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the benchmark regression results testing the differential effects of policy uncertainty on China’s FDI inflows. The analysis reveals two distinct mechanisms through which cross-country policy uncertainty differences affect international investment flows, providing new insights into understanding the political economy of FDI.

The empirical results show significant evidence for both directional effects and consistency effects of policy uncertainty. In column (1), the coefficient for D_EPU is positive and highly significant (β = 0.002, p < 0.001), indicating that FDI flows to China increase when home countries experience higher policy uncertainty relative to China. This finding supports the “push effect” mechanism identified in the international finance literature, where policy uncertainty in home countries drives capital toward relatively stable destinations (Baker et al., 2016; Julio & Yook, 2016). The economic impact magnitude suggests that each one-unit increase in EPU difference leads to a 0.2% increase in FDI inflows, reflecting investors’ risk-averse behavior when facing periods of intensified domestic policy uncertainty.

Complementing this directional effect, column (2) reveals a strong consistency effect captured by the Ad_EPU variable. This coefficient is negative and highly significant (β = −0.004, p < 0.001), indicating that expanding absolute differences between home country and China’s policy environments significantly impede FDI flows. This result is consistent with institutional distance theory, which argues that differences in institutional environments increase cross-border operating costs (W. Henisz, 2000; Kostova, 1999). The economic interpretation suggests that each one-unit increase in policy environment inconsistency reduces FDI inflows by 0.4%, highlighting the substantial friction costs generated by operating across different policy frameworks.

The relative magnitudes of these effects provide important theoretical insights. The consistency effect (|−0.004|) is twice the magnitude of the directional effect (|0.002|), indicating that policy coordination is more important than relative policy advantages for international investment decisions. This finding extends the institutional distance literature by quantifying the trade-off between push factors and institutional friction in cross-border capital allocation. The results support the view that while policy uncertainty can generate push effects favoring relatively stable destinations, the costs of operating in different institutional environments ultimately dominate investment decisions (Xu & Shenkar, 2002).

Regarding control variables, it is noteworthy that traditional macroeconomic determinants of FDI—including market development, economic growth, inflation, and technological progress—show limited significance in this specification. This pattern suggests that institutional and policy factors may play a more prominent role than traditional economic fundamentals in FDI allocation, particularly during periods of heightened global policy uncertainty (Pástor & Veronesi, 2013). The dominance of policy-related variables over economic variables aligns with the recent literature emphasizing the growing importance of institutional quality and policy stability in international investment decisions (Blonigen & Piger, 2014).

Model diagnostics support the robustness of these findings. The R-squared increases from 8.4% in the directional effect specification to 10.2% when including consistency effects, indicating that policy coordination provides substantial additional explanatory power. The observed R² values (8.4–10.2%) in our models are consistent with established FDI literature and reflect the inherent complexity of international investment decisions. In panel data analysis of FDI determinants, R² values typically range from 5 to 20%, as foreign direct investment is influenced by numerous factors beyond policy uncertainty, including market size, institutional quality, exchange rates, and cultural factors. The fixed-effects specification controls for time-invariant country characteristics, focusing on within-country variation, which naturally yields lower R² values compared to cross-sectional analysis. More importantly, the statistical significance of our key variables and their economic meaningfulness demonstrate that our models successfully identify the marginal effects of policy uncertainty on FDI flows, which is the primary objective of this analysis.

The F-statistics (5.379 and 6.704, respectively) confirm the overall significance of both specifications, while the consistent sample size of 379 observations ensures comparability across models.

These findings have important policy implications for both source and host countries. For China, the results indicate that maintaining relative policy stability provides a competitive advantage in attracting foreign investment, particularly during periods of global uncertainty. However, the stronger consistency effect suggests that efforts to coordinate policy frameworks with major source countries—through bilateral dialogue mechanisms, policy coordination agreements, or convergence toward international best practices—may yield greater returns in attracting FDI. For source countries, the results highlight the international spillover effects of domestic policy uncertainty, suggesting that policy instability not only affects domestic investment but also drives capital outflows, potentially weakening the domestic economy’s capital base.

In conclusion, Table 4 provides robust evidence for the dual nature of policy uncertainty’s impact on international investment flows. The coexistence of push effects and institutional friction effects demonstrates the complex interactions of policy environments in shaping cross-border capital allocation. These findings contribute to the growing literature on policy uncertainty and international investment while providing practical insights for policymakers seeking to optimize their countries’ positions in global capital markets.

4.2. Moderating Effects of Institution Quality and Bilateral Political Relations

Table 5. Moderating Effects Regression Results.

	(1)	(2)	(3)	(4)
	lnFDI	lnFDI	lnFDI	lnFDI
D_EPU	0.013 ***		0.003 ***
	(0.002)		(0.001)
D_EPU×WGI	0.030 ***
	(0.007)
D_EPU×Agreement			0.014 ***
			(0.004)
AD_EPU		−0.010 ***		−0.005 ***
		(0.003)		(0.001)
AD_EPU×WGI		−0.017 **
		(0.008)
AD_EPU×Agreement				−0.024 ***
				(0.005)
c_Market	0.307	0.240	0.130	−0.198
	(0.405)	(0.414)	(0.413)	(0.398)
c_Growth	0.046	0.017	0.038	0.024
	(0.031)	(0.032)	(0.031)	(0.031)
c_Inflation	0.024	0.008	0.020	0.035
	(0.030)	(0.032)	(0.030)	(0.029)
c_Tec	−0.184	−0.019	−0.102	0.267
	(0.323)	(0.329)	(0.329)	(0.323)
c_na_res	−0.115	0.088	−0.297	−0.220
	(0.249)	(0.279)	(0.255)	(0.241)
_cons	9.785 ***	9.845 ***	9.655 ***	9.895 ***
	(0.053)	(0.075)	(0.047)	(0.070)
N	379	379	379	379
F	7.826	6.398	6.456	9.926
r2	0.135	0.113	0.114	0.165

Standard errors in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the moderation effects regression results, testing how institutional quality and bilateral political relations moderate the relationship between policy uncertainty differences and China’s FDI inflows. The analysis provides strong evidence that both institutional frameworks and diplomatic relations fundamentally alter the impact of policy uncertainty on international investment decisions, offering nuanced insights into the conditional nature of policy–investment relationships.

The institutional quality moderation results in columns (1) and (2) reveal significant moderation effects on both directional and consistency mechanisms. For the directional effect, the interaction term D_EPU×WGI presents a positive and highly significant coefficient (β = 0.030, p < 0.001), indicating that improvements in China’s institutional quality significantly amplify the push effect of home countries’ relative policy uncertainty. The economic mechanism underlying this finding is that when home countries experience higher policy uncertainty relative to China (D_EPU > 0), China’s higher institutional quality makes it a more attractive “safe haven” destination.

According to the “safe haven effect” theory (Caballero & Krishnamurthy, 2008), investors seek investment destinations with more stable and predictable institutional environments when facing policy uncertainty in their home countries. Improvements in China’s institutional quality (higher WGI values) enhance this safe haven attractiveness by providing investors with more reliable property rights protection, more transparent regulatory environments, and more predictable policy implementation mechanisms (Kaufmann et al., 2011; La Porta et al., 1998). The main effect of D_EPU remains positive and significant (β = 0.013, p < 0.001), confirming the existence of the benchmark push effect, whereby home country policy uncertainty indeed drives capital flows toward the relatively stable Chinese market.

The institutional quality moderation effect results in column (2) reveal an important but counterintuitive finding: the interaction term AD_EPU×WGI presents a negative and significant coefficient (β = −0.017, p < 0.05), indicating that when policy environment differences are already substantial, further improvements in institutional quality actually exacerbate the negative impact of such differences on FDI. This phenomenon can be understood through institutional arbitrage theory and the over-institutionalization hypothesis. According to the institutional arbitrage theory of Khanna and Palepu (1997, 2000), multinational enterprises often seek to exploit institutional differences between countries to gain competitive advantages, particularly in emerging markets where institutional “voids” or imperfections provide unique arbitrage opportunities for firms. When policy environments already exhibit substantial differences, moderate institutional flexibility actually provides investors with more operational space and adaptation possibilities.

The bilateral political relations moderation results in columns (3) and (4) provide additional evidence for the conditional nature of policy uncertainty effects. The D_EPU×Agreement interaction term is positive and highly significant (β = 0.014, p < 0.001), indicating that stronger bilateral political relations amplify the push effect of policy uncertainty differences. This finding supports the diplomatic channel hypothesis, whereby established political relations facilitate capital flows during periods of home country policy uncertainty (Gartzke & Li, 2003; Q. Li & Vashchilko, 2010). Strong diplomatic relations may provide additional confidence and practical mechanisms for investors seeking to reallocate capital during uncertain periods.

The bilateral political relations moderation results in column (4) present an equally interesting pattern. The interaction term AD_EPU×Agreement shows a negative and highly significant coefficient (β = −0.024, p < 0.001), indicating that stronger bilateral political relations paradoxically exacerbate the negative impact of policy environment differences. When policy environment differences are already substantial, improvements in bilateral political relations may indeed produce unexpected negative effects, primarily stemming from unbalanced changes in investment structure. According to political connection theory (Faccio, 2006; Fan et al., 2007), good bilateral political relations mainly benefit enterprises with government backgrounds or political connections, particularly state-owned enterprises. These enterprises are better able to leverage intergovernmental cooperation agreements, policy preferences, and official channels to expand their investments in China. However, this politically driven investment growth often accompanies unexpected deterioration in the investment environment for private enterprises. According to Shleifer and Vishny’s (1994) government intervention theory, close political relations often imply increased government intervention. Private enterprises worry that good bilateral relations will lead both governments to adopt more political considerations rather than pure market principles in investment review, market access, and regulatory enforcement. This politicization trend subjects private enterprises to greater policy uncertainty and regulatory risks. Improvements in bilateral political relations may send negative signals to private enterprises, leading to a significant decline in their investment willingness.

Control variables continue to show limited significance across all specifications, reinforcing the conclusion that policy and institutional factors dominate traditional economic determinants in this context. This pattern is consistent with recent literature emphasizing the growing importance of institutional and political factors relative to economic fundamentals in international investment decisions, particularly during periods of heightened global uncertainty (Julio & Yook, 2016; Pástor & Veronesi, 2013).

Model diagnostics support the robustness and incremental value of the moderating variables. The R-squared gradually increases from 13.5% in the institutional quality moderation to 16.5% in the bilateral relations moderation, indicating that diplomatic factors provide substantial additional explanatory power beyond institutional quality. F-statistics range from 6.398 to 9.526, confirming the overall significance of all specifications while demonstrating the enhanced explanatory power of the complete moderation model.

4.3. Heterogeneity Analysis

This study further explores the heterogeneous characteristics of China’s EPU impact through subgroup regression analysis based on high and low levels of China’s EPU relative to enterprises’ home country EPU. The heterogeneity analysis results in

Table 6. Heterogeneity Analysis Results.

	(1)	(2)
	lnFDI	lnFDI
	${\mathit{E}\mathit{P}\mathit{U}}_{\mathit{c}}\mathbf{>}{\mathit{E}\mathit{P}\mathit{U}}_{\mathit{i}}$	$\mathit{E}\mathit{P}{\mathit{U}}_{\mathit{c}}\mathbf{<}\mathit{E}\mathit{P}{\mathit{U}}_{\mathit{i}}$
lnCEPU	−0.663 ***	0.117
	(0.185)	(0.244)
c_Market	0.620	0.310
	(0.534)	(0.726)
c_Growth	−0.001	0.034
	(0.041)	(0.090)
c_Inflation	0.050	−0.047
	(0.038)	(0.049)
c_Tec	−0.133	−0.183
	(0.403)	(0.620)
c_Natures	−0.104	0.138
	(0.333)	(0.493)
_cons	13.096 ***	9.745 ***
	(1.020)	(1.200)
N	293	86
F	5.360	0.238
r2	0.107	0.023

Standard errors in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01.

reveal differentiated impact mechanisms of policy uncertainty’s relative levels on enterprise investment decisions.

When China’s EPU is higher than the home country’s EPU (column (1)), China’s EPU shows a significant negative impact on enterprise investment, with a coefficient of −0.663 significant at the 1% level, indicating that when China’s policy environment has relatively higher uncertainty, increases in policy uncertainty significantly suppress enterprise investment behavior. This result aligns with traditional uncertainty theory expectations, where high policy uncertainty leads enterprises to postpone investment decisions and adopt a “wait-and-see” strategy. In contrast, when China’s EPU is lower than the enterprise’s home country EPU (column (2)), China’s EPU impact on enterprise investment has a coefficient of 0.117 and is not significant, indicating that under conditions where China’s policy environment is relatively stable, the negative impact of policy uncertainty changes on enterprise investment is substantially weakened.

This heterogeneity result indicates that enterprise investment decisions are more influenced by relative policy environments rather than absolute uncertainty levels. When China presents relative policy advantages, enterprises’ tolerance for policy uncertainty significantly increases, reflecting the existence of “policy environment comparative advantage.” This relative advantage effect may partially offset the negative impact of absolute policy uncertainty through channels such as enhancing enterprise investment confidence and reducing risk premiums. Meanwhile, this result also reflects enterprises’ adaptive learning capabilities under different policy environments, whereby enterprises can adjust their investment strategies and risk management mechanisms according to relative policy stability.

4.4. Robustness Analysis

4.4.1. Endogeneity

This robustness test primarily considers the endogeneity problem of mutual causality in the model. Therefore, to address this endogeneity issue, lagged explanatory variables by one period are employed for regression analysis. The following table presents the regression results analysis of the one-period lagged model. The coefficient significance and directions of each variable in the

Table 7. Endogeneity Results.

	(1)	(2)
	lnFDI	lnFDI
L.D_EPU	0.003 ***
	(0.001)
L.AD_EPU		−0.004 ***
		(0.001)
Market	0.846 *	0.495
	(0.483)	(0.456)
Growth	0.025	0.032
	(0.033)	(0.033)
Inflation	0.010	0.003
	(0.031)	(0.031)
Tec	−0.602 *	−0.264
	(0.354)	(0.338)
Na_res	0.048	−0.060
	(0.263)	(0.251)
_cons	9.679 ***	9.912 ***
	(0.048)	(0.070)
N	358	358
F	6.633	7.633
r2	0.107	0.121

Standard errors in parentheses* p < 0.1, ** p < 0.05, *** p < 0.01.

. below are consistent with the regression results in the main tables above, indicating that the model does not suffer from endogeneity problems of mutual causality, and the model demonstrates robustness.

4.4.2. Robustness Tests

Secondly, this paper further employs the system GMM model approach to conduct robustness tests of the model. System GMM estimation can effectively control for endogeneity problems in explanatory variables and handle unobserved individual heterogeneity and lagged dependent variable bias in dynamic panel data. The

Table 8. Robustness Tests Results.

	(1)	(2)
	lnFDI	lnFDI
D_EPU	0.004 ***
	(0.001)
AD_EPU		−0.008 ***
		(0.001)
Market	0.550 *	0.215
	(0.319)	(0.330)
Growth	0.008	−0.028
	(0.025)	(0.028)
Inflation	0.004	0.033
	(0.023)	(0.024)
Tec	−0.363	0.082
	(0.252)	(0.272)
Na_res	0.136	0.273
	(0.203)	(0.220)
_cons	9.696 ***	10.186 ***
	(0.038)	(0.108)
N	379	379
Wald-chi2	79,042.92	71,068.74
Sargan-p	0.000	0.000
AR(1)-p	0.682	0.338
AR(2)-p	0.583	0.953

Standard errors in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the results of system GMM estimation, where columns (1) and (2) respectively report estimation results under different model specifications.

From the estimation results, the coefficient of D_EPU in column (1) is 0.004 and significant at the 1% level, while the coefficient of AD_EPU in column (2) is −0.008 and significant at the 1% level. These results maintain statistical significance, verifying the robustness of the benchmark regression results presented earlier. Meanwhile, the Sargan test p-values for all models are 0.000, indicating that the over-identification constraints of the instrumental variables are satisfied. The AR(1) test p-values are 0.682 and 0.338, respectively, and the AR(2) test p-values are 0.583 and 0.953, respectively, all of which do not reject the null hypothesis of no serial correlation, indicating that the model specification is reasonable and there are no second-order serial correlation problems. Overall, the system GMM estimation results further confirm the robustness of the impact of EPU on enterprise investment behavior.

5. Findings and Discussion

5.1. Main Empirical Findings

This study conducted a comprehensive examination of how global and domestic EPU affects China’s FDI inflows, revealing a complex network of relationships that vary across different investor characteristics and institutional environments. Using panel data from 20 countries spanning 2005 to 2023, our multilevel analytical approach generated several interconnected insights that collectively paint a nuanced picture of the EPU–FDI relationship.

Our benchmark regression results demonstrate two fundamental mechanisms through which policy uncertainty affects international investment flows. First, we identified a significant “push effect,” whereby higher policy uncertainty in home countries relative to China drives increased FDI inflows (β = 0.002, p < 0.001). This finding aligns with the institutional escape theory proposed by (Witt & Lewin, 2007) indicating that enterprises actively seek to mitigate domestic institutional constraints by investing in relatively more stable policy environments. The economic significance suggests that each one-unit increase in EPU difference leads to a 0.2% increase in FDI inflows, reflecting investors’ risk-averse behavior when facing periods of intensified domestic policy uncertainty.

Complementing this directional effect, we found a strong “consistency effect” captured by the absolute EPU difference, which shows a negative and highly significant coefficient (β = −0.004, p < 0.001). This result supports institutional distance theory (Kostova, 1999; Xu & Shenkar, 2002), indicating that expanding absolute differences between home country and China’s policy environments significantly impede FDI flows. The economic interpretation suggests that each one-unit increase in policy environment inconsistency reduces FDI inflows by 0.4%, highlighting the substantial friction costs generated by operating across different policy frameworks.

The relative magnitudes of these effects provide important theoretical insights. The consistency effect (|−0.004|) is twice the magnitude of the directional effect (|0.002|), indicating that policy coordination is more important than relative policy advantages for international investment decisions. This finding extends the institutional distance literature by quantifying the trade-off between push factors and institutional friction in cross-border capital allocation, consistent with recent research by Beugelsdijk et al. (2018) on the multidimensional nature of institutional distance.

5.2. The Moderating Role of Institutional Quality and Political Relations

Our analysis of moderating effects reveals that both institutional frameworks and diplomatic relations fundamentally alter the impact of policy uncertainty on international investment decisions. The institutional quality moderation results show that improvements in China’s institutional quality significantly amplify the push effect of relative policy uncertainty (β = 0.030, p < 0.001). This finding supports the “safe haven” hypothesis proposed by Caballero and Krishnamurthy (2008), whereby investors seek institutionally stable destinations during periods of home country policy uncertainty.

However, we observed a counterintuitive finding regarding institutional quality’s moderation of the consistency effect. The interaction term shows a negative coefficient (β = −0.017, p < 0.05), indicating that when policy environment differences are already substantial, further improvements in institutional quality may paradoxically exacerbate the negative impact on FDI. This phenomenon can be understood through the institutional arbitrage theory of Khanna and Palepu (2000), which argues that multinational enterprises often seek to exploit institutional differences, particularly in emerging markets where institutional “voids” provide unique arbitrage opportunities.

However, when institutional quality is excessively improved, this arbitrage space becomes significantly compressed. High institutional quality environments typically mean stricter regulatory frameworks, more transparent rule enforcement, and less policy interpretation space. Research by Djankov et al. (2002) indicates that excessive institutional standardization may reduce enterprises’ operational flexibility, particularly for those accustomed to operating under relatively lenient institutional conditions. When policy environment differences already create uncertainty, overly strict institutional requirements further limit enterprises’ adaptive strategies, making policy differences that could originally be resolved through flexible operations even more difficult to handle. This effect is more pronounced in countries with higher institutional quality, as enterprises in these countries may lose their ability to find alternative solutions in institutionally imperfect environments.

From a threshold effect perspective, based on Hansen’s (1999, 2000) threshold regression theory, the moderating effect of institutional quality on policy environment differences may have a critical point. Below this critical point, improvements in institutional quality indeed help mitigate the negative impact of policy differences by providing better risk management tools and more predictable business environments. However, when institutional quality exceeds a certain threshold value, further institutional improvements begin to produce marginal negative effects. Olson’s (1982) concept of “institutional sclerosis” in “The Rise and Decline of Nations” provides theoretical support for this: overly developed institutional systems may reduce economic agents’ adaptability to environmental changes, creating overly complex and rigid operational environments that make it difficult for enterprises to flexibly respond to policy environment differences and changes.

The economic mechanism of this finding can be further explained through Aoki’s (2001) institutional matching theory. Different types of investments and business models require matching corresponding levels of institutional environment complexity. Certain investment projects, particularly those relying on rapid decision-making, flexible resource allocation, or exploiting market imperfections, may perform better under relatively simple and flexible institutional conditions. When policy environments already exhibit significant differences, overly complex institutional frameworks further increase enterprises’ compliance costs and decision-making complexity, thereby amplifying the negative impact of policy differences. This effect suggests that the relationship between institutional quality and cross-border investment facilitation is not simply a linear positive correlation, but rather that there exists an optimal level of institutional complexity, beyond which additional institutional improvements may produce counterproductive effects.

These counterintuitive findings reveal the concept of “optimal institutional complexity levels” in international investment attraction. Our empirical results suggest that there exists a critical threshold in the institutional quality–FDI relationship, beyond which further institutional improvements may become counterproductive, particularly when policy environments already exhibit substantial divergence. This non-linear relationship indicates that moderate institutional flexibility may be more conducive to foreign investment than excessive institutionalization. Future research could employ threshold regression models (Hansen, 1999, 2000) to empirically identify the specific threshold values of institutional quality and bilateral political relations beyond which their marginal effects on FDI attraction turn negative during periods of policy uncertainty. Such threshold analysis would enable policymakers to calibrate institutional reforms and diplomatic strategies more precisely, avoiding the pitfalls of over-institutionalization while maintaining sufficient institutional credibility to attract foreign investment.

The bilateral political relations moderation results provide additional evidence for the conditional nature of policy uncertainty effects. Stronger bilateral relations amplify the push effect of policy uncertainty differences (β = 0.014, p < 0.001), supporting the diplomatic channel hypothesis proposed by Gartzke and Li (2003). This indicates that established political relations facilitate capital flows during periods of source country policy uncertainty by providing additional confidence and practical mechanisms for investors seeking to reallocate capital.

Interestingly, bilateral political relations also show a negative moderating effect on the consistency mechanism (β = −0.024, p < 0.001). This finding can be explained through Faccio’s (2006) political connection theory, whereby improvements in bilateral relations primarily benefit politically connected enterprises, particularly state-owned enterprises, while potentially creating unfavorable conditions for private sector investment due to increased government intervention and political considerations in investment decisions.

Furthermore, based on Acemoglu and Johnson’s (2005) institutional quality theory, excessive political intervention may distort the market competition environment. When bilateral relations improve, state-owned enterprises may receive more policy favoritism and resource support, creating a competitive environment disadvantageous to private enterprises. Private enterprises may anticipate being at a disadvantage in such a politicized investment environment, thus choosing to reduce or postpone investment decisions. Murphy et al.’s (1993) rent-seeking theory further explains this phenomenon: the strengthening of political relations may increase rent-seeking activities, while private enterprises typically lack the capability and channels to participate in such activities.

More critically, the reduction in private enterprise investment often exceeds the increase in state-owned enterprise investment, producing a net negative effect. This asymmetric effect can be explained through several mechanisms. According to Kornai (1979, 1986), soft budget constraint theory, state-owned enterprises’ investment decisions are often driven by political factors rather than pure economic efficiency, prone to over-investment and investment hunger phenomena, with their investment growth potentially being relatively limited and less efficient. In contrast, private enterprises are extremely sensitive to political risks and market environment changes, with their investment cuts potentially being rapid and substantial. Jensen and Meckling’s (1976) agency theory also supports this view: private enterprise owners directly bear investment risks, making them more acutely responsive to political uncertainty.

Furthermore, according to Claessens and Laeven’s (2003) financial constraint theory, private enterprises typically face more severe financing constraints. When improved political relations lead to policy environments more favorable to state-owned enterprises, financial resources may further tilt toward the state-owned sector, exacerbating private enterprises’ financing difficulties. This “double squeeze” effect—facing both competitive disadvantages and financing constraints—causes private enterprises’ investment decline to far exceed state-owned enterprises’ growth, thereby leading to a net decrease in overall FDI inflows.

5.3. Heterogeneity of Policy Uncertainty Effects

The heterogeneity analysis reveals that China’s EPU impact on FDI varies significantly according to the relative policy environment. When China’s EPU exceeds that of source countries, Chinese policy uncertainty significantly suppresses FDI inflows (β = −0.663, p < 0.001), consistent with traditional real options theory predictions (Dixit & Pindyck, 1994). Conversely, when China maintains relatively lower policy uncertainty, the negative impact becomes statistically insignificant (β = 0.117, p > 0.10).

This heterogeneity indicates that investment decisions are more influenced by relative rather than absolute policy environments, supporting the comparative institutional advantage hypothesis. When China exhibits relative policy stability, enterprises demonstrate significantly higher tolerance for policy uncertainty, reflecting the existence of “policy environment comparative advantage.” This relative advantage effect may partially offset the negative impact of absolute policy uncertainty through enhancing investor confidence and reducing risk premiums, consistent with recent findings by F. Chen and Jiang (2023) regarding the role of relative institutional quality in FDI decisions.

5.4. Robustness and Endogeneity Considerations

Our robustness tests confirm the stability of the main findings across different empirical specifications. The lagged variable approach addresses potential reverse causality concerns, producing results consistent with benchmark estimates. System GMM estimation effectively controls for endogeneity and unobserved heterogeneity, with diagnostic tests confirming the validity of our instrumental variables (Sargan test p = 0.000) and the absence of a second-order serial correlation (AR(2) test p > 0.10).

These robustness tests are particularly important because, as identified by Gulen and Ion (2015), there exists potential reverse causality between EPU and FDI. Large-scale FDI from specific countries could theoretically influence China’s domestic policy uncertainty through lobbying, political pressure, or economic interdependence effects. However, our instrumental variable approach and the relative scale of individual country FDI flows suggest this channel is empirically limited.

5.5. Theoretical Contributions and Policy Implications

Our findings make several important theoretical contributions to the international business and political economy literature. First, we extend real options theory beyond the traditional host country perspective by incorporating source country dynamics and their interactive effects. Second, we provide empirical evidence for the dual nature of policy uncertainty effects, showing that both directional (push) and consistency mechanisms operate simultaneously in international investment decisions.

Third, our results highlight the conditional nature of the EPU–FDI relationship, showing that the impact of policy uncertainty critically depends on institutional quality and bilateral political relations. This finding contributes to the growing literature on the conditional effects of institutional factors on international investment (W. J. Henisz & Delios, 2001; Slangen & Beugelsdijk, 2010).

In terms of policy implications, our results indicate that for China, maintaining relative policy stability provides a competitive advantage in attracting foreign investment, particularly during periods of global uncertainty. However, the stronger consistency effect suggests that efforts to coordinate policy frameworks with major source countries—through bilateral dialogue mechanisms, policy coordination agreements, or convergence toward international best practices—may yield greater returns in attracting FDI.

For home countries, our findings highlight the international spillover effects of domestic policy uncertainty, indicating that policy instability not only affects domestic investment but also drives capital outflows, potentially weakening the domestic economy’s capital base. This emphasizes the importance of policy predictability and institutional stability for maintaining the competitiveness of domestic investment environments in an interconnected global economy.

6. Conclusions

This study examines the complex relationship between EPU and China’s FDI inflows, providing new insights into understanding how policy environments across different countries interact to shape international investment decisions. Using comprehensive panel data from 20 countries during 2005–2023, our analysis reveals a nuanced picture of EPU–FDI dynamics that goes beyond traditional single-country perspectives.

Our main findings indicate that policy uncertainty affects international investment through two distinct but complementary channels. The directional effect shows that higher policy uncertainty in home countries relative to China creates a “push effect,” driving increased FDI inflows and supporting institutional escape theory. Simultaneously, the consistency effect reveals that absolute differences between home country and China’s policy environments significantly impede investment flows, consistent with institutional distance theory. Importantly, the consistency effect is proven to be twice the magnitude of the directional effect, indicating that policy coordination is more critical than relative policy advantages for international investment decisions.

The moderating effects analysis provides additional complexity to these relationships. Institutional quality and bilateral political relations fundamentally alter the impact of policy uncertainty on investment decisions, but not always in expected directions. While higher institutional quality amplifies the beneficial push effects of relative policy stability, it paradoxically exacerbates negative impacts when policy differences are already substantial. Similarly, stronger bilateral political relations facilitate investment during periods of source country uncertainty but may create unfavorable conditions for private sector investment due to increased political considerations.

The heterogeneity analysis reveals that China’s policy uncertainty effects vary significantly according to the relative policy environment. When China maintains lower uncertainty than home countries, the traditional negative relationship between uncertainty and investment becomes statistically insignificant, highlighting the importance of comparative rather than absolute policy environments in investment decisions.

These findings have important theoretical implications. Our study extends real options theory by incorporating bilateral perspectives and demonstrating the conditional nature of uncertainty effects. Evidence for dual mechanisms—directional and consistency effects—provides a more complete understanding of how policy environments shape international capital allocation. Furthermore, the moderating role of institutional factors reveals that EPU–FDI relationships are highly context-dependent, challenging simple linear interpretations of uncertainty effects.

From a policy perspective, our results provide valuable guidance for both China and home countries. For China, maintaining relative policy stability provides a competitive advantage in attracting foreign investment, particularly during periods of global uncertainty. However, the dominance of consistency effects suggests that efforts to coordinate policy frameworks with major source countries may yield greater returns than simply maintaining stability. This might involve developing bilateral policy coordination mechanisms, converging toward international best practices, or creating institutional frameworks that reduce cross-border operational friction.

For home countries, our findings emphasize the international spillover effects of domestic policy instability. Policy uncertainty not only affects domestic investment environments but also drives capital outflows, potentially weakening the competitiveness of home economies. This highlights the importance of policy predictability and institutional consistency in maintaining attractiveness for both domestic and foreign investment.

However, it is crucial to acknowledge that China’s unique characteristics as the host country in our analysis may significantly influence the generalizability of our findings. China’s exceptional economic scale, distinctive institutional environment combining market mechanisms with strong state intervention, and unique political system create a contextual framework that may not be readily applicable to other emerging markets. The “push” versus “consistency” framework we identify may be particularly pronounced in China’s case due to its position as both a major global economic power and an emerging market with rapidly evolving institutional structures. This specificity does not diminish the validity of our current findings but rather underscores the critical importance of conducting comparative research across diverse emerging market contexts. Future studies should systematically test whether the dual mechanisms of directional and consistency effects operate similarly in other developing economies with different institutional characteristics, political systems, and economic development levels. Such comparative analysis would be essential for establishing the external validity of our theoretical framework and determining whether China represents a unique case or exemplifies broader patterns in emerging market investment dynamics.

Several limitations of our study provide opportunities for future research. First, our analysis focuses on China as a single host country, which may limit the generalizability of findings to other emerging markets with different institutional characteristics. Future research could examine whether similar dual mechanisms operate in other developing economies or whether China’s unique institutional features drive these results. Second, while our EPU measures capture newspaper-based policy uncertainty, they may not fully reflect all dimensions of policy instability that affect investment decisions. Alternative uncertainty measures, such as policy-specific indices or forward-looking indicators, may provide additional insights.

Alternative uncertainty measures, such as sector-specific policy volatility indices (e.g., trade policy uncertainty or regulatory uncertainty indices), forward-looking indicators derived from options pricing or survey-based expectations, or real-time policy tracking measures, could capture dimensions of policy instability that newspaper-based indices may miss, including anticipatory effects and sector-specific policy risks. Future research incorporating multiple uncertainty measures would provide a more comprehensive understanding of how different types and sources of policy uncertainty interact to influence investment decisions, potentially revealing heterogeneous effects across policy domains that our current aggregate measure may obscure.

Third, our analysis primarily focuses on aggregate FDI flows without distinguishing between different types of investment (greenfield versus M&A) or sectors, which may respond differently to policy uncertainty. Future research could explore these disaggregated relationships to provide more targeted policy recommendations. Finally, the rapid evolution of global economic conditions, including recent developments in trade tensions, technological competition, and geopolitical realignment, suggests that EPU–FDI relationships may continue to evolve, requiring ongoing empirical investigation.

Despite these limitations, our study contributes to understanding how policy uncertainty shapes international investment in an increasingly interconnected global economy. As policy environments continue to evolve and global uncertainty remains elevated, the insights from this study provide valuable guidance for policymakers seeking to optimize their countries’ positions in global capital markets while maintaining domestic economic stability.