Economic Policy Uncertainty, Geopolitical Risk, and the U.S.–China Relations: A Risk Transmission Perspective

Abstract

This study examines risk transmission between the United States and China using integrated economic policy uncertainty (EPU) and geopolitical risk (GPR) indices. We employ a dual methodology that combines Vector Autoregressive (VAR) and Granger causality in quantiles tests to analyze interactions during systemic leadership transitions, a dimension that is currently under-explored. Our dataset covers the period from June 2000 to June 2023. Results indicate that China is narrowing the economic influence gap and strengthening its role as a regional anchor. The U.S., however, maintains predominant global leadership. This dynamic reframes bilateral tensions as a “status dilemma” rather than a security conflict. Crucially, we identify asymmetric spillover effects: the U.S. uncertainty shocks spread globally, while China’s volatility remains regional. Our findings contribute to the understanding of financial stability by demonstrating that leadership asymmetries are critical determinants, providing valuable insights for designing systemic risk monitoring tools and contagion mitigation policies during periods of heightened uncertainty.

1. Introduction

Geopolitical risk is recognized by the Bank of England as a critical factor within the broader context of economic policy uncertainty, forming an “uncertainty trinity” with substantial economic implications (Carney, 2016). Recent crises, including COVID-19, the Russia-Ukraine conflict, and tensions in the Middle East, have challenged global financial stability. In response, academic and policy discussions have intensified around how geopolitical risks and financial uncertainty influence markets. This growing body of literature now encompasses topics such as asset pricing (Caldara & Iacoviello, 2022), stock price crash risk (M. Choi et al., 2025), and capital structure (Chowdhury et al., 2025).

Yet, little is known about how geopolitical risk (GPR) and economic policy uncertainty (EPU) affect risk transmission, especially between major powers. Our research examines how shifts in EPU and GPR drive risk transmission between the U.S. and China and influence regional sub-core economies. We leverage the U.S.–China competition to fill this gap. As China has rapidly grown in economic and global stature, friction with the U.S., epitomized by the trade war, has become inevitable. Crucially, a comparison of these nations must consider their profound historical and demographic differences: the U.S. is a mature, globally entrenched democracy with demographic stability, while China is a rapidly developing, state-capitalist country now facing demographic challenges from its one-child policy. These divergent backgrounds shape the character and evolution of the uncertainty each nation generates. Building upon the complex dynamics previously outlined, it is important to note that, despite extensive policy debates, rigorous empirical analysis of how EPU (Berna et al., 2016; Guo et al., 2018) and GPR (M. Choi et al., 2025) affect the relative hegemonic stability and financial leadership of the U.S. and China remains under-developed.

To fill this empirical gap, our study examines the evolving systemic leadership roles of the U.S. and China through the dual lenses of EPU and GPR. We use a dual methodology framework integrating VAR and Granger causality in quantiles analysis. This captures both system-wide spillover and directional dynamics, especially in extreme conditions. Our analysis distinguishes “core countries” (the U.S. and China) from “sub-core” economies to highlight how systemic leaders propagate shocks globally and regionally. Our “core countries” framework posits the U.S. and China as primary sources of global and regional uncertainty. We examine their spillover effects on four sub-core economies: Japan, as a developed financial hub; India, a major emerging market; Hong Kong, an international financial gateway; and Macao, reflecting regional sentiment towards China. Due to data constraints in mainland China, Hong Kong and Macao serve as economic proxies.

This study makes significant contributions to the literature on international relations and financial stability. First, it presents a theoretical framework that reconceptualizes U.S.–China frictions through a “status dilemma” (Pu, 2022; Schweller & Pu, 2011), suggesting that core rivalry arises from competition over hierarchical standing and prestige, which is reflected in the competing transmissions of uncertainty. Second, it provides empirical evidence of a fundamental asymmetry in systemic risk propagation: The U.S.-originated uncertainty shocks have a global impact through deep capital markets and dollar hegemony, while Chinese volatility remains regionally confined, reinforcing its role as a regional anchor rather than a global leader. Finally, we present a dual methodology framework that integrates VAR model with Granger causality in quantiles tests, improving the analysis of complex financial interdependencies and providing deeper insights into the relationships among financial variables.

These findings are relevant for policymakers and financial institutions. Understanding the origin, magnitude, and channels of financial contagion is essential for developing effective macro-prudential regulations and targeted risk monitoring. Regulators in globally connected economies should prioritize monitoring the U.S.-related economic policy and geopolitical uncertainty, while those in the Asia-Pacific must design systems capable of tracking both global (the U.S.) and regional (Chinese) shocks. Such targeted approaches enable timely and effective responses to periods of volatility or heightened geopolitical tensions.

The remainder of this paper is structured as follows. Section 2 reviews the literature on systemic leadership, and the contextual divergences between the U.S. and China, establishing the theoretical foundation for analyzing the spillover effects of EPU and GPR, and culminating in our testable hypotheses. Section 3 details the empirical methodology, including data sources, and the dual methodology approach. Section 4 presents empirical results and discusses the identified spillover dynamics and causal relationships. Finally, Section 5 concludes by synthesizing the key findings and outlining their theoretical and policy implications.

2. Literature Review and Hypothesis Development

2.1. Literature Review

2.1.1. Systemic Leadership in a Transitional Era

The study of systemic leadership, rooted in hegemonic stability and power transition theories, has focused on the dominant state’s role in providing global public goods and upholding the rules-based order (Organski & Kugler, 1980; Gilpin, 1988). Yet, the rise of China and power diffusion now challenge these views (Allison, 2017; Nye, 2022). Leadership is moving from monopolistic to competitive and dyadic forms (Mastro, 2019; Amr, 2025). As a result, research now explores how established and rising powers interact, reshaping norms and addressing transnational challenges (Ikenberry, 2020; Garba, 2025). In this context, the interplay between EPU and GPR has emerged as a critical analytical framework for understanding the tangible manifestations of systemic leadership in a rapidly transforming world.

2.1.2. Economic Policy Uncertainty (EPU) as a Metric of Economic Stewardship

The concept of EPU, developed by Baker et al. (2016), measures market perceptions of unpredictable fiscal, regulatory, and monetary policies. High EPU lowers investment, consumption, and growth. Global EPU indices allow cross-country comparisons and show large economies’ significant spillover effects (Ahir et al., 2022). From a systemic leadership view, a nation’s EPU extends beyond its borders as a global externality. For the U.S., as issuer of the main reserve currency, domestic policy uncertainty brings volatility to advanced and emerging economies (Kido, 2018; Kwon, 2025), showing the U.S.’s strong impact on global finance. Managing domestic EPU is crucial, promoting global stability through predictable, transparent governance.

Parallel to this, China’s deepening integration into global financial markets and the expansive reach of initiatives like the Belt and Road Initiative (BRI) and Asian Infrastructure Investment Bank (AIIB) mean that uncertainty surrounding its economic policies now carries substantial international implications (Huang & Luk, 2020; Sheng & Nascimento, 2021). The transmission mechanisms of Chinese EPU, however, are distinct, filtered through its state-capitalist model and unique political institutions (Pan, 2019). A growing body of research thus leverages the EPU paradigm to argue that a comparative analysis of U.S. and Chinese EPU spillovers serves as a crucial barometer of their competing economic leadership and the stability they respectively engender or disrupt within the global system (Biljanovska et al., 2021).

2.1.3. Geopolitical Risk (GPR) and the Securitization of Commerce

Complementing the economic focus of EPU, the Geopolitical risk (GPR) index developed by Caldara and Iacoviello (2022) quantifies risks from wars, terrorism, and inter-state tensions that disrupt international relations. Elevated GPR affects global supply chains, triggers energy price shocks, and drives capital flight to safe-haven assets, threatening macroeconomic stability. Historically, the U.S. has acted as a security guarantor, mitigating these risks to maintain a stable environment for commerce (Ikenberry, 2011). A significant shift is evident, as contemporary scholarship identifies U.S. foreign policy and its competition with China as primary drivers of global GPR. This rivalry, manifest in the Indo-Pacific, technological decoupling, and global institutions, creates a pervasive layer of systemic risk (Allison, 2017). Concurrently, China’s assertive foreign policy and territorial claims, particularly in the South China Sea, are recognized as key contributors to elevated GPR (Johnston, 2013; Y. Zhang, 2024).

This reflexivity, where systemic powers act as both managers and sources of geopolitical risk, characterizes the current transitional era. Thus, a nation’s “GPR footprint,” or the extent to which its actions elevate the global risk index, serves as a measurable gauge of its role in fostering or undermining international peace and security, a core aspect of systemic leadership.

2.1.4. The Imperative of Contextual and Historical Nuance

Comparing the U.S. and China requires attention to their significant differences. Overlooking these risks can lead to oversimplification or misleading conclusions. Their unique histories and demographics shape their economies, priorities, and innovation (Y. Zhang, 2024).

The U.S., a market-oriented democracy, shaped the post-WWII liberal order and benefited from early technological advantages. In contrast, China’s rise began only recently, evolving from colonialism and revolution into central planning, and then state-guided integration post-1978 (Wei, 1995; Zheng & Li, 2022). This difference puts the U.S. and China at different development stages and under distinct political-economic systems, shaping policy uncertainty and global impacts (Gilpin, 1987).

The population contrast is also marked. The U.S. has a mature, slow-growing population, supported by immigration and a consumption-based economy. China reaped a “demographic dividend” that fueled export-led growth but now faces aging and a shrinking workforce (Can & Chan, 2022). These trends create different fiscal, social, and growth constraints, which are essential control variables for comparing spillovers (Yaqoob et al., 2024). Critically, the international transmission of the U.S. and Chinese EPU and GPR is not a uniform process but is filtered through these distinct historical and demographic lenses.

2.2. Hypothesis Development

Recently, economic policy uncertainty (EPU) has become a significant factor influencing global economic outcomes. This is largely driven by escalating incidents of political and economic instability (Baker et al., 2016). EPU, now attracting increased scholarly interest, plays a crucial role in shaping economic activities. Its impact on the economy has been examined primarily from two perspectives: economic outcomes and corporate strategies.

Prior research has shown that EPU negatively affects economic outcomes. It diminishes real investments (Bernanke, 1983; Bloom, 2009), reduces economic growth (Bloom, 2009), and elevates the unemployment rate (Fontaine et al., 2020). Another stream of literature explores the influence of EPU on corporate operations. During heightened EPU, companies typically face substantial limits in securing external financing. This affects both stock markets (Çolak et al., 2017) and loan markets. EPU also inhibits corporate capital investments (Gulen & Ion, 2016) and mergers and acquisitions (M&A) activities (Nguyen & Phan, 2017).

Geopolitical instability and policy unpredictability heavily impact financial markets and real economies (Mawusi, 2020). Abrupt moves, such as proposed U.S. tariffs, amplify global uncertainty and strain relations. Biljanovska et al. (2021) found that EPU reduces output growth, consumption, and investment. These works stress two points: uncertainty has a negative impact, and stability is valuable for growth.

Drawing from the theoretical and empirical foundations above, and controlling for key historical and demographic contexts, the following hypotheses are advanced. They form a logical sequence: from a core economic principle to testing whether China is now a systemic risk source, and then to examining persistent structural asymmetry between the two powers.

H1. 

Within-Country Effect: Heightened domestic economic policy uncertainty (EPU) exerts a negative effect on domestic economic performance, as measured by GDP growth, consumer price index (CPI), unemployment rate, and stock market indices for all studied economies (the U.S., China, Japan, India, Hong Kong, and Macao).

H1 establishes a baseline, grounded in the well-established principle that uncertainty induces precautionary savings and delays irreversible investment decisions, thereby reducing aggregate output (Bernanke, 1983; Bloom, 2009; Baker et al., 2016; Tam, 2020).

Second, international transmission of uncertainty is hierarchical. Evidence shows that major economies have a larger impact on peripheral markets (Berna et al., 2016; Das et al., 2019; S. Y. Choi, 2022). If China matches the U.S. in systemic influence, then shocks from both “core” economies should spill over to “sub-core” economies (e.g., Japan, India), with minimal reverse transmission. This hypothesis tests China’s role in global risk. Recent research finds that the connection between uncertainty and financial links changes over time and is sensitive to crises (Leone et al., 2025). In China, EPU shapes foreign direct investment (FDI) by affecting risk perceptions. L. Zhang and Colak (2022) showed China’s EPU restrains outbound FDI but does not significantly deter inflows. Conversely, Dong et al. (2025) suggested that both domestic and global EPU may affect FDI into China, but financial development and political stability can reduce these impacts. Wu and Shao (2023) emphasized that uncertainties in both home and host countries shape outward FDI decisions of Chinese firms, showing varied impacts on international investment behavior.

H2. 

Core-Periphery Spillover: EPU and GPR shocks originating in systemic core economies (U.S. and China) spill over directionally to sub-core economies (e.g., Japan, India, Hong Kong, and Macao), with no significant reverse transmission.

H2 tests the hierarchical nature of international uncertainty transmission. Major “core” economies disproportionately impact smaller, interconnected “peripheral” markets (Berna et al., 2016; Kido, 2018). If supported, this would signal that China has reached a level of systemic influence. Its domestic shocks would have external effects comparable to those of the U.S.

Over the past decade, China has become the world’s second-largest economy. This rise came with increased geopolitical and economic tensions with major powers, especially the U.S. These frictions span trade, technology, security, and human rights, raising systemic risks in Asia-Pacific financial markets and global capital flows. Sino-U.S. relations, now more about strategic competition, have seen trade wars, sanctions, and disputes over Taiwan, the South China Sea, and other flashpoints. Such conflicts create EPU and GPR, disrupting investment, asset pricing, and financial stability. Welsh (2019) argues that China’s move from economic actor to assertive military power challenges U.S. interests and regional stability. This dynamic fits with the “Thucydides Trap,” which warns of conflict when a rising power confronts an established hegemon (Beckley, 2021; Yoder, 2019; Cole, 2023).

Recognizing the historical depth and structural power of the U.S. dominance in global finance and security (Ikenberry, 2020; Mastanduno, 2019), H3 posits a persistent asymmetry. Hegemonic stability theory suggests that uncertainties from the dominant power have outsized international effects. Even if China causes significant spillovers (as tested in H2), the U.S. financial system’s influence may cause a unidirectional effect: from the U.S. to China. This would confirm the ongoing structural power of American hegemony.

H3. 

Asymmetric Hegemonic Transmission: The U.S. EPU and GPR shocks are a significant causal driver of fluctuations in China’s EPU and GPR, with no significant reverse transmission, confirming the persistent hegemonic influence of the U.S. within the global financial system.

3. Data and Methods

3.1. Data

This study uses two primary indices to measure uncertainty. For economic uncertainty, we utilize global and country-specific economic policy uncertainty (EPU) indices from Baker et al. (2016), which analyze newspaper coverage of economic policy and public sentiment regarding future policies. For geopolitical risk, we employ the GPR index (Caldara & Iacoviello, 2022), which tracks risks associated with wars, terrorism, and international tensions. Since Macao is not included in the standard EPU index, we adopt a Macao-specific version from Luk (2018). This monthly index, available from June 2000, uses a similar text-based method and correlates strongly with Macao’s key macroeconomic indicators. As Macao lacks a comprehensive stock market index, we construct one using the stock prices of its six largest casino firms listed in Hong Kong. We apply weighted summation and statistical adjustments to make the index comparable internationally.

Although the EPU and GPR indices start before 2000, we use data from June 2000 to June 2023 to ensure consistent periods across the U.S., China, Japan, India, Hong Kong, and Macao. We process all data using Python (version 3.8.10) and SPSSAU (2023). For macroeconomic variables such as GDP growth, CPI, unemployment rate, and stock market indices, we source data from WIND and CEIC. We use the NYSE Composite Index for the U.S. For China, we convert quarterly macroeconomic data to a monthly frequency using EViews (version 12).

Limited monthly unemployment data for China leads the quantile causality regression to focus on extreme GPR cases: the top tercile (high risk, above the 67th percentile) and the bottom tercile (low risk, below the 33rd percentile). This approach examines causality across varying levels of geopolitical risk.

3.2. Methods

3.2.1. The Vector Autoregressive (VAR) Model

We first implement a standard Vector Autoregressive (VAR) model to test for Granger causality in the conditional mean. This model provides a baseline assessment of average relationships and captures dynamic interdependencies among multiple time series variables. Through the VAR framework, we assess reciprocal feedback mechanisms that support empirically robust macroeconomic analysis (Sims, 1980).

Following Lutkepohl (2005), the VAR(p) model is specified as:

$$y_t={\mathrm{A}}_1{y}_{t-1}+\dots +{\mathrm{A}}_p{y}_{t-p}+\mathrm{B}{x}_t+{\epsilon }_t$$

(1)

where

y_t = a vector of endogenous variables,

x_t = a vector of exogenous variables,

A₁y_t−1 +···+ A₁y_t−p + Bx_t = coefficient matrices,

ε_t = a vector of serially uncorrelated innovations.

To achieve stationarity, logarithmic transformations and differencing are applied where necessary, and unit roots are formally tested using the Augmented Dickey–Fuller (ADF) test. A VAR(1) system (a vector autoregression model with one lag) is specified as:

y_t = α₁ + β₁₁ y_t−1 + β₁₂x_t−1 + u_t

(2)

x_t = α₂ + β₂₁y_t−1 + β₂₂x_t−1 + v_t

where u_t and v_t are white noise error terms. The coefficients β₁₁, β₁₂, β₂₁, and β₂₂ are estimated using ordinary least squares (OLS). To investigate dynamic interrelationships between the variables, we employ both Engle–Granger cointegration tests and Granger causality tests. If a variable Granger-causes another, its past values contribute useful information for predicting the other variable, beyond what is contained in that variable’s own history.

Lag length selection is crucial for model accuracy. We minimize Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Final Prediction Error (FPE), and Hannan-Quinn information criteria (HQIC), all pointing to a lag order of 1. Model stability is checked using an autoregressive root graph. Analysis focuses on the top 90% high-risk and bottom 10% low-risk groups. We exclude more extreme quantiles to ensure at least 20 data points for reliable results.

Impulse response functions trace how endogenous variables react to exogenous shocks. For example, a political risk shock shows dynamic effects on macro variables like GDP, inflation, unemployment, and stock prices, guiding policy to mitigate short-term shocks and consider long-term effects.

3.2.2. Granger Causality in Quantile Tests

To capture nonlinear state-dependent causal relationships that a VAR model alone may miss, we supplement our analysis with Granger causality tests in quantiles. Following Troster et al. (2018), this approach allows us to examine causality across different market conditions, such as bearish, normal, and bullish states, and identify whether the direction of influence varies along the economic distribution.

Following Granger (1969) and Chuang et al. (2009), we define that the random variable x does not Granger-cause the random variable y if:

Fy_t (η|(Y, X)_t−1) = Fy_t (η|Y_t−1), ∀η ∈ R,

(3)

where Fy_t (⋅∣F) denotes the conditional distribution of y_t, and (Y, X)_t−1 represents the information set generated by past values of y_i and x_i up to time t − 1. If this equality fails for any η, x is said to Granger-cause y. A necessary (but not sufficient) condition for (3) is:

E[y_t|(Y, X)_t−1] = E(y_t|Y_t−1).

(4)

Let E(y_t∣F) denote the mean of the conditional distribution Fy_t (⋅∣F). We say that x does not Granger-cause y in mean if condition (3) holds; otherwise, x Granger-causes y in mean. This is typically tested by estimating a linear model for E[y_t|(Y, X)_t−1]:

y_t = α₀ + ∑^p_i=1 α_i y_t−1 + ∑^q_j=1 β_j x_t−1 + ϵt,

(5)

where rejecting the null hypothesis β_j = 0 for all j = 1,…,q implies Granger causality from x to y. Granger non-causality can also be defined in terms of conditional quantiles. Let Qy_t (τ | F) denotes the τ-th quantile of Fy_t (·|F). Condition (3) is equivalent to:

Qy_t (τ|(Y, X)_t−1) = Qy_t (τ|Y_t−1), ∀τ ϵ (0,1).

(6)

If Equation (6) holds, x does not Granger-cause y across any quantile. Alternatively, one may test non-causality at a specific quantile τ by evaluating (6) for a given τ.

This study examines the two-way interactions between the GPR and EPU indices of the U.S. and China, while simultaneously assessing their spillover effects on four Asian economies: Japan, India, Hong Kong, and Macao.

Building on the methodological approaches of Guo et al. (2018) and Kannadhasan and Das (2020), we specify two parallel quantile regression models to capture these dynamics:

For EPU transmission:

Qy_t (τ|(Y, X)_t−1) = Qτ (y_t|Y_t−1, EPU_t)

= α_i (τ) + ∑^p_i=1 β_j (τ) y_t−1 + γ_i (τ) y_t−1 I (y_t−1|> y^q) + θ_i EPU_t + e_t

(7)

For GPR transmission:

Qy_t (τ|(Y, X)_t−1) = Qτ (y_t|Y_t−1, GPR_t)

= α_i (τ) + ∑^p_i=1 β_j (τ) y_t−1 + γ_i (τ) y_t−1 I (y_t−1|> y^q) + θ_i GPR_t + e_t

(8)

The variable y_t measures index or macroeconomic changes across countries. We first examine U.S. leadership by regressing its EPU/GPR (log-differenced) on China’s indices and regional variables, then reverse the analysis. Y_t−1 denotes lagged values, with β_j(τ) showing lagged effects and γ_i(τ) capturing extreme shocks via indicator I(|y_t−1| > y^q) (threshold: 95th quantile). The coefficient θ_i(τ) captures the impacts of EPU_t and GPR_t on other countries’ indices or economic variables, which are critical for testing our hypotheses.

Although sensitivity analysis of composite methods is limited, our dual VAR-quantile approach aligns with data mining literature on method-specific performance dependencies (Ruiz & Nieto, 2000), integrating VAR with quantile causality, as shown by Oztekin et al. (2013) and Sevim et al. (2014).

y = f (x₁, x₂, …, x_n)

(9)

Our EPU/GPR prediction model employs a linear regression framework:

f (x₁, x₂, …, x_n) = β + ∑ a_i x_i

(10)

For neural networks, this generalizes to:

f (x₁, x₂, …, x_n) = Φ (w₀ + ∑ w_i x_i)

(11)

where Φ is the activation function and wi are weights. The fusion model combines m predictors:

y_f = ψ (y_i, j) = ψ (f_1,(x), f₂ (x),…, f_m (x))

(12)

For simplicity, this study assumes that the operator ψ is linear, thereby transforming the equation into a “composite model” which can be expressed as:

y_f = ∑ ω_i f_i (x) = ω₁ f₁ (x) + ω₂ f₂ (x) + … + ω_m f_m (x)

(13)

where ω_i represents the weights assigned to each model f_i (x). The normalized weights (∑ω_i = 1) reflect each model’s predictive accuracy, prioritizing superior performers. For this study, we treat the VAR model and Granger causality in quantiles as equally weighted components (ω₁ = ω₂ = 0.5), suggesting their joint efficiency in forecasting EPU/GPR dynamics. Following Sevim et al. (2014), we employ sensitivity analysis to optimize Granger causality tests. The sensitivity measure for variable n combines inputs from m models, is expressed as:

S_n,f = ∑_i^m ω_i S_i,n = ω₁ S_1, n + ω₂ S_2, n + … + ω_m S_m,n

(14)

where S_i,n represents the i-th model’s sensitivity to variable n, weighted by ω_i. This aggregated measure evaluates each variable’s comprehensive influence. In our framework (m = 2 models, n = 4 variables), this approach identifies the most impactful economic indicators for uncertainty transmission analysis.

4. Empirical Results and Discussion

4.1. Empirical Results

The following subsections report empirical findings from the previously introduced econometric framework. To improve inference reliability, this framework is specifically calibrated to address potential biases stemming from endogeneity, heteroscedasticity, and multicollinearity. Results and analysis are structured to build logically from domestic impacts through to global transmission.

4.1.1. The Domestic Impact of Economic Policy Uncertainty

Empirical results demonstrate that elevated domestic EPU consistently depresses macroeconomic performance across all studied economies.

Table 1. Descriptive statistics.

Country	Variable Type	Number	Mean	SD	Skewness	Kurtosis
China	GPR	317	0.008	0.382	0.442	0.836
	EPU	317	0.002	0.460	0.066	1.092
	GDP	88	0.012	0.431	−0.487	0.846
	UNEMPLOYMENT	88	0.010	0.147	−0.619	3.615
	CPI	88	−0.001	0.006	−0.514	1.992
	STOCK	88	0.005	0.076	−0.125	1.903
USA	GPR	317	0.003	0.258	2.189	8.740
	EPU	317	0.003	0.269	0.499	1.678
	GDP	88	0.006	0.028	−0.689	−0.135
	UNEMPLOYMENT	88	0.021	0.271	1.084	0.789
	CPI	88	0.005	0.007	−1.003	5.295
	STOCK	88	0.013	0.086	−0.992	0.993
Japan	GPR	317	−0.001	0.533	0.805	2.901
	EPU	317	−0.000	0.190	−0.091	0.920
	GDP	88	0.005	0.025	−2.059	5.796
	UNEMPLOYMENT	88	0.040	0.011	−0.351	−0.683
	CPI	88	0.003	0.030	1.045	2.605
	STOCK	88	0.001	0.111	−0.766	0.262
India	GPR	317	0.003	0.486	0.962	3.715
	EPU	317	0.001	0.447	−0.139	0.831
	GDP	88	0.006	0.288	0.078	2.647
	UNEMPLOYMENT	88	0.019	0.003	−1.693	1.580
	CPI	88	0.002	0.206	0.105	0.438
	STOCK	88	0.029	0.119	−0.269	1.475
Hong Kong	GPR	317	−0.003	0.842	0.320	0.869
	EPU	317	−0.001	0.463	0.160	0.367
	GDP	88	0.028	0.041	−0.874	1.038
	UNEMPLOYMENT	88	−0.011	0.105	1.115	4.486
	CPI	88	0.054	0.995	−0.312	0.778
	STOCK	88	0.007	0.103	−0.459	0.874
Macao	GPR	n/a	n/a	n/a	n/a	n/a
	EPU	317	0.019	0.477	−0.031	0.434
	GDP	88	0.007	0.008	0.147	0.410
	UNEMPLOYMENT	88	−0.000	0.002	−0.134	0.863
	CPI	88	0.003	0.115	1.084	5.096
	STOCK	88	0.044	0.299	1.240	7.316

Note: GPR data are not available for Macao.

’s descriptive statistics illustrate this broad deterioration. A rise in domestic EPU coincides with a statistically significant unemployment rate increase (SD: China 0.147, U.S. 0.271, Japan 0.011, India 0.003, Hong Kong 0.105, and Macao 0.002), after adjusting for other factors. EPU shocks also disrupt consumer price stability, raising price volatility. The consumer price index (CPI) standard deviation rises sharply in the U.S. (0.007), China (0.006), Japan (0.030), India (0.206), Hong Kong (0.995), and Macao (0.115). These results establish the domestic role of EPU before analyzing international transmissions.

Financial markets react negatively to policy uncertainty, as indicated by a causal deterioration in investor sentiment. Table 1 reports negative skewness in China’s (−0.125) and the U.S.’s (−0.992) stock markets, reflecting a higher probability of extreme negative returns due to EPU-induced shocks. India’s stock market, in turn, responds with higher return volatility (SD = 0.119). Granger causality tests within the VAR framework confirm that increases in domestic EPU cause these negative effects. In China, a rise in domestic EPU Granger-causes GDP declines. Additionally, the Granger causality in quantiles test shows that domestic EPU’s effect is stronger in lower quantiles (e.g., during downturns) than in median or upper quantiles, underscoring that policy uncertainty is especially critical during stress periods.

4.1.2. The International Transmission of Uncertainty and GPR

Moving from the domestic channel, the analysis identifies a hierarchical and uneven pattern in the transmission of uncertainty globally. The U.S. and China primarily shape these flows.

Table 2. (a). Pearson correlation among countries (EPU-level). (b). Pearson test of major countries (Changes in EPU).

(a)
Country	Mean	SD	Macao	Hong Kong	USA	China	Japan	India	Global
Macao	138.519	76.936	1
Hong Kong	135.702	72.398	0.483 **	1
USA	124.041	45.337	0.481 **	0.433 **	1
China	138.624	114.206	0.483 **	0.407 **	0.570 **	1
Japan	107.184	32.364	0.382 **	0.408 **	0.622 **	0.286 **	1
India	92.060	46.930	0.181 **	0.340 **	0.460 **	−0.030	0.565 **	1
Global	140.387	71.218	0.629 **	0.550 **	0.780 **	0.871 **	0.533 **	0.177 **	1
(b)
Country	Mean	SD	China-EPU	Japan-EPU	India-EPU	USA-EPU
China-EPU	0.005	0.423	1.000
Japan-EPU	0.000	0.368	0.125	1.000
India-EPU	−0.002	0.190	0.023	0.343	1.000
USA-EPU	0.003	0.269	0.244 **	0.217	0.199 **	1.000

** p < 0.01.

a details a hierarchy of vulnerability: sub-core economies (SD: Japan 32.3, India 46.9, Hong Kong 72.3, and Macao 76.9) have lower EPU volatility than China (SD = 114.2), whose volatility is over 150% above that of the U.S. (SD = 45.3). This hierarchy is also supported by cross-country EPU correlations. The clear positive relationship between U.S. and Chinese EPU (ρ = 0.244, Table 2b) is a major driver of global co-movement, connecting to the subsequent discussion of spillovers.

Spillovers of uncertainty are asymmetric, as Granger causality in quantiles tests reveal one-way causal influence: shocks to the U.S. EPU cause changes in China’s macroeconomic variables, but not the reverse. This spillover reaches its peak when Chinese markets decline. Conversely, China’s GPR can causally transmit shocks upstream, impacting the U.S. variables directly. The diagnostic tests for model adequacy, presented in

Table 3. Portmanteau test results.

Table Impact Country	Impacted Country	H0	χ2	df	p	5% Critical Value
Hong Kong	Hong Kong	No Auto	33.570	25	0.117	37.652
China	China	No Auto	35.806	25	0.075	37.652
USA	USA	No Auto	36.208	25	0.069	37.652
Macao	Macao	No Auto	36.217	25	0.068	37.652
Japan	Japan	No Auto	36.239	25	0.068	37.652
India	India	No Auto	27.171	25	0.347	37.652
China	Macao	No Auto	46.476	36	0.113	50.998
USA	Macao	No Auto	46.498	36	0.113	50.998

, confirm the robustness of these findings.

GPR transmission patterns confirm the global causal hierarchy.

Table 4. Correlation of the geopolitical risk of China, the U.S., Japan, and India.

	Mean	SD	China-GPR	Japan-GPR	India-GPR	USA-GPR
Pearson test
China-GPR	0.008	0.382	1.000
Japan-GPR	−0.001	0.533	0.628	1.000
India-GPR	0.003	0.486	0.438	0.386	1.000
USA-GPR	0.003	0.258	0.587	0.546	0.456	1.000
Spearman test
China-GPR	0.008	0.382	1.000
Japan-GPR	−0.001	0.533	0.569	1.000
India-GPR	0.003	0.486	0.374	0.302	1.000
USA-GPR	0.003	0.258	0.555	0.449	0.390	1.000
Kendall test
China-GPR	0.007	0.382	1.000
Japan-GPR	−0.001	0.533	0.408	1.000
India-GPR	0.003	0.486	0.255	0.202	1.000
USA-GPR	0.003	0.258	0.397	0.309	0.264	1.000

’s correlation analysis shows that the U.S. exerts the strongest causal links with other nations, such as 0.587 with China, 0.546 with Japan, and 0.456 with India, establishing the U.S. as the main node in the global GPR network. The interrelationship between geopolitical risk and economic policy uncertainty within each country is further detailed in

Table 5. Correlation of the GPR and EPU of China, the U.S., Japan, and India.

	China-GPR	Japan-GPR	India-GPR	USA-GPR
Pearson test
China-EPU	0.208 **	0.107	0.085	0.023
Japan-EPU	0.035	0.044	−0.117	−0.116
India-EPU	−0.034	0.019	−0.111	−0.004
USA-EPU	0.086	0.142	−0.017	0.085
Spearman test
China-EPU	0.137 *	0.024	0.063	0.007
Japan-EPU	0.022	0.012 *	−0.116 *	−0.141
India-EPU	−0.026	0.033	−0.123 *	0.014
USA-EPU	0.039	0.085	−0.054	−0.028
Kendall test
China-EPU	0.089 *	0.014	0.042	0.002
Japan-EPU	0.013	0.008	−0.079 *	−0.089 *
India-EPU	−0.017	0.023	−0.084 *	0.008
USA-EPU	0.025	0.057	−0.037	−0.017

* p < 0.05, ** p < 0.01.

, which presents the correlation of GPR and EPU for China, the U.S., Japan, and India. The coupling of economic and geopolitical forces creates “asymmetric interdependence”: the U.S. primarily exports EPU, while China causally drives GPR independently. These empirical findings are supported by stationary time series data, as confirmed by the Augmented Dickey-Fuller test results in

Table 6. Augmented Dickey–Fuller (ADF) test results.

Variable	$\mathit{t}$-Value	$\mathit{\rho }$-Value	Critical Values
			1%	5%	10%
USA-GPR	−6.553	0.000 ***	−3.450	−2.870	−2.571
USA-EPU	−2.948	0.040 **	−3.450	−2.870	−2.571
China-GPR	−9.179	0.000 ***	−3.450	−2.870	−2.571
China-EPU	−17.394	0.000 ***	−3.450	−2.870	−2.571
China-GDP	−4.083	0.001 ***	−3.451	−2.871	−2.572
China-UNEMPLOYMENT	−4.151	0.001 ***	−3.451	−2.870	−2.572
China-CPI	−4.687	0.000 ***	−3.451	−2.870	−2.572
China-STOCK	−7.827	0.000 ***	−3.450	−2.870	−2.571

Notes: *** 1%, ** 5%.

. The detailed outcomes of the Granger causality in quantiles tests that underpin the analysis of asymmetric spillovers are presented in

Table 7. Granger causality in quantiles test results.

Independent Variable	USA- GPR	USA-EPU	China-GPR	China-EPU	China-GDP	China- UNEMPLOYMENT	China-CPI	China-STOCK
USA-GPR	---	Y	Y	---	---	---	---	---
USA-EPU	Y	---	Y	Y	Y	Y	Y	N
China-GPR	Y	Y	---	Y	N	Y	Y	Y
China-EPU	Y	N	N	---	Y	Y	Y	Y
China-GDP	Y	N	Y	N	---	N	Y	Y
China-UNEMPLOYMENT	Y	N	Y	Y	N	---	Y	N
China-CPI	Y	Y	Y	Y	Y	Y	---	Y
China-STOCK	Y	Y	Y	Y	Y	Y	Y	---

Notes: Y indicates acceptance of the Granger causal null hypothesis, that is, not the interpretive change. N indicates the rejection of the null hypothesis. --- indicates that the test was not performed, either because a variable cannot Granger-cause itself or due to the specific design of the test.

.

4.1.3. Robustness Checks and Model Validation

A Variance Inflation Factor (VIF) check shows severe multicollinearity among some explanatory variables. Out of thirty-seven variables, thirteen, primarily CPI, EPU, and stock market indicators, exhibited VIF values exceeding the critical threshold of 10 (Belsley et al., 1980), ranging from 10.99 to 245.67 (see

Table A1. Analysis of the Variance Inflation Factor (VIF).

Variable Category	Number of Variables	VIF Range	Severity Level	Economic Interpretation
CPI	5	98.65–245.67	Extreme	Global inflation synchronization and price transmission mechanisms
STOCK	5	18.54–32.65	Severe	Financial market integration and contagion effects
EPU	3	10.99–15.88	Moderate	Cross-country policy uncertainty spillovers
Other variables	24	2.11–9.88	Acceptable	No significant multicollinearity concerns

Source: Prepared by the authors based on econometric data.

in Appendix C). This aligns with global economic and financial synchronization. To address this, we used Principal Component Analysis (PCA), as proposed by Stock and Watson (2002). PCA transforms highly correlated variables into orthogonal components that capture shared variance. By retaining only the most informative components, we reduce multicollinearity while preserving key economic details.

We tested for endogeneity using the Durbin–Wu–Hausman test (Hausman, 1978). The test strongly rejects exogeneity (p = 0.0046) (

Table 8. Comprehensive diagnostic test results and methodological implications.

Diagnostic Concern	Test Method	Test Results	Statistical Threshold	Interpretation
Multicollinearity	Variance Inflation Factor (VIF)	13 out of 37 variables exceed thresholds	VIF > 10 indicates severe multicollinearity	Global synchronization of economic policies and financial market integration
Endogeneity	Durbin–Wu–Hausman Test	t-statistic for residual coefficient: 2.8764 p-value: 0.0046	p < 0.05 rejects exogeneity	Conclusion: Reject null hypothesis— USA-EPU variable appears endogenous; supports simultaneous determination theory
Heteroscedasticity	White Test	Multiple specifications: Model 1: p = 0.0048 Model 2: p = 0.0032 Model 3: p = 0.0108	p < 0.05 indicates heteroscedasticity	Volatility clustering characteristics of financial time series
Functional Form/ Nonlinearity	Ramsey RESET Test & Quantile Diagnostics	Evidence of nonlinear relationships and state- dependent effects	Rejection of linear specification	Relationships vary across economic regimes and market conditions
Instrument Validity	First-Stage F-test & Sargan Test	First-stage F-statistic: 28.45 Sargan test p-value: 0.2663	F > 10: Strong instruments p > 0.05: Valid instruments	Instruments are relevant and exogenous

), indicating clear causal endogeneity in the USA-EPU variable. This aligns with theoretical views on how policy uncertainty and financial markets influence each other. To address this and obtain unbiased causal parameter estimates, we employed an Instrumental Variables (IV) method with Two-Stage Least Squares (2SLS) estimation. The endogenous USA-EPU is instrumented with Global-EPU and Japan-EPU.

Table A2. Correlations matrix for Endogeneity Diagnostics.

Variable	USA- STOCK	USA-EPU (Endogenous)	Global-EPU (Instrument)	Japan-EPU (Instrument)	USA-GPR	USA- CPI	USA- UNEMPLOYMENT	China-EPU
USA-STOCK	1.000
USA-EPU	−0.345	1.000
Global-EPU	0.234	0.645	1.000
Japan-EPU	0.267	0.598	0.723	1.000
USA-GPR	−0.289	0.567	0.456	0.445	1.000
USA-CPI	−0.456	0.389	0.298	0.321	0.412	1.000
USA-UNEMPLOYMENT	−0.412	0.276	0.198	0.223	0.309	0.432	1.000
China-EPU	0.198	0.523	0.678	0.589	0.387	0.254	0.167	1.000

Source: Prepared by the authors based on econometric data.

in Appendix C shows that these instruments are causally associated with the endogenous regressor (correlations: 0.645 and 0.598), satisfying the relevance condition. Their much lower correlations with the US financial outcome, USA-STOCK (0.234 and 0.267), tentatively fulfill the exclusion restriction. Both instruments pass validity checks: the first-stage F-statistic is 28.45 (well above the threshold of 10 established by Staiger and Stock (1997)), and the Sargan test (p = 0.2663) confirms they are not correlated with the error term, supporting the validity of the IV model for causal inference.

The White test for heteroscedasticity shows it is present across many model types (p-values: 0.0032 to 0.0108; Table 8). This fits the common pattern of volatility clustering in financial time series (Engle, 1982). As a result, we do not use ordinary least squares standard errors. Instead, we report results using the White (1980) heteroscedasticity-consistent estimator. This yields strong t-statistics, confidence intervals, and p-values even when the variance is not constant. Additionally, the Ramsey RESET test and initial quantile analyses reveal possible nonlinearity (Ramsey, 1969). This supports the use of quantile regression to capture changes that may be missed by mean models.

The integrity and reliability of the dynamic models are confirmed through extensive validation checks. The stability of each VAR system, essential for valid impulse response analysis, is verified via AR root graphs (see Appendix A, Figure A1, Figure A2, Figure A3 and Figure A4), confirming that all eigenvalues for China, the U.S., Hong Kong, and Macao lie within the unit circle. The model specification is refined by selecting the optimal lag length based on established information criteria. Residual diagnostics indicate no autocorrelation through Portmanteau tests (e.g., Hong Kong: χ² = 33.570, p-value = 0.117), as shown in Table 3, and confirm stationarity across all variables using Augmented Dickey–Fuller tests (p < 0.01) in Table 6.

Robustness in correlation and regression analyses is ensured through methodological safeguards. Correlation analyses between EPU and GPR, detailed in Table 5, reveal cross-country heterogeneity, showing a strong positive correlation in China (ρ = 0.208), a moderate correlation in the U.S. (ρ = 0.086), and a negative association in India (Kendall’s τ = −0.084), indicating that the analyses remain free from specification biases. Heteroscedasticity-robust standard errors are used for regression-based inferences to correct for non-constant variance. The orthogonal impulse response functions, illustrated in Figure A5, Figure A6, Figure A7, Figure A8, Figure A9 and Figure A10 in Appendix B, quantify the dynamic impact of shocks across the system.

4.1.4. Global Financial Landscape and Risk Analysis

Bringing together all empirical findings reveals a multilayered global structure of uncertainty transmission that links the domestic and international vulnerabilities of the modern economy. First, as demonstrated in previous sections, domestic EPU universally imposes systemic risk by undermining macroeconomic stability through suppressed output, higher unemployment, and financial market stress. This persistent effect, observed in all economies studied, underscores the broad economic costs of policy ambiguity that impede decision-making.

The international analysis further highlights a core-periphery hierarchy. The U.S. emerges as the main transmitter of economic policy shocks, while China acts both as a recipient of the U.S.-induced shocks and as a source of geopolitical risk. This asymmetric interdependence, detailed in earlier sections, creates a multi-tiered global risk structure. Shocks from central economies can be amplified by financially interconnected hubs. Therefore, global investors must consider the influence of the U.S. EPU and China’s GPR distinctly, linking empirical evidence to risk management strategies for regional and sectoral exposures.

4.2. Discussion of the Results

4.2.1. The Domestic Channel: EPU as a Systemic Risk Factor

Domestic economic policy uncertainty (EPU) undermines macroeconomic stability in both advanced and emerging economies. This validates the “cognitive rigidity” channel caused by policy ambiguity (Bloom, 2009; Baker et al., 2016). This represents a causal relationship: an opaque policy environment prompts widespread caution. When policy is unclear, firms delay investments, hiring, and capital spending (Dixit & Pindyck, 1994). Households spend less and save more (Bernanke, 1983). This hesitation lowers GDP, increases unemployment, and creates price instability. Financially, asset pricing and risk management must account for increased volatility and negative skewness in returns (Table 1), as investors demand a higher risk premium for EPU (Engle, 2002). Granger causality tests show that EPU has a stronger impact at lower quantiles, revealing that risk rises much more during downturns. This state-dependent effect means risk models must be nonlinear; constant-volatility models miss tail risk.

4.2.2. The International Channel: A Hierarchy of Asymmetric Interdependence

The international spillover results reveal a core-periphery system (Forbes & Warnock, 2012) defined by asymmetric interdependence. Unidirectional spillovers from the U.S. EPU to China position the U.S. as the global financial anchor. The U.S. spreads uncertainty through supply chains, financial markets, and dollar dominance. In contrast, China can now project GPR upstream to influence the U.S., demonstrating its evolution into a geostrategic shaper. This reflects a modern reality: economic leadership and geopolitical influence are now distinct global risk channels. Asymmetric interdependence creates a complex environment by diversifying sources of systemic risk. Financial stability now depends not just on the policies of traditional core economies but also on geopolitical events from rising powers.

The financial implications of this asymmetry are significant. Addressing them requires a two-step approach to global risk. For top-down allocation, use the EPU of core countries, particularly the U.S., as a primary leading indicator. When the U.S. EPU rises, it reduces risk in vulnerable markets with strong trade and financial ties. For bottom-up regional or sector planning, rely on the GPR of core countries and the EPU of peripheral ones. China’s GPR is important for pricing assets in sectors like commodities, defense, and shipping, as well as for evaluating companies linked to Asian supply chains. High EPU volatility in major emerging markets like China and India (Table 2a) justifies applying higher country-risk premiums in their equity and bond markets. This adjustment reflects added, non-diversifiable risk for investors.

5. Conclusions and Policy Implications

This study has developed and validated a hierarchical framework for analyzing how EPU and GPR move through the global economy. Using a robust econometric approach, we find three key results. First, higher domestic EPU consistently hurts macroeconomic performance. It is linked to higher unemployment, increased price volatility, and less stable financial markets in both advanced and emerging economies. This indicates that EPU is a powerful, unavoidable risk stemming from national policies.

Second, our analysis shows that uncertainty between countries does not spread evenly in both directions but follows a core-periphery pattern. Shocks from core economies, especially the U.S., primarily flow into peripheral countries. A key detail is the “asymmetric interdependence” between the U.S. and China. The U.S. is the main sender of EPU, but China has become a significant source of GPR and can transmit shocks back that affect U.S. economic outcomes.

The study makes three primary contributions. Methodologically, it demonstrates the effectiveness of combining VAR modeling with quantile-based Granger causality tests to capture both average and dynamic transmission patterns. Theoretically, it elucidates the complex, nonlinear nature of international spillovers and introduces the concept of asymmetric interdependence between economic and geopolitical risks to the existing literature. Practically, it identifies key transmission nodes and offers a refined map of risk pathways for enhanced management.

These findings carry significant implications for both policy and investment strategies. Policymakers in peripheral countries must contend with the risk of imported uncertainty from core nations. They should closely monitor trends in core-country EPU and implement fiscal and monetary protections to bolster their economies. Investors and asset managers are encouraged to adopt a multi-layered risk approach, utilizing U.S. EPU for global decisions and GPR alongside peripheral-country EPU for regional and sectoral considerations. This strategy facilitates more accurate pricing of risk.

Despite these contributions, this study is subject to limitations that suggest avenues for future research. The geographical focus on the U.S., China, and select Asian economies may restrict the generalizability of the findings to other regions, such as Europe or Latin America. Methodologically, while robust, the framework could be expanded through network analysis or dynamic spatial econometric models to better elucidate the complex topology and time-varying nature of the identified contagion networks. Additionally, as noted by Chen et al. (2024), using keywords to construct EPU and GPR indices can introduce errors. Future research could employ more advanced methods to obtain more accurate measures. Finally, investigating sector-specific vulnerabilities to these spillovers would provide valuable micro-level insights for corporate and financial risk management.