Economic policy uncertainty and the predictability of stock returns: impact of China in Asia

Abstract

Purpose

Motivated by the significant role of uncertainty in affecting investment decisions and China's economic leadership in Asia, this paper investigates the predictive role of exposure to Chinese economic policy uncertainty at the individual stock level in large Asian markets.

Design/methodology/approach

We estimate the monthly uncertainty exposure (beta) for each stock and then employ the portfolio-level sorting analysis to investigate the relationship between the China’s uncertainty exposure and the future returns of major Asian markets over multiple trading horizons. The raw returns of the high-minus-low portfolios are then adjusted using conventional asset pricing models to investigate whether the relationship is explained by common risk factors. Finally, we check the robustness of the portfolio-level results through firm-level Fama and MacBeth (1973) regressions.

Findings

Applying portfolio-level sorting analysis, we reveal that exposure to Chinese uncertainty is negatively related to the future returns of large stocks over multiple trading horizons in Japan, Hong Kong and India. We discover this is unexplained by common risk factors, including market, size, value, profitability, investment and momentum, and is robust to the specification of stock-level Fama and MacBeth (1973) regressions.

Research limitations/implications

Our analysis demonstrates the spillover effects of Chinese economic policy uncertainty across the region, provides evidence of China's emerging economic leadership, and offers trading strategies for managing uncertainty risks.

Originality/value

The findings of the study significantly improve our understanding of stock return predictability in Asian markets. Unlike previous studies, our results challenge the leading role of the US by providing a new intra-regional return predictor, namely, China’s uncertainty exposure. These results also evidence the continuing integration of the Asian economy and financial markets. However, contrary findings for some Asian markets point toward certain market-specific features. Compared with market-level research, our analysis provides deeper insights into the performance of individual stocks and is of particular importance to investors and other market participants.

Citation

Chen, X., Li, B., Singh, T. and Worthington, A.C. (2024), "Economic policy uncertainty and the predictability of stock returns: impact of China in Asia", China Accounting and Finance Review, Vol. 26 No. 5, pp. 645-679. https://doi.org/10.1108/CAFR-11-2023-0144

1. Introduction

After rebounding from the 1997 Asian financial crisis, Asia has rapidly become a dominant force in the global economy (Goto, Endo, & Ito, 2021). Figure 1 illustrates the significant growth in the Gross Domestic Product (GDP) of the East and South Asian and Pacific regions, measured in trillions of US dollars, over the past two decades. This expansion has led to the region now accounting for a substantial portion of the world's economy.

Concomitantly, stock markets in Asia are now one of the focal points of attraction for investors around the globe. As elsewhere, these investment decisions hinge heavily on stock market performance in terms of both current and future expected returns. In this regard, the conventional intertemporal capital pricing model (ICAPM) of Merton (1973) assumes that investors hedge against unfavorable risks by adjusting consumption and investment based on the shifts in future economic conditions and investment opportunities over the long run (Chen, Li, Worthington, & Singh, 2024).

These hedging needs regarding economic turbulence are argued to drive stock market performance and risk, and macroeconomic indicators that predict the future economic and investment conditions are considered useful return predictors under the ICAPM (Rossi & Timmermann, 2015; Chen et al., 2024). Consequently, the empirical literature is replete with the repeated use of some standard factors and their variants to predict stock returns. Some of these studies investigate factors at the aggregate market level, while others consider factors related to specific asset characteristics.

Of these, uncertainty about economic conditions and policy environment remains one of the most important factors governing investment decisions. Many studies have found that economic policy uncertainty reshapes both the macroeconomic and financial environment (Ko & Lee, 2015; Caggiano, Castelnuovo, & Figueres, 2017; Chen, Jiang, & Tong, 2017). Economic policy-related risk has also been found to be closely related to uncertainty in the aggregate economy, where newly issued policies or political events can trigger uncertainty in the economic environment, while high economic uncertainty can result in policy changes (Baker, Bloom, & Davis, 2016). Therefore, economic policy uncertainty has been widely used as one of the measures for capturing aggregate economic uncertainty risk, especially since Baker et al. (2016) developed an index based on newspaper coverage frequency, focusing particularly on policy-related economic uncertainty.

The role of uncertainty in the aggregate economy has been recognized in many studies, including Jones and Olson (2013), Caggiano et al. (2017), and Bahmani-Oskooee and Nayeri (2018). However, less effort has been devoted to examining its role in predicting stock returns. This is important as higher levels of economic policy uncertainty raise concerns about investment conditions, potentially postpone investment, reduce output and employment (Baker et al., 2016), push down aggregate stock market returns (Brogaard & Detzel, 2015), and perhaps cause large recessions and economic downturns (Karnizova & Li, 2014).

Although limited in number, these studies have typically revealed a negative relationship between economic policy uncertainty and stock market returns in many Asian markets (Dzielinski, 2012; Ko & Lee, 2015; Li, Balcilar, Gupta, & Chang, 2016; Chen et al., 2017; Phan, Sharma, & Tran, 2018), where uncertainty could arise from both exogenous and policy-induced shocks to demand and supply. In a more contemporary setting, the recent COVID-19 pandemic has triggered uncertainty around the globe, adversely affected investment, and raised concerns about the future economy (Baker, Terry, Bloom, & Davis, 2020).

Of course, it is impossible to talk of Asia without highlighting the leading role of China in information diffusion. Policy-related uncertainty is particularly important in China, given the predominant role of the Chinese government in determining its macroeconomic conditions and developments (Paul & Mas, 2016). Economic policy uncertainty shocks in China are easily transmitted to neighboring markets via regionwide trade, foreign direct investment (FDI) flows (Tong, Chen, Singh, & Li, 2022), and financing (Nguyen, Kim, & Papanastassiou, 2018). Since the 1997 Asian financial crisis, many Asian economies have moved away from their close relationships with Western markets and have instead strengthened their economic and financial integration within the Asian region (Das, 2012). Institutional frameworks like the Association of Southeast Asian Nations (ASEAN) and the Asia-Pacific Economic Cooperation (APEC), alongside increasing regional trade agreements and initiatives like the Regional Comprehensive Economic Partnership (RCEP), have enhanced regional economic and financial integration (Yuan, 2010; Acharya, 2011; Australian Government, 2020). China actively engages in these organizations for regional economic cooperation and thereby plays a leading role in the region (Wong, 2013).

China’s remarkable economic growth likewise has been pivotal, driving economic development and integration in the region (Jorgenson & Vu, 2011; Pasierbiak, 2019; Goto et al., 2021), with the Chinese economy growing much faster than that of the US over the past few years (Morrison, 2019; Li, 2021). The 2008 global financial crisis (GFC) further raises questions about the sustainability of the US’s leading role in the world economy (Ba, 2014). Being the largest economy in Asia and the second largest in the world, China is now ranked highest for manufacturing, international trading, and foreign exchange reserves (Morrison, 2019). Most investors therefore believe that China has provided new regional leadership in Asia, while the engagement of Asian markets with the US has consequently weakened (Tang, Thuzar, Hoang, Chalermpalanupap, Pham, & Saelaow, 2019). For this reason, Asian stock markets tend to exhibit strong comovements in light of the increased economic and financial market integration within the region (Awokuse, Chopra, & Bessler, 2009; Fang & Bessler, 2018), such that uncertainty risk in China, not the US, increasingly drives market risk in many Asian markets (Tsai, 2017).

Motivated by the significant role of uncertainty in affecting investment decisions, stock market returns, macroeconomic conditions, and the leading role of China in Asia, this paper investigates the predictive role of exposure to economic policy uncertainty in China at the individual stock level in other large Asian stock markets. In doing so, the analysis contributes to the extant literature on cross-sectional return predictability in the following ways. First, existing research has only been able to document how China plays a leading role in the Asian economic development and integration via regionwide trading, FDI, and regional economic cooperation. No study to date has assessed the predictive power of China’s economic policy uncertainty in the context of stock returns, except for recent work by Balcilar, Gupta, Kim, and Kyei (2019), which focuses on the effects of global economic uncertainty, including uncertainty in the US, Japan, China, and Europe, on Asian stock market returns.

Such market level analysis is, however, not informative of firm-level performance and risks. Our study is then the first to investigate the predictive role of exposure to Chinese economic policy uncertainty for individual stock returns in major Asian markets. Compared with market-level research, our analysis provides deeper insights into the performance of individual stocks and is of particular importance to investors and other market participants.

Second, previous studies concluding a close relationship between uncertainty risk and stock returns in China have focused narrowly on the domestic economy (Pástor & Veronesi, 2013; Brogaard & Detzel, 2015; Chen et al., 2017). Given the leading role of China in Asia, its own policy uncertainty also tends to spread across other markets in the region. Our study is then the first to extend the existing analysis to a broader set of other major Asian stock markets, including the Japanese, Hong Kong, Indian, South Korean, and Taiwan stock markets. Along with the Shanghai and Shenzhen stock markets in mainland China, these markets account for most Asian stock market capitalization (Yao, Wang, Zhang, & Ou, 2016; Hong, Lee, Liao, & Seneviratne, 2017; Lee & Yin, 2017).

Third, as the US has been and continues to be the world’s largest economy, the existing literature has long concentrated on the spillover effects of the US uncertainty across many markets around the world (e.g. Chuliá, Gupta, Uribe, & Wohar, 2017; Das & Kumar, 2018; Wen, Li, Chen, & Singh, 2023). The remarkable growth of the Chinese economy, however, has challenged the dominant role of the US economy in Asia, particularly in recent decades (Tsai, 2017). Asian stock markets are also less exposed to global risks and less related to Western markets, including the US (Aityan, Ivanov-Schitz, & Izotov, 2010; Chien, Lee, Hu, & Hu, 2015; Tsai, 2017). Unlike previous studies, we investigate the predictive power of exposure to Chinese uncertainty in major Asian markets, providing novel evidence for the emerging leadership of China in the region. Finally, unlike existing methods that myopically focus only on the next month’s return predictability, we evaluate the relationship between exposure to China’s economic policy uncertainty and future stock returns over multiple trading horizons as long as 24 months. This enables investors to better identify the most profitable trading windows for hedging their uncertainty risks.

Our portfolio-level sorting analysis reveals that exposure to China’s economic policy uncertainty is negatively related to future returns for large stocks in the Japanese, Hong Kong, and Indian stock markets over multiple trading horizons. This predictive power is unexplained by common risk factors in prevailing asset pricing models, including market, size, value, profitability, investment, and momentum. These findings remain consistent when using firm-level Fama and MacBeth (1973) regressions. We argue that this negative relationship is well explained by the ICAPM theory (Merton, 1973), which postulates that conservative investors are willing to pay higher prices and accept lower expected returns by holding positive-beta stocks to hedge uncertainty risks (Chen, Li, Worthington, & Singh, 2021, 2024). Interestingly, exposure to China’s uncertainty is positively related to stock returns in South Korea, although this apparent relationship is explained by common risk factors, particularly the value factor.

The findings of the study significantly improve our understanding of stock return predictability in Asian markets. Unlike previous work, our results challenge the leading role of the US by providing a new intra-regional return predictor, namely, China’s uncertainty exposure. These results also evidence the continuing integration of the Asian economy and financial markets. Moreover, contrary findings for some Asian markets point toward certain market-specific features and highlight the need to provide additional robustness checks in different markets when investigating stock return predictors, with investors needing to incorporate market characteristics into their investment decision-making.

The remainder of the study is organized as follows. Section 2 discusses economic policy uncertainty and provides a brief overview of the leading role of China in the Asian economy. Section 3 discusses the sources of data and definitions of the variables used in the model, while Section 4 specifies the model and discusses the estimation strategy. Section 5 presents and analyzes the empirical results for which Section 6 provides some robustness checks. Section 7 concludes.

2. Economic policy uncertainty and integration in Asia — the leading role of China

Economic policy uncertainty affects future investment opportunities and drives stock market performance (Brogaard & Detzel, 2015). Some studies investigate the relationship between such uncertainty, as measured by the economic policy uncertainty (EPU) index in Baker et al. (2016), and aggregate stock market returns in Asia, where it is shown that uncertainty drags down future stock market returns in China (Chen et al., 2017), Japan, and South Korea (Phan et al., 2018). Conversely, Li et al. (2016) find only a weak relationship between domestic uncertainty and stock market returns in China and India, while Balcilar et al. (2019) conclude that domestic and global uncertainty do not drive aggregate stock market returns in either Hong Kong or Malaysia. However, the global EPU exhibits a stronger predictive power than the domestic one for South Korea (Balcilar et al., 2019).

The rise of the Chinese economy has played a significant role in economic growth and the integration in Asia. Over the past two decades, China has actively engaged in intra-regional trading, with its growing importance both as a supplier and as a consumer market (Pasierbiak, 2019; Goto et al., 2021). Its influence is exemplified by its overtaking of the US as the largest trading partner for India (Batra, 2011).

FDI also strengthens China's economic power (Tong et al., 2022). China's FDI inflows surpass those of other Asian economies, contributing to its rapid economic growth (Yao et al., 2016). Neighboring markets relocate industries to mainland China for reasons of improved cost efficiency (Kim, Driffield, & Temouri, 2016; Lee & Yin, 2017). Conversely, FDI outflows from China have increased significantly and have overtaken its FDI inflows since 2016 (Yu, Qian, & Liu, 2019). Asia is the most important recipient of China’s outward FDI, driving economic development across the region (Yao et al., 2016).

China likewise actively engages in economic cooperation in the region, such as signing the Tripartite Cooperation Agreement with South Korea and Japan (Wong, 2013). Despite strong competition and potential conflict, bilateral cooperation between China and India has grown over the past few decades. For example, China has pledged to invest in pharmaceuticals, IT, industrial parks, and high-speed rail infrastructure in India (Devadason, 2012). Both have also engaged in multilateral forums and organizations, such as the South Asian Association for Regional Cooperation (SAARC) (Kumar, 2015).

With its strong economic power in Asia, China plays a leading role in information diffusion in the region. Uncertainty shocks in China are easily transmitted to other financial markets in Asia (Jayasuriya, 2011; Fang & Bessler, 2018). For example, Chinese market uncertainty, as measured by stock market volatility, has been transmitted significantly to neighboring markets since 2005, and this has increased with the continuing rise of the Chinese economy (Zhou, Zhang, & Zhang, 2012). China has also increasingly had a much stronger impact on market crashes in most Asian markets than traditional regional leaders, such as Japan (Fang & Bessler, 2018).

Being the largest economy in Asia, China is now ranked highest in manufacturing, international trading, and foreign exchange reserves (Morrison, 2019). The remarkable growth of the Chinese economy has also challenged the leading role of the US economy in Asia, particularly in recent decades (Tsai, 2017). The contagious effect of China’s economic policy uncertainty is much stronger than that of the US uncertainty in many Asian stock markets (Tsai, 2017). Most Asian members now believe that their relationships with the US have been weakened by the newly established regional leadership of China (Tang et al., 2019).

Although economic conditions and policy environment in China significantly drive stock market performance and risk in Asia, no study has yet assessed the predictive power of Chinese economic policy uncertainty for individual stock returns in Asian markets. Our study addresses this gap and investigates the relationship between exposure to Chinese uncertainty and future individual stock returns in major Asian markets, including the Japanese, Hong Kong, Indian, South Korean, and Taiwan stock markets. Together, these markets account for a major share of Asian stock market capitalization, and their economies are closely linked to economic conditions and policy changes in China. For instance, China has developed its Asian production networks with neighboring developed markets, such as Japan and South Korea (Hong et al., 2017), receives substantial inbound investment from Japan and South Korea (Lee & Yin, 2017), and heavily invests itself in other Asian markets (Tong, Li, & Singh, 2018; Li, Gallagher, & Mauzerall, 2020).

3. Data, variables, and summary statistics

The study uses the sample period from January 2000 to December 2020 to estimate models and investigate the relationship between exposure to Chinese economic policy uncertainty and individual stock returns in five major Asian markets, including the Japanese, Hong Kong, Indian, South Korean, and Taiwan stock markets [1]. The rationale for January 2000 as the start of the sample period is twofold. First, our measure of economic policy uncertainty discussed later is only available after January 2000. Second, although the Chinese economy began to take off as early as 1978, economic growth remained slow in the initial stages of economic reform. With the surge in economic growth and the resultant rise in the importance of its economy, China has been a hub of the Asian region since 2000. Its accession to the World Trade Organization (WTO) in 2001 further stimulated region-wide FDI flows and trading with neighboring markets, enhancing its leading role in the region and promoting cross-regional economic and financial cooperation (Pasierbiak, 2019; Goto et al., 2021).

3.1 Economic policy uncertainty in China: the EPU index

We use the EPU index of Huang and Luk (2020) to measure economic policy uncertainty in China [2]. Following the methodology of Baker et al. (2016), Huang and Luk (2020) count the frequency of articles with terms related to economic or economic policy uncertainty in ten major Chinese newspapers, and subsequently construct a monthly EPU index by taking the simple average value. Huang and Luk's (2020) index has recently been used in several related studies (He, Ma, & Zhang, 2020; Fang, Jing, Shi, & Zhao, 2021) and is widely used to capture uncertainty in macroeconomic conditions.

In this regard, high uncertainty results in new policy changes, as governments typically respond by creating new policies during turbulent economic conditions. News about economic conditions or newly implemented economic policies is frequently reported in newspapers, and thus the fluctuations in the EPU index can reflect economic uncertainty (Baker et al., 2016). Alternatively, high economic uncertainty may be an outcome of newly issued policies or political events. When new policies are introduced or political events take place, newspapers publish news of market interest. Investors then become more cautious about their consumption and investments (Bachmann, Elstner, & Sims, 2013), and prefer to hold their investment decisions to save more (Jurado, Ludvigson, & Ng, 2015). This behavior further pushes uncertainty in both macroeconomic activity and financial markets to a higher level (Baker et al., 2016). Consequently, many studies find that policy changes reshape both the macroeconomic and financial environment (Ko & Lee, 2015; Caggiano et al., 2017; Chen et al., 2017).

Policy-related uncertainty is particularly important in the economic environment of China, given the dominating role of the Chinese government in determining domestic macroeconomic conditions and developments (Paul & Mas, 2016). Many studies have found that policy-related uncertainty significantly determines many aspects of the Chinese economy, including corporate investment (Wang, Chen, & Huang, 2014), firm market values (Yang, Yu, Zhang, & Zhou, 2019), corporate innovation (Cui, Wang, Liao, Fang, & Cheng, 2021), and housing (Huang & Luk, 2020). Since China first launched its economic reforms in 1978 (Chen et al., 2017), it has constantly implemented dynamic economic reforms, and therefore, unexpected economic shocks and turbulences are closely related to policy changes in China (Chen et al., 2017; Huang & Luk, 2020).

Moreover, compared with the US, the economic power of China is particularly linked to political factors (Zhang, Lei, Ji, & Kutan, 2019). China plays a leading role in Asia through intra-regional trading, FDI, and regional economic collaboration. These cross-border activities are typically government-led and carried out by state-owned organizations controlled by government agencies. For example, the “Belt and Road Initiative” (BRI) has introduced a series of government-led economic engagements aimed at boosting China’s economic power by enhancing intra-regional investment and trading. Chinese policy banks also offer substantial financing to support foreign investments and acquisitions in the Asian region (Chan, 2017).

Figure 2 plots Huang and Luk’s (2020) EPU index for China and Baker’s et al. (2016) EPU index for the US over the period January 2000 to December 2020 [3]. As shown, both indexes move together with major global economic shocks. For instance, they bounced highly following the news of the Lehman Brothers bankruptcy around October 2008, and then increased to a peak with the rating downgrade of the US sovereign credit in August 2011. Noneconomic turbulence, such as the 9/11 attack in 2001 and the Iraq War in 2003, also impact upon economic uncertainty in the US. In contrast, economic shocks mainly drive China’s EPU index, a notable example being the domestic turbulence associated with changes in the Renminbi fixing mechanism in August 2015. Interestingly, the global stock market crash arising from the COVID-19 pandemic around early 2020 also had less impact on Chinese economic policy uncertainty than in the US. Therefore, the EPU index is an especially useful proxy for measuring economic uncertainty in China given that it reflects economic-related turbulence.

The EPU index developed by Huang and Luk (2020) provides several advantages over other EPU indexes available for China. For instance, the Chinese EPU index from Baker et al. (2016) relies on English-language newspapers in Hong Kong for its construction, resulting in a greater focus on economic conditions there than in mainland China. Huang and Luk (2020) also find that Baker’s et al. (2016) index is not particularly sensitive to economic conditions in China. Alternatively, Davis, Liu, and Sheng (2019) constructed an EPU index by counting articles in two leading newspapers in mainland China. Like Baker’s et al. (2016) index, relying on only two newspapers makes it difficult to capture overall macroeconomic turbulence, potentially leading to biased results. Huang and Luk’s (2020) index contains a broader range of information from various sources and separately reflects uncertainty in both macroeconomic outcomes (e.g. employment and industrial output) and financial market performance (e.g. stock market returns). For these reasons, we consider it a better index than the alternatives.

3.2 Stock returns

We collect monthly stock returns from DataStream. For Japan, the sample comprises stocks listed on the Japan Exchange Group (JPX). Indian stocks in the sample are from the two major exchanges, namely, the Bombay Stock Exchange (BSE) and the National Stock Exchange (NSE). For South Korea, our sample includes stocks listed on the Korea Exchange (KRX), covering both the Korea Stock Exchange (KSE) and the Korean Securities Dealers Automated Quotations (KOSDAQ) Stock Market. The sample for the Hong Kong stock market includes stocks listed on the Hong Kong Stock Exchange (SEHK), and that for the Taiwan stock market includes stocks listed on the Taiwan Stock Exchange (TWSE).

In addition to the full sample, we investigate whether the predictive role of China’s EPU exposure is sensitive to firm size, with a focus on top-100 and top-200 large stocks. The justifications for concentrating on large stocks are many. First, although many stocks listed on Asian markets are small or microcap, large stocks primarily drive these markets. In evidence, over the sample period, the shares of market capitalization of the top-200 stocks in the Japanese, Hong Kong, Indian, South Korean, and Taiwan markets are 71, 91, 89, 88, and 89%, respectively. Second, small stocks are usually undesirable for many investors, especially institutional investors, resulting in very thin trading and correspondingly low liquidity (Ko, Kim, & Cho, 2007; Kang, Lee, & Lee, 2011). By contrast, large stocks are more frequently traded and are easier to be deployed to form valid buy–sell portfolios.

Third, small stocks may contain a more unsystematic risk than large stocks (Bali, Cakici, & Levy, 2008). Economic policy uncertainty represents systematic risk, so focusing on large stocks will avoid correspondingly noisy estimates. Finally, many studies on Asian markets also focus exclusively on large stocks (e.g. Au, Peng, & Wang, 2000; Kang et al., 2011; Dharani, Hassan, & Paltrinieri, 2019). To achieve a low estimation error, we require that the stocks used to estimate the uncertainty betas have at least 24 consecutive months of monthly returns within the last three years. This criterion excludes very thinly traded small stocks. Nonetheless, it yields a sample covering about 83, 75, 91, 79, and 62% of total market capitalization for the Japanese, Hong Kong, Indian, South Korean, and Taiwan stock markets, respectively.

Table 1 details the mean, volatility, and skewness of excess monthly returns for each of the five markets for each year of the sample. Using the data set starting in January 2000 and the three-year estimation window, we identify the valid stocks that satisfy this requirement since January 2003. The risk-free rates used to calculate excess returns are sourced from Bloomberg or the Federal Reserve Bank of St. Louis, except for the Taiwan market, where they are sourced from DataStream. Returns are measured in logarithmic form, which allows for additive calculations over multiple time windows. Returns are also expressed in local currency, as our primary objective is to identify a potential return predictor of local investors in these markets.

Overall, the returns in all five markets decreased significantly during the 2008 GFC, with India exhibiting the most negative monthly excess return. Compared with other markets, excess stock returns in Japan show the lowest volatility, indicating that stock returns in this market are less risky. The values of skewness in Japan are all larger than one, indicating the highly positively skewed distribution of returns, whereas in most of the other markets, skewness is usually less than one, suggesting more moderate skewness in returns.

3.3 Common risk factors

Following the previous studies analyzing returns and uncertainty (e.g. Chen et al., 2021, 2024), we use common asset pricing models, including the capital asset pricing model (CAPM) (Sharpe, 1964; Lintner, 1965) and the Fama and French (1993, 2015, 2018) three-, five-, and six-factor models, to further test whether the predictive role of uncertainty is instead captured by common risk factors, including market, size, value, profitability, investment, and momentum.

The market return for Japan is the return on the Tokyo Stock Exchange Stock Price Index (TOPIX). The market returns for the Hong Kong and Taiwan stock markets are the returns of the Hang Seng Index (HSI) and Taiwan Capitalization Weighted Stock Index (TAIEX), respectively. Market returns for India and South Korea are the returns from the corresponding Morgan Stanley Capital International (MSCI) market indexes. The other common risk factors, being size, value, profitability, investment, and momentum for Japan are downloaded from the Kenneth French data library (French, 2021). For other included markets, we collect data from DataStream to construct corresponding factors ourselves.

Size is measured by market capitalization and value is measured by the book-to-market ratio (Fama & French, 1993). Profitability is measured by return on equity (ROE), calculated as the net profit before tax divided by the total shareholder’s equity investment (Chiah, Chai, Zhong, & Li, 2016). Investment is measured by the annual growth rate of total assets (AG) (Fama & French, 2015). Momentum is constructed by using past monthly returns (Fama & French, 2015).

Following previous studies focusing on the Hong Kong (Brown, Du, Rhee, & Zhang, 2008; Lam, Li, & So, 2010), Indian (Mehta & Chander, 2010), South Korean (Kang & Jang, 2016), and Taiwan (Foye, 2018) stock markets, we construct factor mimicking portfolios at the end of June in each year t–1. We then calculate portfolio returns from July in year t–1 to the next June in year t. At the end of June in year t–1, stocks are ranked based on market capitalization to form size groups. The size breakpoint is set following previous studies. Like Lam et al. (2010) and Nartea, Gan, and Wu (2008), we divide stocks in the Hong Kong market into large (L) and small (S) with an equal number of stocks. For India and South Korea, the breakpoint is the median value of size (Kim, Kim, & Shin, 2012; Sehrawat, Kumar, Nigam, Singh, & Goyal, 2020). For the Taiwan market, large (small) stocks account for 90% (10%) of market capitalization (Foye, 2018), so we use this unequal division.

At the same time, we sort stocks into three groups based on their book-to-market ratios, ROE, AG ratios, and past returns to construct value, profitability, investment, and momentum portfolios, respectively. Using value portfolios as an example, stocks are ranked based on their book-to-market ratios: the top 30% are classified as value stocks (H), the bottom 30% as growth stocks (L), and the remaining as medium stocks (M). We then construct six interacted portfolios based on size and values (S/H, S/M, S/L, B/H, B/M, and B/L), and calculate their value-weighted monthly returns from July in year t–1 to June in year t. The return for the value-mimicking portfolio (HML) is calculated by subtracting the average return of the low-value portfolios (S/L and B/L) from the average return of the high-value portfolios (S/H and B/H).

Applying the same method, the return of the profitability mimicking portfolio (RMW) is the return spread between the robust and weak profitability portfolios constructed with size portfolios. The returns of the investment mimicking portfolio (CMA) are the return differences between the conservative (high AG) and aggressive (low AG) portfolios constructed with the size portfolios. Following Fama and French (2015, 2018), we calculate the monthly returns of size mimicking portfolio (SMB) using the average return differences between nine small- and large-sized portfolios constructed with the value, profitability, and investment groups. Following Fama and French (2018), we rank stocks based on their past performance, using the returns from the past 11 months with a 1-month waiting period, to construct momentum portfolios. The monthly returns of the momentum-mimicking portfolio (WML) are calculated by subtracting the average returns of the loser portfolios from the average returns of the winner portfolios, constructed with the size portfolios.

4. Model specification and estimation strategy

Following previous studies (e.g. Bali, Brown, & Tang, 2017; Chen et al., 2021, 2024), we conducted an analysis of the predictive power of China’s uncertainty exposure in four steps. First, we apply rolling regressions with a three-year time window to estimate the monthly uncertainty exposure (beta) for each stock throughout the sample period. Second, we employ the portfolio-level sorting analysis to investigate the relationship between the China’s uncertainty exposure and the future returns of major Asian markets over multiple trading horizons. The raw returns of the high-minus-low portfolios are then adjusted using conventional asset pricing models to investigate whether the relationship is explained by common risk factors. Finally, we check the robustness of the portfolio-level results through firm-level Fama and MacBeth (1973) regressions.

4.1 Uncertainty beta estimation

Similar to previous studies (e.g. Bali, Brown, & Caglayan, 2011; Chen et al., 2021, 2024; Bali, Brown, & Tang, 2023), we employ a monthly rolling regression over a three-year time window to estimate the monthly beta of China's economic policy uncertainty for each stock in each tested market. Prior studies suggest that 24 observations can reduce estimation errors to a reasonable level (Ang, Hodrick, Xing, & Zhang, 2006; Bali et al., 2017). We therefore require that stocks in our sample are listed on the market for at least three years and have at least 24 sequential monthly returns in the past three-year time window. The rolling regression used to estimate betas is specified as follows:

(1)Ri,t−Rf,t=αi+βi,tMRT(Rm,t−Rf,t)+βi,tCEPUCEPUt+εit

where Rit is the log return of stock i, and Rft is the risk-free log return in month t. CEPUt is the EPU index of China in logarithmic form at month t [4]. βi,tCEPU is the uncertainty beta of China’s EPU. We draw on previous studies (e.g. Ang et al., 2006; Bali et al., 2011, 2023; Li, 2016; Chen et al., 2021) and specify a single control variable, namely, market excess returns (Rmt−Rft). βi,tMRT is the market beta for each stock.

4.2 Portfolio sorting analysis

Following prior studies (Amaya, Christoffersen, Jacobs, & Vasquez, 2015; Zhong & Gray, 2016; Bali et al., 2017, 2023; Chen & Lu, 2017; Chen et al., 2021, 2024), we apply portfolio-level sorting analysis to test the relationship between exposure to China’s EPU and future stock returns over multiple trading horizons. The uncertainty betas for a formation period longer than one month are calculated as the monthly averages over that period.

For each month t, we rank stocks into quintiles using their monthly uncertainty betas over the past F (=1, 3, 6, 9, 12) months formation period. We then form a high-minus-low portfolio of buying the highest-beta quintile and selling the lowest-beta quintile and hold this zero-cost portfolio over the following H (=1, 3, 6, 9, 12) months holding period. Both equal- and value-weighted holding-period monthly excess returns are calculated for the high-minus-low portfolio. Section 5 reports the main results for the 1/1 (i.e. 1-month formation and 1-month holding), 3/3, 6/6, 9/9, and 12/12 horizons, while Section 6 shows the results for additional trading horizons.

If these raw returns are significantly different from zero, they are then adjusted using common asset pricing models, being the CAPM (Sharpe, 1964; Lintner, 1965) and the Fama and French (1993, 2015, 2018) three-, five-, and six-factor models. Significant abnormal returns in these models confirm the relationship between exposure to Chinese economic policy uncertainty and future stock returns unexplained by common risk factors. To show whether these adjusted returns are significantly different from zero, we report Newey and West (1987) adjusted t-statistics with four lags [5].

4.3 Firm-level Fama and MacBeth (1973) regression analysis

In addition to the portfolio sorting analysis, we follow similar studies (Amaya et al., 2015; Zhong & Gray, 2016; Bali et al., 2017) and apply the firm-level Fama and MacBeth (1973) regressions to check the robustness of portfolio-level results.

There are two steps for this firm-level analysis. First, we estimate the economic policy uncertainty beta for each stock i in month t by applying a time-series regression over a three-year time window, controlling for market (Rm,t−Rf,t), size (SMBt), value (HMLt), profitability (RMWt), investment (CMAt), and momentum (WMLt) factors. The regression is specified as follows:

(2)Ri,t−Rf,t=αi+βi,tMKT(Rm,t−Rf,t)+βi,tCEPUCEPUt+βi,tSMBSMBt+βi,tHMLHMLt+βi,tRMWRMWt+βi,tCMACMAt+βi,tWMLWMLt+εi,t

The next step is to obtain the economic policy uncertainty risk loading (premium) by employing a cross-sectional regression for all stocks in each month. Similar to portfolio-level analysis, we apply multiple time windows up to 24 months for this firm-level analysis. The formation-period betas are calculated as the monthly average values. We regress the holding-period monthly average return on the formation-period betas in the cross-sectional regression, which is specified as follows:

(3)R̅i,t−Rf,t=λo,t+λMKT,tβi,tMKT+λCEPU,t βi,tCEPU+λSMB,tβi,tSMB+λHML,tβi,tHML+λRMW,tβi,tRMW+λCMA,tβi,tCMA+λWML,tβi,tWML+εi,t

where R̅i,t−Rf,t is the average monthly excess returns of stock i over the holding period t. λCEPU,t is the slope coefficient (risk premium) on economic policy uncertainty betas over the formation period t; λMKT,t, λSMB,t, λHML,t, λRMW,t, λCMA,t, and λWML,t are the risk premiums on the market, size, value, profitability, investment, and momentum betas.

5. Empirical results and analysis

5.1 Portfolio sorting results

We obtain uncertainty betas and apply the portfolio-level sorting analysis to reveal the relationship between exposure to the China’s EPU and individual stock returns in the five Asian markets. Table 2 presents monthly excess returns (in percentages) of the quintile with the lowest (Low) and highest (High) uncertainty betas, and those of the high-minus-low (H-L) portfolios for both full-sample and large stocks ranked in the top-100 and top-200 stocks over the 1/1, 3/3, 6/6, 9/9, and 12/12 trading horizons. The insignificant results for small caps out of the top-200 stocks are unreported.

For the full sample, most high-minus-low returns on the EPU betas are insignificant, and only the value-weighted returns over the 9/9 and 2/12 trading horizons in the Hong Kong market, the 12/12 trading horizon in the Indian market, and the 6/6, 9/9, and 2/12 trading horizons in the South Korean market are significant [6]. The high-minus-low returns of large (i.e. top-100 and top-200) stocks, however, are more significant than the full sample results in most markets, except for the Taiwan stock market. These results indicate that economic policy uncertainty in China mostly drives returns for large stocks in major Asian markets.

For the Japanese and Hong Kong markets, the high-minus-low returns of the top-100 stocks are significant over more trading horizons than those of the top-200 stocks. More importantly, the value-weighted returns of the top-100 and top-200 stocks are significant over more trading horizons than the equal-weighted returns. The most negative value-weighted returns of the top-100 stocks, for instance, are generated over the 3/3 trading horizon, suggesting that buying the lowest-beta portfolio and selling the highest-beta portfolio will generate annual excess returns for around 7.608% [=(0.634%)×12] in Japan and 9.516% [=(0.793%)×12] in Hong Kong. These results indicate again that larger stocks in the Japanese and Hong Kong stock markets are more sensitive to economic policy uncertainty from mainland China.

In the Indian stock market, the equal- and value-weighted results display less difference. For instance, both equal- and value-weighted returns for the top-200 stocks are negative and significant over the 9/9 and 12/12 trading horizons in this market. However, the value-weighted returns are positive in South Korea, which is inconsistent with the negative findings in other markets.

By contrast, high-minus-low returns for both large stocks and the full sample are insignificant over the 1/1 trading horizon, showing that there may be a delay in returns reflecting information transpiring in the past few months. These insignificant results suggest a slow information dissemination mechanism and thus a source of market inefficiency in these markets (Chen et al., 2021, 2024). Exposure to uncertainty in mainland China also does not affect returns in the Taiwan stock market [7].

To test whether the spillover effect of Chinese economic policy uncertainty is explained by common risk factors, we further adjusted the raw high-minus-low returns using the CAPM and Fama and French (1993, 2015, 2018) three-, five-, and six-factor models. Table 3 reports the abnormal returns adjusted by these models for the top-100 and top-200 large stocks that drive the uncertainty effect in the Japanese, Hong Kong, Indian, and South Korean stock markets. The raw results for the Taiwan market are all insignificant, therefore they are omitted from this step.

In an unreported analysis (available from the authors upon request), the adjusted results for the full sample are all explained by common risk factors, except the value-weighted return for the 12/12 trading horizon in India. However, Table 3 shows that adjusted returns of large stocks are still negative and highly significant over most trading horizons in the Japanese, Hong Kong, and Indian markets, indicating the unexplained predictive power of exposure to Chinese economic policy uncertainty. In Japan, for instance, all significant raw returns of the top-100 and top-200 large stocks are unexplained by common asset pricing models. Similar results are found in India, where only the value-weighted returns for the 12/12 trading horizon are explained.

In the Hong Kong market, common risk factors can explain more returns of the top-200 stocks than those of the top-100 stocks. For the top-100 stocks, common risk factors can explain only the high-minus-low returns for the 12/12 trading horizon. However, these factors explain many returns of the top-200 stocks, including all significant equal-weighted returns and the value-weighted returns over the 3/3 and 12/12 trading horizons. The adjusted results, consistent with the raw results, again indicate a stronger predictive role of exposure to uncertainty in mainland China for larger stocks in the Hong Kong stock market.

This negative relationship between exposure to Chinese economic policy uncertainty and future returns in the Japanese, Hong Kong and Indian stock markets is consistent with the predictive role of domestic uncertainty in the US (Bali et al., 2017) and Australia (Chen et al., 2021). These unexplained results suggest that Chinese economic uncertainty, however defined, is a potential and novel candidate for predicting returns in Asian markets.

Interestingly, the positive high-minus-low returns of large stocks are all explained by the Fama-French models in South Korea. In these Fama-French models, we find that the estimated value betas are significant over most trading horizons. Further analysis also shows that the high return earned on the spread of China’s EPU betas may result not from the uncertainty risk, but from value risk in Korea. Ancillary findings (unreported but available upon request) are consistent with Kang, Kang, and Kim (2019) that the value effect is pronounced in Korea.

This opposing result for Korea may also be explained by the irrational behavior of individual investors dominating this market. These investors are irrational, show overconfidence, and have inadequate knowledge to process financial information (Bae, Min, & Jung, 2011; Kim & Park, 2015). Indeed, many studies find that individual investors in Korea normally trade in the opposite direction to institutional investors (Bae et al., 2011; Kim & Park, 2015; Eom, Hahn, & Sohn, 2019). According to the ICAPM (Merton, 1973), risk-averse investors prefer to hold positive uncertainty beta stocks to hedge against uncertainty risk. Irrational individual investors, however, may purchase negative-beta stocks to pursue high risks (Bae et al., 2011; Kim & Park, 2015). These negative uncertainty beta stocks will then incur high current prices due to the demand from irrational individual investors and thereby exhibit lower or even negative future returns owing to the price correction mechanism ensuing. By contrast, stocks with positive uncertainty betas will earn higher future returns.

We believe that the strong predictive power for large stocks can be explained by the close relationship between the Chinese economy and large firms in other Asian markets. Compared with small firms, large firms in most Asian stock markets are engaged more in cross-border business and are therefore more sensitive to uncertainty from other trading markets (Nguyen et al., 2018). In Japan and South Korea, for example, large enterprises supported by the government play a leading role in cross-border trading (Baek, 2005). In Japan (Gallagher & Irwin, 2014) and India (Nölke, Tobias, Claar, & May, 2015), policy banks also generously provide loans to these large firms to encourage cross-border investment. Most importantly, large firms in Asia concentrate their operations within the region and focus less on cross-regional sales, including sales to North America and Europe (Rugman & Hoon Oh, 2008). As China exhibits the strongest economic power in the Asian region, large stocks in Asian markets are influenced by the China’s economic policy uncertainty.

The stronger effect for large stocks may also be explained by the measurement employed to proxy for uncertainty in China, which is the EPU index that pays more attention to uncertainty related to policy changes. In general, large firms are more sensitive to policy changes than small firms, such as changes in trade and labor regulations and economic turbulence (Das & Das, 2014). For instance, large firms are exposed more to law enforcement and inspections than small firms and therefore spend more time addressing government regulations or learning about newly issued policies (Aterido, Hallward-Driemeier, & Pages, 2011). Large firms are also likely to have closer relationships with governments or government authorities than small firms, and may even use political influence to shape economic policies in their favor (Aterido et al., 2011). As large firms in other Asian markets are closely related to the Chinese economy, they are then exposed to any uncertainty related to economic policy changes in China.

5.2 Firm-level Fama and MacBeth (1973) regression results

We now apply the Fama and MacBeth (1973) regressions to check whether the significant adjusted results found in the portfolio-level sorting strategies are robust at the firm-level when controlling for multiple risk factors. Table 4 provides the time-series monthly average slope coefficients (risk premiums) earned on China’s EPU betas and the control variables, including market (Mar), size, value, profitability (Pro), investment (Inv), and momentum (Mom) betas over the 3/3, 6/6, 9/9, and 12/12 trading horizons for large stocks in these markets. The full sample estimations are all insignificant over multiple trading horizons for these tested markets, which are consistent with the portfolio-level results and indicate stronger predictive power of the exposure to China’s uncertainty for large stocks. Insignificant risk premiums of uncertainty betas, such as those for the 1/1 trading horizon and for the Taiwan and South Korean markets are not reported for brevity but are available upon request. Consistent with the portfolio-level results, the insignificant risk premiums of uncertainty betas indicate that the relationship between uncertainty beta exposures and future returns is explained by common risk factors over the 1/1 trading horizon.

Overall, Table 4 shows negative slope coefficients, particularly for the top-100 stocks, over most trading horizons in the tested markets. For the top-100 stocks in the Japanese and Hong Kong stock markets, for example, the exposure to the EPU in China results in negative and significant slope coefficients over most trading horizons. Although other common risk factors, such as the size, value, profitability, investment, and momentum factors, also earn significant premiums in Japan, they diminish in no way the predictive power of the uncertainty betas. Similar results are obtained in India where China’s EPU betas invoke negative and significant slope coefficients over the 9/9 trading horizon.

If stocks earn positive (negative) uncertainty betas, the negative average slope coefficient on uncertainty betas will lead to lower (higher) future returns. Moving from the lowest-to the highest-beta quintile, the average future returns will therefore exhibit a significant decrease, indicating a negative relationship between uncertainty betas and subsequent returns, after controlling for common risk factors. Slope coefficients earned on the top-100 stocks are also significant over more trading horizons than those of the top-200 stocks. Those results, together with the portfolio-level analysis, confirm a negative relationship between the exposure to China’s uncertainty and future returns for large stocks over multiple trading horizons in major Asian markets.

5.3 Uncertainty exposure and characteristics

Following Bali et al. (2017), we further analyze the characteristics of stocks with different exposures to China’s economic policy uncertainty, investigating how uncertainty risk is related to common risk factors. The monthly Fama and MacBeth (1973) cross-sectional regression is applied over the sample period. The regression is as follows:

(4)βi,tCEPU=λo,t+λMKT,tβi,tMKT+λSMB,tβi,tSMB+λHML,tβi,tHML+λRMW,tβi,tRMW+λCMA,tβi,tCMA+λWML,tβi,tWML+εi,t

where βi,tCEPU is the China’s economic policy uncertainty beta of stock i at month t. Risk factors measured are market (βi,tMKT), size (βi,tSMB), value (βi,tHML), profitability (βi,tRMW), investment (βi,tWML), and momentum (βi,tWML) for stock i at month t. λMKT,t, λSMB,t, λHML,t, λRMW,t, λCMA,t, and λWML,t are the corresponding risk premiums on those common risk factors.

Table 5 shows the results for the top-100 and top-200 large stocks in the Japanese, Hong Kong, and Indian stock markets that have been demonstrated to have close relationships with China’s economic policy uncertainty risk. Overall, stocks with higher exposures to China’s uncertainty have higher market betas in these tested markets. Our analysis shows that those stocks will earn lower future returns, which is consistent with the negative relationship between market beta and future returns found in Bali et al. (2017) and Frazzini and Pedersen (2014). Stocks with higher uncertainty betas also show higher exposures to the size risk but lower exposures to probability and investment risks in the Hong Kong market. Different results are found in India, where high uncertainty beta stocks have higher size and profitability betas but lower value and investment betas. Those results suggest that the impacts of different characteristics on China’s economic policy uncertainty beta are subject to market-specific conditions.

5.4 Source of the uncertainty effect: the preference-based explanation

Both our portfolio- and firm-level analyses find a negative relationship between the exposure to China’s economic policy uncertainty and future returns in major Asian markets. Such a negative relationship can be explained by the ICAPM theory (Merton, 1973), which postulates that rational investors are risk-averse, and that they intend to hedge against the unfavorable economic policy uncertainty risk. They are therefore willing to pay high prices to hold stocks with high uncertainty betas, which earn high returns with an increase in economic uncertainty. The popularity of these high-beta stocks pushes their current prices even higher, and their future returns, in turn, will decrease or even become negative due to the price correction mechanism.

When uncertainty is high, the hedging demand for high-beta stocks would be even higher, resulting in a more significant decrease in their future returns compared to normal times. To investigate this, Figure 3 plots the monthly returns of the highest-beta quintile constructed in our portfolio-level analysis, with the turbulence in the China’s economic policy uncertainty. We particularly focus on the top-100 stocks that exhibit the strongest relationship between uncertainty exposure and future returns in the Japanese, Hong Kong, and Indian stock markets.

Consistent with the ICAPM theory (Merton, 1973), Figure 3 plots that future returns of the quintile with the highest uncertainty betas decrease significantly with the increase in the EPU index. This trend is observed in high uncertainty periods, such as in September 2008 during the GFC, in August 2011 following the rating downgrade of US sovereign credit, around the middle of 2015 coinciding with the change in the RMB fixing mechanism, at the end of 2018 amidst heightened US-China trade policy tensions, and in early 2020 during the onset of the COVID-19 pandemic.

These results indicate stronger demand for those high-beta stocks during uncertain periods. The more negative future returns generated on high-beta stocks during uncertainty periods are consistent with the ICAPM theory. When uncertainty is high, investors are more concerned about the future economy (Loh & Stulz, 2018) and are therefore willing to pay higher current prices to hold positive-beta stocks that earn higher returns with the increase in EPU. The extra hedging demand for positive-beta stocks during uncertainty periods gradually pushes their current prices higher. Their future prices and returns, however, will fall even more because of the price correction mechanism.

The results are then consistent with the preference-based explanations in the respective analyses of Bali et al. (2017, 2023) and Chen et al. (2021) for domestic economic uncertainty, as well as Chen et al. (2024) for global uncertainty. Other studies also found a stronger return predictability over recessionary or uncertain periods (Cujean & Hasler, 2017; Loh & Stulz, 2018; Wen & Li, 2020).

6. Robustness check

6.1 Portfolio-level sorting analysis over additional trading horizons

In this subsection, we extend the analysis and test whether the predictive role of China’s economic policy uncertainty betas is robust over additional trading horizons by applying portfolio-level sorting analysis. Table 6 reports raw and adjusted results obtained on the spread of uncertainty betas for the Japanese, Hong Kong, and Indian stock markets. We particularly report the results for the top-100 stocks that reflect stronger EPU spillover effects than smaller stocks. We do not show the insignificant results that can be explained by common risk factors for the South Korean and Taiwan stock markets.

Overall, the negative relationship between exposure to China’s EPU and future individual returns is robust over additional trading horizons in these tested markets. Many abnormal returns also show large t-statistics (>3), satisfying the high threshold suggested by Harvey, Liu, and Zhu (2016) to minimize data mining concerns. For Japan, most equal- and value-weighted raw results are negative and significant over multiple trading horizons, and all adjusted returns remain highly significant. Similar results are obtained in the Hong Kong market, although some value-weighted adjusted returns are explained by common risk factors for the trading strategies over a short formation period or a long holding period. Consistent with the results in Section 5, the China’s EPU exposure predicts future returns over the 9- and 12-month holding periods in India, with some returns explained by common risk factors.

6.2 The China’s EPU vs the US EPU

Although China shows economic power in the Asian region, the US still dominates in terms of affecting economic conditions and financial risk in many markets around the world. Figure 2 compares the China’s EPU index of Huang and Luk (2020) and the US EPU index of Baker et al. (2016) over the sample period. As discussed in Subsection 3.1, political-related turbulences, such as the 9/11 attack in 2001 and the Iraq War in 2003, influence economic-based uncertainty in the US. However, the China’s EPU index is driven primarily by economic shocks. The global stock market crash following the COVID-19 pandemic around early 2020 also has less influence on the uncertainty in China than that in the US, indicating the power of the Chinese government in its aggregate economic stability (Paul & Mas, 2016).

In this subsection, we investigate whether exposure to the US uncertainty has a similar impact to China’s uncertainty to drive individual stock returns in major Asian markets. We use the EPU index of Baker et al. (2016) to measure uncertainty in the US and apply our portfolio-sorting strategies on the estimated US uncertainty betas for large stocks in the Japanese, Indian, and Hong Kong stock markets over the sample period. Table 7 shows the results over multiple trading horizons.

In Japan and India, high-minus-low returns on the spread of the US uncertainty betas are insignificant for both the top-100 and top-200 large stocks over various trading horizons, indicating that there is no relationship between the exposure to the US economic policy uncertainty and future stock returns in these markets. As raw returns in Japan and India are insignificant, there is no need to further adjust these results using common risk factors; therefore, it is unnecessary to provide the adjusted results.

For the Hong Kong market, although value-weighted high-minus-low returns are significant over some trading horizons, they are explained by common risk factors. These results indicate that the exposure to the US uncertainty risk has already been captured in domestic conventional risk factors in the Hong Kong market. Overall, the results for US uncertainty display less significance than those for Chinese uncertainty, highlighting China’s strong economic influence in the Asian region.

7. Conclusion

Motivated by the significant role of uncertainty in affecting investment decisions, stock market returns, and macroeconomic conditions, and the leading role of China in Asia, this study investigates the predictive power of exposure to Chinese economic policy uncertainty for individual stock returns in other major Asian stock markets. Following prior studies on cross-sectional return predictability, we apply both portfolio-level sorting analysis and firm-level Fama and MacBeth (1973) regressions to test whether exposure to Chinese EPU predicts Asian stock returns over multiple trading horizons.

We find that the risk exposure to economic policy uncertainty in China is significantly and negatively related to future individual returns in the Japanese, Hong Kong, and Indian stock markets over multiple trading horizons. More importantly, this predictive role is much stronger for large than for small stocks. The relationship between uncertainty exposures and future returns of the large stocks is unexplained by common risk factors, indicating that China’s uncertainty spillover effect is a market anomaly, and that this can be used to predict and to earn abnormal returns in these markets. The negative relationship is robust when using firm-level regression analysis and when extending the analysis to additional trading horizons.

These negative results are also consistent with the ICAPM, which postulates that risk-averse investors intend to hedge unfavorable uncertainty risk and are therefore willing to accept lower future returns to hold positive uncertainty-beta stocks. However, uncertainty risk from mainland China does not drive returns in the Taiwan stock market. While the relationship between the exposure to China’s economic policy uncertainty and Korean stock returns is positive, the relationship is captured by the conventional value factor.

Our results have three key implications. First, they provide a new return predictor, namely, exposure to Chinese economic policy uncertainty, for investors to better understand and predict stock returns in leading Asian markets. The negative relationship found suggests that stock returns in other Asian markets contain uncertainty risk from China, so this predictor should be considered along with other risk factors when predicting stock returns. Second, our results, which differ from those of previous studies primarily focusing on the spillover effects of the US uncertainty, challenge the leading role of the US in Asia by providing a new intra-regional return predictor. Our results also evidence the ongoing process integration in the Asian economy and its financial markets. Finally, we provide different results in tested markets that reflect market-specific features and thereby reinforce the need to provide additional robustness checks when investigating stock return predictors, and to consider market characteristics when making investment decisions in the real world.

Figures

Figure 1

Regional GDP at purchaser prices in trillions of US dollars

Figure 2

Economic policy uncertainty in China

Figure 3

High-beta returns during uncertain periods

Table 1

Summary statistics for monthly excess returns

Year	Mean					Volatility					Skewness
	Japan	Hong Kong	India	Korea	Taiwan	Japan	Hong Kong	India	Korea	Taiwan	Japan	Hong Kong	India	Korea	Taiwan
2003	2.769	2.462	5.876	0.572	2.136	0.104	0.160	0.184	0.137	0.116	0.886	−0.162	0.539	0.468	0.484
2004	1.783	0.430	2.922	−0.186	−0.009	0.090	0.148	0.169	0.148	0.104	1.079	−0.307	0.624	0.495	−0.352
2005	3.502	−0.803	2.970	5.561	−0.443	0.085	0.141	0.170	0.175	0.116	1.250	1.167	0.507	1.090	0.239
2006	−1.447	1.744	0.157	−0.781	2.555	0.080	0.153	0.158	0.131	0.109	−0.003	1.109	0.720	1.028	0.771
2007	−1.854	3.239	4.395	1.303	−0.063	0.090	0.192	0.166	0.145	0.108	0.240	1.747	0.967	1.017	0.327
2008	−4.248	−8.948	−10.312	−5.821	−5.505	0.125	0.182	0.162	0.149	0.122	−0.182	0.045	0.359	0.268	0.325
2009	0.833	6.362	5.437	3.673	6.385	0.115	0.174	0.164	0.151	0.112	0.779	0.793	0.555	0.600	0.794
2010	0.519	1.231	0.905	0.596	1.941	0.090	0.132	0.139	0.130	0.110	1.196	0.879	0.713	0.812	0.886
2011	−0.756	−3.978	−4.720	−0.884	−2.911	0.091	0.148	0.137	0.136	0.089	1.720	−0.271	−0.129	0.649	0.113
2012	1.413	1.030	0.682	0.321	1.032	0.088	0.124	0.140	0.139	0.085	1.559	−0.132	0.214	0.482	0.573
2013	3.506	1.160	−2.327	−0.026	1.661	0.110	0.128	0.159	0.114	0.084	2.425	1.326	0.305	0.531	1.146
2014	1.103	−0.222	2.888	0.502	0.250	0.091	0.148	0.167	0.118	0.083	1.630	1.217	0.244	0.835	0.976
2015	0.727	−0.507	0.495	1.976	−1.160	0.086	0.152	0.157	0.143	0.090	1.565	0.366	0.229	1.459	0.709
2016	0.412	−0.906	−0.693	−0.036	0.642	0.089	0.137	0.138	0.126	0.078	1.078	−0.707	0.150	0.524	0.885
2017	2.380	0.319	2.325	−0.320	1.477	0.086	0.129	0.136	0.116	0.084	1.546	−0.745	0.368	0.363	1.288
2018	−2.160	−2.770	−4.206	−1.449	−1.121	0.092	0.135	0.135	0.136	0.095	0.440	−0.146	−0.194	0.766	0.864
2019	1.425	−1.583	−3.701	−0.172	1.357	0.086	0.170	0.143	0.125	0.074	0.957	−1.791	−0.281	0.590	0.797
2020	−0.411	−1.447	1.405	1.584	1.443	0.105	0.177	0.167	0.153	0.103	1.084	−0.025	0.543	1.116	1.682

Note(s): This table provides the means, volatility, and Skewness of monthly excess returns of the valid stocks in the sample in the five markets over the period from January 2003 to December 2020

Source(s): Table created by the authors

Table 2

China’s EPU beta rankings and portfolio returns

	F/H		Full sample					Top-100					Top-200
			Japan	HK	India	Korea	Taiwan	Japan	HK	India	Korea	Taiwan	Japan	HK	India	Korea	Taiwan
Equal-weighted returns	1/1	Low	0.572	−0.941	0.065	0.082	0.461	1.295	1.869	1.399	1.231	0.868	1.117	1.715	1.608	1.446	1.029
		High	0.357	−0.288	0.015	0.086	0.099	0.590	1.289	1.334	1.132	1.066	0.769	1.171	1.223	1.264	0.810
		H-L	−0.215	0.652	−0.049	0.004	−0.362	−0.705	−0.580	−0.065	−0.099	0.198	−0.348	−0.544	−0.386	−0.182	−0.219
		t-stat	(−0.374)	(0.765)	(−0.050)	(0.005)	(−0.518)	(−1.278)	(−0.827)	(−0.081)	(−0.142)	(0.303)	(−0.623	(−0.699)	(−0.457)	(−0.251)	(−0.326)
	3/3	Low	0.531	−0.789	0.088	0.058	0.451	0.841	1.286	0.909	0.530	0.542	0.683	0.894	0.974	0.460	0.650
		High	0.358	−0.264	0.004	0.083	0.191	0.285	0.681	0.519	0.669	0.449	0.314	0.471	0.437	0.368	0.364
		H-L	−0.172	0.524	−0.084	0.025	−0.260	−0.556	−0.605	−0.389	0.139	−0.094	−0.369	−0.423	−0.536	−0.092	−0.286
		t-stat	(−0.453)	(0.904)	(−0.138)	(0.058)	(−0.600)	(−1.489)	(−1.365)	(−0.770)	(0.332	(−0.226)	(−1.011)	(−0.831)	(−0.989)	(−0.221)	(−0.667)
	6/6	Low	0.369	−0.738	−0.103	0.100	0.334	0.584	1.010	0.659	0.270	0.519	0.472	0.716	0.664	0.194	0.374
		High	0.275	−0.352	−0.206	0.084	0.166	0.012	0.333	0.051	0.425	0.263	0.067	0.139	0.128	0.055	0.181
		H-L	−0.094	0.385	−0.103	−0.016	−0.168	−0.572*	−0.677**	−0.608*	0.155	−0.256	−0.405	−0.577	−0.536	−0.139	−0.194
		t-stat	(−0.339)	(0.893)	(−0.227)	(−0.052)	(−0.529)	(−1.956)	(−2.065)	(−1.652)	(0.496)	(−0.843)	(−1.460)	(−1.565)	(−1.360)	(−0.456)	(−0.631)
	9/9	Low	0.332	−0.666	0.150	0.095	0.288	0.497	0.809	0.467	0.211	0.509	0.373	0.605	0.514	0.078	0.333
		High	0.230	−0.352	−0.131	0.153	0.132	−0.053	0.108	−0.123	0.397	0.274	−0.012	−0.033	−0.072	−0.011	0.195
		H-L	−0.102	0.314	−0.281	0.058	−0.156	−0.550**	−0.702***	−0.590**	0.186	−0.236	−0.385*	−0.638**	−0.586*	−0.089	−0.138
		t-stat	(−0.446)	(0.852)	(−0.670)	(0.238)	(−0.605)	(−2.304)	(−2.682)	(−2.019)	(0.746)	(−0.977)	(−1.701)	(−2.107)	(−1.838)	(−0.366)	(−0.569)
	12/12	Low	0.317	−0.595	0.182	0.116	0.323	0.400	0.601	0.335	0.082	0.471	0.291	0.486	0.427	0.021	0.312
		High	0.223	−0.254	−0.244	0.247	0.172	−0.044	0.093	−0.091	0.521	0.306	−0.006	−0.066	−0.078	0.064	0.309
		H-L	−0.094	0.341	−0.426	0.131	−0.150	−0.444**	−0.509**	−0.426	0.439**	−0.165	−0.297	−0.552**	−0.505*	0.043	−0.004
		t-stat	(−0.455)	(1.048)	(−1.110)	(0.627)	(−0.661)	(−2.085)	(−2.184)	(−1.636)	(2.009)	(−0.814)	(−1.459)	(−2.015)	(−1.779)	(0.205)	(−0.018)
Value-weighted returns	1/1	Low	0.583	0.187	0.257	0.035	0.264	0.917	0.972	0.728	0.348	0.426	0.746	1.116	0.744	0.257	0.447
		High	0.069	−0.025	0.100	0.377	0.233	0.170	0.209	0.759	1.040	0.715	0.209	0.158	0.523	0.933	0.451
		H-L	−0.514	−0.212	−0.156	0.342	−0.031	−0.747	−0.764	0.031	0.692	0.289	−0.537	−0.958	−0.221	0.676	0.004
		t-stat	(−0.910)	(−0.270)	(−0.163)	(0.487)	(−0.049)	(−1.319)	(−1.144)	(0.038)	(0.968)	(0.473)	(−0.958)	(−1.375)	(−0.272)	(0.938)	(0.007)
	3/3	Low	0.489	0.277	0.398	0.037	0.301	0.676	0.908	0.662	0.072	0.289	0.629	0.733	0.655	0.159	0.350
		High	0.033	−0.141	0.004	0.648	0.305	0.042	0.115	0.256	1.002	0.377	0.014	0.006	0.162	0.992	0.283
		H-L	−0.456	−0.418	−0.395	0.611	0.004	−0.634*	−0.793**	−0.406	0.930**	0.088	−0.615	−0.728*	−0.493	0.833**	−0.067
		t-stat	(−1.228)	(−0.839)	(−0.699)	(1.513)	(0.011)	(−1.665)	(−1.961)	(−0.840)	(2.291)	(0.214)	(−1.641)	(−1.685)	(−0.992)	(2.035)	(−0.168)
	6/6	Low	0.334	0.120	0.324	0.009	0.362	0.487	0.783	0.433	−0.020	0.336	0.454	0.715	0.452	0.047	0.337
		High	−0.066	−0.155	−0.138	0.592	0.072	−0.118	0.106	−0.103	0.900	0.373	−0.091	−0.146	−0.092	0.735	0.115
		H-L	−0.399	−0.275	−0.462	0.583*	−0.289	−0.604**	−0.677**	−0.536	0.920***	0.036	−0.546*	−0.862***	−0.544	0.688**	−0.222
		t-stat	(−1.425)	(−0.792)	(−1.144)	(1.979)	(−0.980)	(−2.070)	(−2.432)	(−1.528)	(3.134)	(0.126)	(−1.910)	(−2.741)	(−1.541)	(2.329)	(−0.744)
	9/9	Low	0.235	0.155	0.467	−0.134	0.333	0.375	0.615	0.271	−0.052	0.339	0.341	0.648	0.333	−0.016	0.307
		High	−0.024	−0.336	−0.121	0.483	0.043	−0.103	0.005	−0.283	0.824	0.341	−0.084	−0.200	−0.254	0.687	0.151
		H-L	−0.259	−0.492*	−0.587	0.618**	−0.290	−0.478**	−0.610***	−0.555**	0.876***	0.001	−0.425*	−0.848***	−0.587**	0.703***	−0.156
		t-stat	(−1.157)	(−1.772)	(−1.569)	(2.535)	(−1.257)	(−2.035)	(−2.765)	(−2.009)	(3.520)	(0.006)	(−1.860)	(−3.328)	(−2.104)	(2.876)	(−0.684)
	12/12	Low	0.219	0.209	0.359	−0.092	0.387	0.289	0.466	0.167	−0.158	0.317	0.274	0.522	0.277	−0.088	0.283
		High	−0.028	−0.383	−0.321	0.503	0.128	−0.071	−0.043	−0.289	0.898	0.272	−0.047	−0.116	−0.253	0.700	0.224
		H-L	−0.247	−0.592**	−0.680*	0.595***	−0.258	−0.360*	−0.509***	−0.456*	1.056***	−0.045	−0.322	−0.638***	−0.530**	0.788***	−0.058
		t-stat	(−1.214)	(−2.199)	(−1.924)	(2.729)	(−1.307)	(−1.718)	(−2.620)	(−1.833)	(4.889)	(−0.240)	(−1.572)	(−2.839)	(−2.111)	(3.659)	(−0.309)

Note(s): This table provides monthly excess returns (in percentage, %) of portfolios with the lowest (Low), highest (High), and high-minus-low (H-L) quintiles of betas on the EPU in China for the full-sample, top-100, and top-200 stocks in five Asian markets from January 2000 to December 2020. F/H are trading horizons for 1- (1/1) to 12-month (12/12) estimation and holding periods. *, **, *** denote that returns are significant at the 10, 5, and 1% levels, respectively. Significant estimates in italic

Source(s): Table created by the authors

Table 3

Adjusted returns on the spread of the China’s EPU betas for large stocks

F/H		Japan		Hong Kong		India		South Korea
		Equal	Value	Equal	Value	Equal	Value	Equal	Value
Panel A: Top-100
3/3	αC	−0.0053***	−0.0060***	−0.0061**	−0.0062*	−0.0043	−0.0054*	0.0017	0.0093***
		(−2.924)	(−2.794)	(−2.207)	(−1.785)	(−1.637)	(−1.936)	(0.633)	(3.157)
	α3	−0.0056***	−0.0063***	−0.0068**	−0.0065**	−0.0043*	−0.0057**	−0.0033	0.0045
		(−3.079)	(−2.935)	(−2.457)	(−1.986)	(−1.658)	(−1.960)	(−1.008)	(1.217)
	α5	−0.0056***	−0.0061***	−0.0078**	−0.0086**	−0.0034	−0.0056	0.0000	0.0069
		(−3.065)	(−2.922)	(−2.367)	(−2.047)	(−1.231)	(−1.547)	(0.004)	(1.528)
	α6	−0.0058***	−0.0063***	−0.0083**	−0.0099**	−0.0044	−0.0063*	0.0002	0.0071
		(−3.294)	(−3.103)	(−2.201)	(−1.998)	(−1.556)	(−1.741)	(0.036)	(1.589)
6/6	αC	−0.0052***	−0.0053***	−0.0073***	−0.0056*	−0.0065***	−0.0064***	0.0014	0.0089***
		(−3.268)	(−3.039)	(−2.842)	(−1.760)	(−2.712)	(−2.762)	(0.696)	(4.009)
	α3	−0.0053***	−0.0056***	−0.0079***	−0.0056*	−0.0063***	−0.0059**	−0.0031	0.0039
		(−3.370)	(−3.271)	(−2.733)	(−1.801)	(−2.612)	(−2.529)	(−1.243)	(1.459)
	α5	−0.0051***	−0.0051***	−0.0152***	−0.0103**	−0.0064*	−0.0074**	0.0007	0.0073*
		(−3.300)	(−3.279)	(−3.377)	(−1.988)	(−1.877)	(−2.449)	(0.173)	(1.873)
	α6	−0.0052***	−0.0052***	−0.0177***	−0.0130**	−0.0072***	−0.0071**	0.0006	0.0073*
		(−3.285)	(−3.277)	(−3.622)	(−1.972)	(−2.180)	(−2.293)	(0.167)	(1.843)
9/9	αC	−0.0049***	−0.0041***	−0.0076***	−0.0049*	−0.0064***	−0.0066***	0.0017	0.0082***
		(−3.387)	(−2.601)	(−2.901)	(−1.842)	(−3.045)	(−3.231)	(0.985)	(3.949)
	α3	−0.0055***	−0.0046***	−0.0081***	−0.0055*	−0.0056***	−0.0057***	−0.0012	0.0045*
		(−4.157)	(−3.259)	(−2.580)	(−1.872)	(−2.646)	(−2.773)	(−0.492)	(1.696)
	α5	−0.0053***	−0.0043***	−0.0161***	−0.0097**	−0.0057*	−0.0063*	0.0018	0.0044
		(−4.097)	(−3.333)	(−3.409)	(−2.204)	(−1.658)	(−1.667)	(0.401)	(0.903)
	α6	−0.0055***	−0.0044***	−0.0210***	−0.0135**	−0.0057*	−0.0062*	0.0020	0.0046
		(−4.203)	(−3.360)	(−3.781)	(−2.121)	(−1.673)	(−1.686)	(0.457)	(0.916)
12/12	αC	−0.0042***	−0.0032**	−0.0060***	−0.0043*	−0.0049**	−0.0060**	0.0036**	0.0097***
		(−3.018)	(−2.261)	(−2.675)	(−1.697)	(−2.434)	(−2.708)	(2.212)	(4.579)
	α3	−0.0047***	−0.0036***	−0.0049	−0.0024	−0.0037*	−0.0044**	0.0032	0.0088***
		(−3.941)	(−3.008)	(−1.471)	(−0.699)	(−1.916)	(−2.138)	(1.455)	(3.002)
	α5	−0.0045***	−0.0035***	−0.0107*	−0.0062	−0.0054**	−0.0042	0.0015	0.0051
		(−3.845)	(−2.929)	(−1.915)	(−1.254)	(−2.017)	(−1.043)	(0.415)	(1.066)
	α6	−0.0046***	−0.0031***	−0.0190***	−0.0118*	−0.0050*	−0.0042	0.0015	0.0050
		(−3.739)	(−2.662)	(−3.227)	(-1.912)	(−1.823)	(−1.056)	(0.419)	(1.057)
Panel B: Top-200
3/3	αC	−0.0031*	−0.0057***	−0.0034	−0.0053	−0.0057**	−0.0064**	−0.0012	0.0079**
		(−1.490)	(−2.744)	(−1.175)	(−1.362)	(−2.430)	(−2.252)	(−0.508)	(2.571)
	α3	−0.0034*	−0.0060***	−0.0048	−0.0055	−0.0055**	−0.0066**	−0.0062**	0.0016
		(−1.648)	(−2.925)	(−1.636)	(−1.490)	(−2.341)	(−2.250)	(−1.992)	(0.421)
	α5	−0.0034	−0.0058***	−0.0077*	−0.0098*	−0.0054**	−0.0072**	−0.0035	0.0032
		(−1.602)	(−2.865)	(−1.834)	(−1.858)	(−2.235)	(−1.995)	(−0.855)	(0.661)
	α6	−0.0034	−0.0059***	−0.0094*	−0.0130**	−0.0066**	−0.0082**	−0.0035	0.0033
		(−1.640)	(−3.006)	(−1.879)	(−2.180)	(−2.571)	(−2.312)	(−0.873)	(0.686)
6/6	αC	−0.0032*	−0.0048***	−0.0056**	−0.0072*	−0.0057***	−0.0063***	−0.0018	0.0063***
		(−1.817)	(−2.828)	(−2.039)	(−1.821)	(−2.709)	(−2.921)	(−0.931)	(2.704)
	α3	−0.0032*	−0.0049***	−0.0060**	−0.0065*	−0.0052**	−0.0059***	−0.0068**	0.0
		(−1.861)	(−3.017)	(−2.136)	(−1.924)	(−2.385)	(−2.653)	(−2.509)	(0.001)
	α5	−0.0032*	−0.0045***	−0.0114**	−0.0144**	−0.0067**	−0.0088***	−0.0054*	0.0032
		(−1.826)	(−2.917)	(−2.433)	(−2.381)	(−2.419)	(−3.529)	(−1.706)	(0.848)
	α6	−0.0031*	−0.0045***	−0.0139**	−0.0194***	−0.0082***	−0.0091***	−0.0053*	0.0033
		(−1.801)	(−2.901)	(−2.347)	(−2.847)	(−2.915)	(−3.449)	(−1.725)	(0.875)
9/9	αC	−0.0032**	−0.0037***	−0.0072**	−0.0074**	−0.0066***	−0.0070***	−0.0010	0.0065***
		(−2.403)	(−2.763)	(−2.296)	(−2.106)	(−3.388)	(−3.752)	(−0.70)	(3.205)
	α3	−0.0033**	−0.0039***	−0.0039	−0.0067*	−0.0057***	−0.0063***	−0.0062***	0.0014
		(−2.529)	(−3.202)	(−1.187)	(−1.852)	(−2.880)	(−3.308)	(−3.237)	(0.547)
	α5	−0.0033**	−0.0038***	−0.0087*	−0.0126**	−0.0082***	−0.0090***	−0.0051	0.0012
		(−2.486)	(−3.101)	(−1.724)	(−2.355)	(−2.905)	(−3.129)	(−1.536)	(0.251)
	α6	−0.0032**	−0.0038***	−0.0139**	−0.0167**	−0.0089***	−0.0092***	−0.0052*	0.0010
		(−2.448)	(−3.112)	(−2.20)	(−2.381)	(−3.120)	(−3.296)	(−1.648)	(0.213)
12/12	αC	−0.0026**	−0.0028**	−0.0069**	−0.0058*	−0.0058***	−0.0067***	0.0001	0.0068***
		(−2.176)	(−2.285)	(−2.211)	(−1.861)	(−2.979)	(−3.261)	(0.076)	(3.340)
	α3	−0.0027**	−0.0031***	−0.0032	−0.0041	−0.0046**	−0.0053***	−0.0035**	0.0050**
		(−2.336)	(−2.795)	(−0.893)	(−0.988)	(−2.482)	(−2.733)	(−2.114)	(2.076)
	α5	−0.0029**	−0.0030**	−0.0085	−0.0081	−0.0071***	−0.0052	−0.0030	0.0027
		(−2.317)	(−2.571)	(−1.260)	(−1.267)	(−2.812)	(−1.380)	(−1.143)	(0.690)
	α6	−0.0025**	−0.0026**	−0.0182**	−0.0159**	−0.0077***	−0.0055	−0.0031	0.0023
		(−2.034)	(−2.286)	(−2.567)	(−2.215)	(−2.979)	(−1.466)	(−1.299)	(0.630)

Note(s): This table provides the adjusted returns earned on the spread of the China’s EPU betas in major Asian markets from January 2000 to December 2020. Panel A and B report results for the top-100 and top-200 stocks, respectively. αC, α3, α5, and α6 are returns adjusted by the CAPM and Fama and French three-, five-, and six-factor models, respectively. The second row is the trading horizon from 3-month (3/3) to 12-month (12/12) estimation and holding periods; “Equal (Value)” are equal- (value-) weighted returns. *, **, *** denote significance at the 10, 5, and 1% levels, respectively. Significant estimates in italic

Source(s): Table created by the authors

Table 4

China’s EPU and Fama-MacBeth results for large stocks

		Top-100								Top-200
		3/3		6/6		9/9		12/12		3/3		6/6		9/9		12/12
Japan	α	0.0068**	(2.562)	0.0063**	(2.523)	0.0058**	(2.423)	0.0050**	(2.079)	0.0058**	(2.266)	0.0053**	(2.215)	0.0051**	(2.350)	0.0048**	(2.360)
	EPU	−0.0215*	(−1.819)	−0.0190*	(−1.870)	−0.0178**	(−2.025)	−0.0140	(−1.638)	−0.0196*	(−1.776)	−0.0148	(−1.506)	−0.0120	(−1.632)	−0.0074	(−1.209)
	Mar	−0.0015	(−0.545)	−0.0030	(−1.341)	−0.0033*	(−1.806)	−0.0027*	(−1.788)	−0.0009	(−0.381)	−0.0023	(−1.211)	−0.0028*	(−1.801)	−0.0028**	(−2.125)
	Size	−0.0037**	(−1.988)	−0.0047***	(−2.802)	−0.0049***	(−3.399)	−0.0046***	(−3.586)	−0.0031**	(−2.387)	−0.0040***	(−3.265)	−0.0038***	(−3.604)	−0.0036***	(−4.020)
	Value	−0.0048**	(−2.377)	−0.0045***	(−2.619)	−0.0041***	(−2.764)	−0.0033**	(−2.541)	−0.0031*	(−1.693)	−0.0030**	(−1.974)	−0.0028**	(−2.175)	−0.0025**	(−2.228)
	Pro	−0.0038**	(−2.315)	−0.0035***	(−2.668)	−0.0032***	(−2.851)	−0.0029***	(−2.687)	−0.0025*	(−1.848)	−0.0028**	(−2.549)	−0.0028***	(−3.174)	−0.0027***	(−3.781)
	Inv	−0.0029**	(−1.995)	−0.0030***	(−2.723)	−0.0029***	(−2.953)	−0.0024**	(−2.599)	−0.0020*	(−1.710)	−0.0022**	(−2.342)	−0.0022***	(−2.792)	−0.0022***	(−2.744)
	Mom	−0.0051***	(−3.081)	−0.0058***	(−3.811)	−0.0062***	(−4.541)	−0.0053***	(−4.531)	−0.0040***	(−2.999)	−0.0043***	(−3.406)	−0.0045***	(−3.781)	−0.0043***	(−3.968)
	Adj.R2	3.17%		12.75%		28.74%		36.43%		13.69%		16.71%		31.16%		24.54%
Hong Kong	α	0.0041	(1.221)	0.0014	(0.485)	0.0008	(0.293)	−0.0004	(−0.174)	0.0052	(1.395)	0.0055	(1.636)	0.0045	(1.376)	0.0023	(0.838)
	EPU	−0.0354**	(−2.347)	−0.0302**	(−2.296)	−0.0271*	(−1.838)	−0.0274**	(−2.103)	−0.0218	(−1.577)	−0.0272**	(−2.270)	−0.0219*	(−1.933)	−0.0196*	(−1.774)
	Mar	0.0031	(1.014)	0.0042	(1.585)	0.0039*	(1.704)	0.0051***	(2.608)	0.0012	(0.470)	−0.0003	(−0.121)	−0.0005	(−0.212)	0.0010	(0.492)
	Size	0.0039	(1.189)	0.0020	(0.671)	0.0016	(0.624)	0.0020	(0.825)	−0.0021	(−0.750)	−0.0023	(−0.903)	−0.0013	(−0.582)	0.0016	(0.700)
	Value	0.0014	(0.810)	0.0008	(0.505)	−0.0003	(−0.203)	−0.0022	(−1.579)	0.0008	(0.599)	−0.0003	(−0.221)	−0.0012	(−0.883)	−0.0018	(−1.370)
	Pro	−0.0037	(−1.257)	−0.0012	(−0.413)	−0.0008	(−0.313)	−0.0029	(−1.248)	0.0001	(0.028)	0.0005	(0.174)	0.0000	(−0.011)	−0.0025	(−1.044)
	Inv	0.0004	(0.167)	0.0005	(0.220)	−0.0010	(−0.623)	−0.0027**	(−2.315)	−0.0011	(−0.384)	−0.0002	(−0.097)	−0.0011	(−0.806)	−0.0024	(−2.466)
	Mom	−0.0034	(−1.335)	−0.0041**	(−1.978)	−0.0034*	(−1.666)	−0.0034*	(−1.785)	−0.0056	(−1.308)	−0.0020	(−0.691)	0.0000	(−0.017)	−0.0024	(−1.307)
	Adj.R2	12.82%		10.18%		29.34%		20.85%		13.90%		14.54%		20.79%		15.48%
India	α	0.0151***	(4.122)	0.0121***	(3.968)	0.0097***	(3.352)	0.0091***	(3.601)	0.0137***	(3.797)	0.0110***	(3.571)	0.0096***	(3.077)	0.0095***	(3.031)
	EPU	−0.0144	(−0.992)	−0.0166	(−1.329)	−0.0183*	(−1.907)	−0.0137	(−1.557)	−0.0067	(−0.522)	−0.0116	(−1.124)	−0.0119	(−1.497)	−0.0077	(−1.071)
	Mar	−0.0066**	(−1.994)	−0.0069**	(−2.510)	−0.0062***	(−3.455)	−0.0063***	(−4.630)	−0.0050	(−1.477)	−0.0052*	(−1.910)	−0.0053***	(−2.641)	−0.0058***	(−3.191)
	Size	0.0002	(0.093)	0.0021	(0.963)	0.0018	(0.964)	0.0012	(0.686)	−0.0007	(−0.379)	0.0002	(0.123)	−0.0002	(−0.159)	−0.0006	(−0.419)
	Value	0.0002	(0.110)	0.0005	(0.263)	0.0007	(0.479)	0.0008	(0.549)	−0.0027	(−1.475)	−0.0027*	(−1.741)	−0.0022	(−1.542)	−0.0021	(−1.525)
	Pro	0.0007	(0.404)	0.0008	(0.523)	0.0005	(0.363)	0.0010	(0.916)	0.0022*	(1.709)	0.0017	(1.501)	0.0010	(0.904)	0.0014	(1.210)
	Inv	0.0017	(0.817)	0.0014	(0.803)	0.0011	(0.764)	0.0013	(1.035)	0.0030	(1.545)	0.0021	(1.311)	0.0022	(1.525)	0.0029**	(2.138)
	Mom	0.0011	(0.304)	0.0021	(0.632)	0.0019	(0.658)	0.0012	(0.492)	−0.0001	(−0.047)	0.0007	(0.292)	0.0003	(0.128)	0.0004	(0.169)
	Adj.R2	24.87%		16.19%		4.77%		20.99%		24.93%		9.31%		4.16%		10.90%

Note(s): This table shows abnormal returns (α) and risk premiums earned on China’s EPU betas, market (Mar), size, value, profitability (Pro), investment (Inv), and momentum (Mom) factors estimated by the Fama-MacBeth regressions over selected trading horizons (F/H) for Japan, Hong Kong, and India from January 2000 to December 2020. *, **, *** denote significance at the 10, 5, and 1% levels, respectively. Significant EPU premiums in italic

Source(s): Table created by the authors

Table 5

China’s EPU betas and average stock characteristics

	Japan		Hong Kong		India
Panel A: Top-100
α	−0.0309***	(−5.085)	−0.0415***	(−6.214)	−0.0380***	(−5.825)
Mar	0.0242***	(3.783)	0.0259***	(4.993)	0.0421***	(6.566)
Size	−0.0028	(−0.477)	0.0624***	(10.481)	0.0254***	(3.451)
Value	−0.0044	(−0.476)	−0.0032	(−0.761)	−0.0439***	(−5.732)
Pro	0.0093*	(1.750)	−0.0541***	(−5.811)	0.0368***	(5.084)
Inv	−0.0043	(−0.512)	−0.0061	(−0.890)	−0.0423***	(−4.230)
Mom	0.0083	(0.927)	0.0166	(1.549)	0.0119	(1.211)
Panel B: Top-200
α	−0.0176***	(−3.386)	−0.0519***	(−5.910)	−0.0391***	(−5.853)
Mar	0.0139**	(2.555)	0.0322***	(4.976)	0.0432***	(8.147)
Size	−0.0041	(−0.699)	0.0369***	(7.037)	0.0324***	(5.442)
Value	−0.0096	(−1.198)	−0.0062	(−1.236)	−0.0355***	(−4.789)
Pro	0.0054	(1.137)	−0.0255***	(−5.489)	0.0235***	(3.659)
Inv	−0.0143	(−1.558)	−0.0183***	(−4.092)	−0.0479***	(−5.454)
Mom	0.0026	(0.288)	0.0151**	(2.005)	0.0079	(0.839)

Note(s): This table shows the average slope coefficients from the regressions of the China’s economic policy uncertainty betas on common risk factors, including market (Mar), size, value, profitability (Pro), investment (Inv), and momentum (Mom), for the top-100 (Panel A) and top-200 (Panel B) large stocks in Japan, Hong Kong, and India from January 2000 to December 2020. *, **, *** denote significance at the 10, 5, and 1% levels, respectively

Source(s): Table created by the authors

Table 6

China’s EPU effects of the top-100 stocks over additional trading horizons

F/H	1/3	1/6	1/9	1/12	3/6	3/9	3/12	6/3	6/9	6/12	9/3	9/6	9/12	12/3	12/6	12/9
Panel A: Japan
Equal-weighted returns
H-L	−0.693*	−0.647**	−0.608***	−0.604***	−0.565**	−0.571**	−0.575***	−0.520	−0.555**	−0.564***	−0.545	−0.609**	−0.505**	−0.639*	−0.591**	−0.494**
	(−1.926)	(−2.331)	(−2.695)	(−3.040)	(−1.975)	(−2.497)	(−2.834)	(−1.390)	(−2.394)	(−2.753)	(−1.431)	(−2.069)	(−2.420)	(−1.658)	(−1.966)	(−2.046)
αC	−0.0067***	−0.0065***	−0.0060***	−0.0059***	−0.0054***	−0.0054***	−0.0054***	−0.0048***	−0.0050***	−0.0053***	−0.0050***	−0.0055***	−0.0047***	−0.0058***	−0.0053***	−0.0046***
	(−3.621)	(−3.930)	(−4.60)	(−4.781)	(−3.553)	(−4.248)	(−4.376)	(−2.644)	(−3.810)	(−4.003)	(−2.778)	(−3.544)	(−3.441)	(−3.382)	(−3.366)	(−3.080)
α3	−0.0071***	−0.0067***	−0.0063***	−0.0063***	−0.0055***	−0.0057***	−0.0058***	−0.0050***	−0.0055***	−0.0058***	−0.0052***	−0.0057***	−0.0053***	−0.0061***	−0.0056***	−0.0050***
	(−3.876)	(−4.100)	(−5.061)	(−5.698)	(−3.682)	(−4.740)	(−5.361)	(−2.794)	(−4.421)	(−5.071)	(−2.943)	(−3.740)	(−4.574)	(−3.518)	(−3.583)	(−3.694)
α5	−0.0072***	−0.0067***	−0.0063***	−0.0061***	−0.0055***	−0.0056***	−0.0056***	−0.0050***	−0.0053***	−0.0055***	−0.0051***	−0.0055***	−0.0051***	−0.0061***	−0.0055***	−0.0049***
	(−3.805)	(−4.218)	(−5.067)	(−5.640)	(−3.704)	(−4.655)	(−5.217)	(−2.768)	(−4.303)	(−4.847)	(−2.919)	(−3.709)	(−4.426)	(−3.515)	(−3.567)	(−3.682)
α6	−0.0073***	−0.0069***	−0.0065***	−0.0063***	−0.0056***	−0.0057***	−0.0057***	−0.0050***	−0.0054***	−0.0056***	−0.0051***	−0.0055***	−0.0052***	−0.0061***	−0.0056***	−0.0051***
	(−4.073)	(−4.407)	(−5.218)	(−5.807)	(−3.781)	(−4.802)	(−5.322)	(−2.786)	(−4.383)	(−4.845)	(−2.921)	(−3.682)	(−4.454)	(−3.518)	(−3.548)	(−3.764)
Value-weighted returns
H-L	−0.707**	−0.617**	−0.555***	−0.485***	−0.630**	−0.599***	−0.540***	−0.574	−0.538**	−0.502**	−0.496	−0.545*	−0.417**	−0.586	−0.507*	−0.411*
	(−2.001)	(−2.317)	(−2.631)	(−2.745)	(−2.183)	(−2.585)	(−2.667)	(−1.50)	(−2.318)	(−2.482)	(−1.309)	(−1.889)	(−2.042)	(−1.516)	(−1.709)	(−1.727)
αC	−0.0067***	−0.0060***	−0.0054***	−0.0046***	−0.0059***	−0.0054***	−0.0046***	−0.0052**	−0.0046***	−0.0044***	−0.0042**	−0.0046***	−0.0037***	−0.0051***	−0.0044**	−0.0036**
	(−3.131)	(−3.384)	(−3.923)	(−3.779)	(−3.271)	(−3.788)	(−3.577)	(−2.438)	(−3.319)	(−3.271)	(−2.109)	(−2.789)	(−2.611)	(−2.580)	(−2.568)	(−2.281)
α3	−0.0070***	−0.0063***	−0.0057***	−0.0050***	−0.0062***	−0.0057***	−0.0049***	−0.0056***	−0.0050***	−0.0048***	−0.0045**	−0.0049***	−0.0042***	−0.0054***	−0.0047***	−0.0040***
	(−3.284)	(−3.571)	(−4.288)	(−4.468)	(−3.467)	(−4.234)	(−4.338)	(−2.654)	(−3.865)	(−4.236)	(−2.326)	(−3.054)	(−3.615)	(−2.807)	(−2.890)	(−2.845)
α5	−0.0069***	−0.0058***	−0.0053***	−0.0046***	−0.0057***	−0.0052***	−0.0045***	−0.0053***	−0.0047***	−0.0045***	−0.0044**	−0.0046***	−0.0039***	−0.0054***	−0.0046***	−0.0038***
	(−3.229)	(−3.657)	(−4.503)	(−4.679)	(−3.546)	(−4.374)	(−4.388)	(−2.656)	(−4.005)	(−4.208)	(−2.321)	(−3.083)	(−3.477)	(−2.846)	(−2.948)	(−2.911)
α6	−0.0070***	−0.0060***	−0.0054***	−0.0046***	−0.0058***	−0.0054***	−0.0045***	−0.0054***	−0.0048***	−0.0045***	−0.0044**	−0.0046***	−0.0038***	−0.0054***	−0.0046***	−0.0039***
	(−3.365)	(−3.780)	(−4.455)	(−4.567)	(−3.642)	(−4.398)	(−4.340)	(−2.690)	(−3.997)	(−4.071)	(−2.322)	(−3.054)	(−3.350)	(−2.821)	(−2.884)	(−2.885)
Panel B: Hong Kong
Equal-weighted returns
H-L	−0.508	−0.540*	−0.568**	−0.549**	−0.599*	−0.545**	−0.488**	−0.609	−0.621**	−0.515**	−0.862**	−0.844***	−0.571**	−0.783*	−0.767**	−0.649**
	(−1.140)	(−1.654)	(−2.104)	(−2.293)	(−1.840)	(−2.038)	(−2.056)	(−1.399)	(−2.344)	(−2.165)	(−2.007)	(−2.666)	(−2.461)	(−1.802)	(−2.418)	(−2.514)
αC	−0.0049*	−0.0052**	−0.0055**	−0.0062***	−0.0058**	−0.0057***	−0.0059***	−0.0062**	−0.0070***	−0.0065***	−0.0087***	−0.0088***	−0.0069***	−0.0079**	−0.0081***	−0.0069***
	(−1.808)	(−2.061)	(−2.504)	(−3.046)	(−2.287)	(−2.624)	(−2.826)	(−2.220)	(−2.901)	(−2.687)	(−2.694)	(−3.038)	(−2.861)	(−2.540)	(−2.774)	(−2.680)
α3	−0.0057**	−0.0063**	−0.0056**	−0.0053**	−0.0068**	−0.0052**	−0.0048	−0.0068**	−0.0072**	−0.0052	−0.0085***	−0.0093***	−0.0067*	−0.0077**	−0.0086***	−0.0071**
	(−2.042)	(−2.334)	(−2.367)	(−2.059)	(−2.461)	(−2.015)	(−1.641)	(−2.386)	(−2.472)	(−1.465)	(−2.709)	(−2.879)	(−1.917)	(−2.563)	(−2.703)	(−2.263)
α5	−0.0061*	−0.0092**	−0.0124***	−0.0122***	−0.0118***	−0.0138***	−0.0129***	−0.0096***	−0.0165***	−0.0138**	−0.0125***	−0.0156***	−0.0143**	−0.0103***	−0.0126***	−0.0128***
	(−1.891)	(−2.50)	(−3.057)	(−2.724)	(−2.850)	(−3.158)	(−2.698)	(−2.674)	(−3.588)	(−2.508)	(−3.197)	(−3.426)	(−2.523)	(−2.737)	(−3.002)	(−2.752)
α6	−0.0064*	−0.0113***	−0.0178***	−0.0212***	−0.0147***	−0.0189***	−0.0214***	−0.0110***	−0.0215***	−0.0227***	−0.0141***	−0.0178***	−0.0235***	−0.0118***	−0.0137**	−0.0175***
	(−1.744)	(−2.746)	(−4.117)	(−4.978)	(−3.146)	(−3.847)	(−4.726)	(−2.738)	(−4.438)	(−4.355)	(−3.231)	(−3.364)	(−3.943)	(−2.646)	(−2.563)	(−3.219)
Value-weighted returns
H-L	−0.623	−0.607**	−0.504**	−0.405**	−0.726**	−0.698***	−0.654***	−0.645*	−0.672***	−0.586***	−0.628*	−0.669**	−0.554***	−0.527	−0.611**	−0.566**
	(−1.597)	(−2.217)	(−2.347)	(−2.326)	(−2.526)	(−2.977)	(−3.258)	(−1.678)	(−2.986)	(−3.009)	(−1.649)	(−2.440)	(−2.902)	(−1.358)	(−2.208)	(−2.560)
αC	−0.0041	−0.0044	−0.0034	−0.0028	−0.0060*	−0.0057**	−0.0056**	−0.0047	−0.0057**	−0.0048**	−0.0047	−0.0056*	−0.0044*	−0.0040	−0.0053*	−0.0048*
	(−1.276)	(−1.469)	(−1.451)	(−1.573)	(−1.801)	(−2.125)	(−2.382)	(−1.328)	(−2.045)	(−1.982)	(−1.279)	(−1.746)	(−1.897)	(−1.165)	(−1.655)	(−1.665)
α3	−0.0044	−0.0045	−0.0029	−0.0005	−0.0061*	−0.0049*	−0.0052	−0.0047	−0.0059*	−0.0043	−0.0035	−0.0054*	−0.0044	−0.0033	−0.0047	−0.0045
	(−1.418)	(−1.486)	(−1.025)	(−0.219)	(−1.955)	(−1.675)	(−1.579)	(−1.330)	(−1.894)	(−1.196)	(−0.996)	(−1.766)	(−1.301)	(−1.021)	(−1.583)	(−1.585)
α5	−0.0057	−0.0062	−0.0047	−0.0033	−0.0093*	−0.0094*	−0.0093*	−0.0080*	−0.0109**	−0.0097*	−0.0070	−0.0105**	−0.0088*	−0.0067	−0.0072*	−0.0065
	(−1.510)	(−1.381)	(−1.019)	(−0.730)	(−1.846)	(−1.873)	(−1.692)	(−1.766)	(−2.202)	(−1.767)	(−1.608)	(−2.085)	(−1.780)	(−1.565)	(−1.645)	(−1.598)
α6	−0.0062	−0.0098*	−0.0112**	−0.0120**	−0.0134**	−0.0161**	−0.0182***	−0.0094*	−0.0153**	−0.0181***	−0.0078	−0.0125*	−0.0156***	−0.0082	−0.0082	−0.0092
	(−1.382)	(−1.832)	(−2.026)	(−2.358)	(−2.169)	(−2.553)	(−3.075)	(−1.751)	(−2.335)	(−3.041)	(−1.504)	(−1.899)	(−2.654)	(−1.609)	(−1.303)	(−1.462)
Panel C: India
Equal-weighted returns
H-L	−0.370	−0.513	−0.550*	−0.537**	−0.519	−0.557*	−0.533**	−0.498	−0.585*	−0.520**	−0.597	−0.674*	−0.496*	−0.587	−0.586*	−0.492*
	(−0.734)	(−1.336)	(−1.735)	(−2.007)	(−1.367)	(−1.797)	(−2.049)	(−1.016)	(−1.992)	(−2.051)	(−1.232)	(−1.893)	(−1.950)	(−1.224)	(−1.655)	(−1.675)
αC	−0.0038	−0.0052**	−0.0059***	−0.0059***	−0.0056**	−0.0062***	−0.0060***	−0.0056**	−0.0064***	−0.0058***	−0.0064**	−0.0071***	−0.0055***	−0.0066**	−0.0064***	−0.0056***
	(−1.369)	(−2.260)	(−2.746)	(−2.80)	(−2.361)	(−2.823)	(−2.820)	(−2.157)	(−2.927)	(−2.744)	(−2.393)	(−3.006)	(−2.644)	(−2.428)	(−2.916)	(−2.764)
α3	−0.0039	−0.0052**	−0.0057***	−0.0053**	−0.0054**	−0.0058***	−0.0053**	−0.0054**	−0.0059***	−0.0049**	−0.0061**	−0.0067***	−0.0044**	−0.0063**	−0.0059***	−0.0046**
	(−1.467)	(−2.218)	(−2.628)	(−2.517)	(−2.30)	(−2.642)	(−2.472)	(−2.126)	(−2.625)	(−2.281)	(−2.323)	(−2.791)	(−2.136)	(−2.349)	(−2.651)	(−2.316)
α5	−0.0031	−0.0055**	−0.0067**	−0.0075**	−0.0047	−0.0062*	−0.0073**	−0.0052*	−0.0061	−0.0066*	−0.0064**	−0.0071**	−0.0056*	−0.0061**	−0.0056*	−0.0048*
	(−1.155)	(−2.121)	(−2.421)	(−2.282)	(−1.644)	(−1.935)	(−2.082)	(−1.925)	(−1.639)	(−1.697)	(−2.215)	(−1.993)	(−1.658)	(−2.069)	(−1.702)	(−1.672)
α6	−0.0050	−0.0066***	−0.0075***	−0.0086***	−0.0055**	−0.0071**	−0.0078**	−0.0059**	−0.0067*	−0.0067*	−0.0071**	−0.0073**	−0.0054	−0.0065**	−0.0053*	−0.0042
	(−1.733)	(−2.673)	(−2.680)	(−2.619)	(−2.0)	(−2.231)	(−2.267)	(−2.080)	(−1.830)	(−1.750)	(−2.476)	(−2.149)	(−1.566)	(−2.238)	(−1.673)	(−1.487)
Value-weighted returns
H-L	−0.221	−0.403	−0.405	−0.384*	−0.478	−0.458	−0.466*	−0.587	−0.505*	−0.470**	−0.656	−0.614*	−0.488**	−0.580	−0.557	−0.496*
	(−0.462)	(−1.162)	(−1.466)	(−1.696)	(−1.335)	(−1.574)	(−1.925)	(−1.239)	(−1.819)	(−1.996)	(−1.394)	(−1.790)	(−2.026)	(−1.221)	(−1.614)	(−1.761)
αC	−0.0035	−0.0051**	−0.0056***	−0.0054***	−0.0058***	−0.0060***	−0.0060***	−0.0072***	−0.0062***	−0.0058***	−0.0078***	−0.0071***	−0.0061***	−0.0073***	−0.0068***	−0.0063***
	(−1.106)	(−2.359)	(−2.853)	(−2.917)	(−2.711)	(−2.966)	(−3.048)	(−2.704)	(−3.054)	(−2.838)	(−2.864)	(−3.048)	(−2.831)	(−2.630)	(−3.041)	(−2.962)
α3	−0.0042	−0.0051**	−0.0052***	−0.0047***	−0.0057***	−0.0054***	−0.0051***	−0.0071**	−0.0056***	−0.0047**	−0.0076***	−0.0066***	−0.0046**	−0.0073**	−0.0062***	−0.0051**
	(−1.411)	(−2.388)	(−2.766)	(−2.629)	(−2.625)	(−2.755)	(−2.658)	(−2.535)	(−2.660)	(−2.299)	(−2.711)	(−2.743)	(−2.259)	(−2.575)	(−2.739)	(−2.472)
α5	−0.0020	−0.0040	−0.0064**	−0.0057*	−0.0057**	−0.0065**	−0.0070**	−0.0094***	−0.0068*	−0.0064	−0.0108***	−0.0082**	−0.0050	−0.0095***	−0.0057	−0.0043
	(−0.485)	(−1.244)	(−2.418)	(−1.916)	(−2.094)	(−2.240)	(−2.206)	(−3.217)	(−1.957)	(−1.512)	(−3.611)	(−2.503)	(−1.145)	(−2.982)	(−1.619)	(−1.072)
α6	−0.0037	−0.0045	−0.0065**	−0.0063**	−0.0056**	−0.0064**	−0.0070**	−0.0102***	−0.0065*	−0.0060	−0.0117***	−0.0079**	−0.0050	−0.0098***	−0.0052	−0.0040
	(−0.951)	(−1.474)	(−2.521)	(−2.258)	(−2.073)	(−2.318)	(−2.270)	(−3.333)	(−1.946)	(−1.441)	(−3.872)	(−2.398)	(−1.164)	(−3.155)	(−1.441)	(−1.017)

Note(s): This table shows portfolio-sorting results of the China’s EPU betas for the top-100 large stocks over more trading horizons in Japan, Hong Kong, and India over the sample period from January 2000 to December 2020. It reports monthly returns (in percentage, %) of the high-minus-low portfolios (H-L) formed on the uncertainty betas and abnormal returns adjusted by the CAPM (αC) and Fama-French three- (α3), five- (α5), and six-factor (α6) models, respectively. “F/H” row shows trading horizons from 3-month (3/3) to 12-month (12/12) estimation and holding periods; “Equal (Value)” are equal- (value-) weighted returns. Panel A, B, C report both equal- and value-weighted results for Japan, Hong Kong, and India. *, **, *** denote significance at the 10, 5, and 1% levels, respectively

Source(s): Table created by the authors

Table 7

US uncertainty beta ranking and portfolio returns

F/H		3/3		6/6		9/9		12/12
		Equal	Value	Equal	Value	Equal	Value	Equal	Value
Panel A: Top-100
Japan	H-L	−0.158	−0.253	−0.257	−0.334	−0.269	−0.288	−0.293	−0.260
		(−0.413)	(−0.662)	(−0.912)	(−1.195)	(−1.155)	(−1.210)	(−1.380)	(−1.163)
India	H-L	−0.543	−0.034	−0.477	−0.144	−0.379	−0.184	−0.233	−0.151
		(−1.101)	(−0.069)	(−1.291)	(−0.398)	(−1.306)	(−0.648)	(−0.898)	(−0.596)
Hong Kong	H-L	−0.346	−0.582	−0.355	−0.738**	−0.325	−0.711***	−0.275	−0.692***
		(−0.751)	(−1.340)	(−1.004)	(−2.287)	(−1.141)	(−2.773)	(−1.085)	(−3.025)
	αC	−0.0016	−0.0039	−0.0011	−0.0046	−0.0012	−0.0051*	−0.0016	−0.0054*
		(−0.453)	(−1.073)	(−0.336)	(−1.523)	(−0.517)	(−1.788)	(−0.778)	(−1.874)
	α3	−0.0035	−0.0027	−0.0037	−0.0068*	−0.0044**	−0.0092***	−0.0059***	−0.0110***
		(−1.030)	(−0.642)	(−1.211)	(−1.816)	(−1.992)	(−3.229)	(−2.692)	(−3.890)
	α5	−0.0022	−0.0018	−0.0066	−0.0089	−0.0068	−0.0090	−0.0097**	−0.0091
		(−0.477)	(−0.328)	(−1.321)	(−1.555)	(−1.418)	(−1.617)	(−2.001)	(−1.501)
	α6	−0.0029	−0.0025	−0.0098	−0.0098	−0.0105	−0.0120	−0.0156**	−0.0169**
		(−0.530)	(−0.388)	(−1.570)	(−1.305)	(−1.608)	(−1.494)	(−2.426)	(−2.144)
Panel B: Top-200
Japan	H-L	−0.124	−0.301	−0.245	−0.311	−0.310	−0.303	−0.286	−0.291
		(−0.345)	(−0.814)	(−0.929)	(−1.149)	(−1.425)	(−1.322)	(−1.449)	(−1.366)
India	H-L	−0.455	−0.097	−0.456	−0.011	−0.430	−0.167	−0.278	−0.052
		(−0.849)	(−0.191)	(−1.154)	(−0.031)	(−1.327)	(−0.562)	(−0.972)	(−0.198)
Hong Kong	H-L	−0.122	−0.596	−0.240	−0.661**	−0.263	−0.534**	−0.172	−0.484**
		(−0.233)	(−1.347)	(−0.627)	(−2.011)	(−0.842)	(−2.055)	(−0.595)	(−2.078)
	αC	0.0015	−0.0036	0.0003	−0.0038	−0.0005	−0.0034	−0.0002	−0.0031
		(0.431)	(−0.938)	(0.100)	(−1.234)	(−0.195)	(−1.167)	(−0.072)	(−1.073)
	α3	0.0008	−0.0025	−0.0008	−0.0051	−0.0012	−0.0051*	−0.0017	−0.0074
		(0.242)	(−0.628)	(−0.286)	(−1.541)	(−0.527)	(−1.827)	(−0.796)	(−2.863)
	α5	−0.0002	−0.0026	−0.0016	−0.0061	−0.0017	−0.0051	−0.0034	−0.0062
		(−0.042)	(−0.437)	(−0.304)	(−1.051)	(−0.334)	(−0.874)	(−0.598)	(−1.003)
	α6	−0.0008	−0.0040	−0.0035	−0.0088	−0.0041	−0.0092	−0.0100	−0.0144*
		(−0.144)	(−0.532)	(−0.524)	(−1.119)	(−0.540)	(−1.110)	(−1.349)	(−1.857)

Note(s): This table shows monthly excess returns (in percentage, %) of the high-minus-low (H-L) portfolio constructed on the US EPU betas for the top-100 (Panel A) and top-200 (Panel B) large stock in Japan, India, and Hong Kong. It also shows adjusted returns (α) for Hong Kong. “F/H” row shows trading horizons from 3-month (3/3) to 12-month (12/12) estimation and holding periods; “Equal (Value)” are equal- (value-) weighted returns. *, **, *** denote that returns are significant at the 10, 5, and 1% levels, respectively

Source(s): Table created by the authors

Notes

1. Stocks satisfying our estimation requirements in other Asian markets are too few for constructing diversified sorting portfolios. For example, in Singapore, the next largest market after Taiwan, only 14% of stocks by number satisfy our requirement. In some years (e.g. 2004, 2005, 2019, and 2020), the number of valid Singaporean stocks is less than 90 and those even among the top-100 stocks number less than 20. The valid stocks also account for just 31% of total Singaporean market capitalization.

2. Available at https://economicpolicyuncertaintyinchina.weebly.com/

3. The Baker et al. (2016) EPU index is available at www.PolicyUncertainty.com

4. Previous studies (e.g. Bilgin, Gozgor, Lau, & Sheng, 2018) measure macroeconomic variables (e.g. GDP and CPI) in natural logarithmic form to filter the upward time trend. As the Chinese index in Figure 2 exhibits a clear upward trend, we measure this index in logarithmic form following similar studies (Gao & Zhang, 2016; Demir, Gozgor, Lau, & Vigne, 2018; Gozgor & Demir, 2018; Guo, Zhu, & You, 2018; Gozgor, Tiwari, Demir, & Akron, 2019).

5. Following many studies (Ho & Siu, 2007; Ho & Li, 2008; Feldman, 2010; Wang & Huang, 2020), we apply the plug-in rule of Newey and West (1987, 1994) to determine the lag length for the Newey and West (1987) estimation. The plug-in rule is calculated as floor [4(N/100)2/9], where N is the number of observations (Hoechle, 2007). Using a dataset starting in January 2000 and applying a three-year rolling window to estimate betas, we construct the sorting portfolios over the period from January 2003 to December 2020, which yields a maximum of 216 adjusted returns over the sample period. Thus, the lag length is four based on the plug-in rule. Some studies also use four lags when reporting adjusted t-statistics (e.g. Genesove & Mullin, 1998; Erutku, 2012; Chen et al., 2024).

6. Compared with equal-weighted returns, value-weighted returns weighting more on large stocks may be more accurate, as they contain less bias on noisy prices of mean returns (Asparouhova, Bessembinder, & Kalcheva, 2013; Hahn & Yoon, 2016). Large stocks are also frequently traded by investors, while small stocks with liquidity issues are thinly traded and have difficulties to construct portfolio sorting strategies in the real world (Ko et al., 2007; Kang et al., 2011).

7. This may be because the Taiwan stock market is dominated by domestic individual investors. Many individual investors are less likely to be fully rational and to have professional knowledge than institutional investors (Larson, 2007), and thus may not be able to fully digest market information or have the ability to analyze market risk for making appropriate investment decisions (Munkh-Ulzii, McAleer, Moslehpour, & Wong, 2018). As they may not react to uncertainty risk, uncertainty exposure does not drive stock returns in the Taiwan market.