Cross-country categorical economic policy uncertainty spillovers: evidence from a conditional connectedness TVP-VAR framework

Kingstone Nyakurukwa (School of Economics and Finance (SEF), University of the Witwatersrand, Johannesburg, South Africa)
Yudhvir Seetharam (School of Economics and Finance (SEF), University of the Witwatersrand, Johannesburg, South Africa)

Journal of Financial Economic Policy

ISSN: 1757-6385

Article publication date: 28 February 2023

Issue publication date: 13 March 2023

891

Abstract

Purpose

This study aims to investigate the dynamic interconnectedness of economic policy uncertainty (EPU), fiscal policy uncertainty (FPU) and monetary policy uncertainty (MPU) in four nations, the USA, Japan, Greece and South Korea, between 1998 and 2021.

Design/methodology/approach

To comprehend the cross-category/cross-country evolution of uncertainty connectedness, the authors use the conditional connectedness approach. By using an inclusive network, this strategy lessens the bias caused by omitted variables. The TVP-VAR method is advantageous as it eliminates outliers that may potentially skew the results and reduces the bias caused by picking arbitrary rolling windows.

Findings

Based on the findings, aggregate EPU is a net transmitter of policy uncertainties across all countries when conditional-country connectedness is used. MPU receives significantly more spillovers than FPU does across all countries, even though both are primarily recipients of uncertainties. The USA appears to be a transmitter of categorical spillovers before COVID-19, while Greece appears to be a net receiver of all category spillovers in terms of category-specific connectedness. The existence of extreme global events is also seen to cause an increase in category-specific and country-specific connectedness. Additionally, the authors report that conditional country-specific connectedness is greater than conditional category-specific connectedness.

Originality/value

This study expands existing literature in several ways. Firstly, the authors use a novel conditional connectedness approach, which has not been used to untangle cross-category/cross-country policy uncertainty connectedness. Secondly, they use the TVP-VAR approach which does not depend on rolling windows to understand dynamic connectedness. Thirdly, they use an expanded number of countries in their analysis, a departure from existing studies that have in most cases used two countries to understand categorical EPU connectedness.

Keywords

Citation

Nyakurukwa, K. and Seetharam, Y. (2023), "Cross-country categorical economic policy uncertainty spillovers: evidence from a conditional connectedness TVP-VAR framework", Journal of Financial Economic Policy, Vol. 15 No. 2, pp. 164-181. https://doi.org/10.1108/JFEP-10-2022-0256

Publisher

:

Emerald Publishing Limited

Copyright © 2023, Kingstone Nyakurukwa and Yudhvir Seetharam.

License

Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial & non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode.


1. Introduction

The past few years have witnessed an increase in the rate at which globalisation is taking place. This has resulted in economies and country-specific institutions and policies being more connected than ever before. As a result, changes in economic policy in one country are likely to have far-reaching consequences on other countries not directly privy to the policies. Several global extreme events have occurred in the past few years, including the 2008 Global Financial Crisis, the 2018 Sino–US trade war, the 2020 COVID-19 pandemic and the 2022 Russian–Ukrainian war. These events have amplified economic policy uncertainty (EPU), as previously untested policies have often been used to ameliorate the effects of different global events. It is well-documented that EPU affects the stock market, international trade, the crypto market, industrial production, employment and other macroeconomic variables and extant literature has been directed at these linkages (Abaidoo, 2018; Roma et al., 2020; Paule-Vianez et al., 2021). However, Thiem (2018) notes that only a handful of studies have attempted to understand the phenomenon itself (connectedness among different categories of EPU). At the same time, economic uncertainty is not a “one size fits all” concept. There are different categories of EPU, and besides understanding the connectedness of aggregate measures of uncertainty, it is also essential to understand the most influential categories of uncertainty within a network.

Several theoretical and empirical studies have established the connectedness of EPU and the real economy. According to Song et al. (2022), the interconnectedness of EPU and macroeconomic as well as financial variables can be theoretically explained by two mechanisms. The first is the supply–demand channel in which increasing uncertainty is associated with dwindling production motivation leading to volatility in financial markets and instability in macroeconomic variables. Several studies have proved this empirically and shown that the uncertainty associated with the EPU suppresses investing (Neuenkirch, 2012), reduces oil demand (Kang and Ratti, 2013), moderately influences stock market liquidity (Debata and Mahakud, 2018) and amplifies fluctuations in the exchange rate (Chen et al., 2020). Fluctuations in the exchange rate caused by economic uncertainty exacerbate the volatility in commodity demand in the short run and thereby affecting the commodity supply–demand relationship in the long term. This spillover of economic uncertainty can cascade to the real economy and cause various economic imbalances. The second theoretical channel in which EPU affects the real economy is through market sentiment (Bernanke, 1983; Albulescu, 2021). Due to the imperfect rationality of market players, information asymmetry increases the likelihood of the convergence effect and the herd effect. Participants are more likely to react to policies during times of crisis. Theoretically, EPU will have an impact on market players' choices and higher levels of uncertainty will slow down choices about production, investment and consumption. This could result in a flight of safe-haven capital from some nations and thereby fuelling currency devaluations, which will have a greater impact on commodity prices and further affect the macroeconomic environment and financial markets.

In this study, we, examine the connectedness of categorical EPU indices of four countries, namely, the USA, Japan, Greece and South Korea. The USA and Japan are some of the largest economies in the world, whose policy uncertainty can have far-reaching consequences beyond them. Greece and South Korea are smaller but open economies and are crucial in understanding EPU spillovers as studies have shown that consequences of policy uncertainty are more pronounced in small and open economies (Cerda et al., 2018). We specifically choose the two categories of EPU, namely, monetary policy uncertainty (MPU) and fiscal policy uncertainty (FPU), because they exert unique kinds of effects in terms of magnitude and direction (Aye, 2021), and these two policies are the main macro-control economic tools used globally (Song et al., 2022). These two categories are also the only categorical EPU variables that are universally found in the data sets of all four countries sampled for this study. In trying to understand the interdependencies of the variables, we also include the aggregate EPU [1] in our network. This is a departure from existing studies on categorical EPU connectedness which only include the categories of EPU without including the aggregate metric. Intuitively, as MPU and FPU are extracted from EPU, the implication is that EPU is most likely to be a net transmitter of uncertainty spillovers in a within-country connectedness network. This then allows us to test the validity of our methods.

The majority of the studies that have looked at the connectedness of EPU have rather used the aggregate measure. Cui and Zou (2020) use the Baruník and Křehlík (2018) framework to untangle the connectedness of EPU indices among G20 countries. They report significant EPU spillover among the countries, with the USA, France and Australia being the net transmitters, while Brazil, Italy, Mexico and Russia act as the net receivers of uncertainty spillovers. The study also reports that EPU connectedness is amplified by significant global events. Finally, the study also emphasised the importance of the frequency domain in understanding the dynamics of EPU connectedness as it reports that EPU spillovers are mostly transmitted in the short term. Klößner and Sekkel (2014) examine the spillover dynamics of EPU among six developed economies, namely, Canada, France, Germany, Italy, the UK and the USA. The study reports that the USA and UK are responsible for a significant portion of the spillovers since the financial crisis while the remaining countries are mostly net receivers.

Few studies have used categorical uncertainty indices. Antonakakis et al. (2019) examine the propagation mechanism of EPU shocks between Greece and Europe between 1998 and 2018. The Greek categorical uncertainty indices used in the study include banking policy uncertainty, currency policy uncertainty, taxes policy uncertainty, debt policy uncertainty, pension policy uncertainty and MPU. In terms of the total connectedness index (TCI), the study reports that connectedness is not static but fluctuates and seems to be sensitive to extreme global events. Thus, according to the study, TCI peaks in response to the period around the Global Financial Crisis, a period when Greece was downgraded because of increased unsustainability in its national debt. In terms of net transmissions, the European policy uncertainty is reported as being driven by the Greek uncertainty indices. Thiem (2018) uses a within-country categorical connectedness of policy uncertainties in the USA. The categorical policies used are monetary policy, fiscal policy, health-care policy, national security policy, regulatory policy and trade policy. Using the Diebold and Yilmaz (2012) [2] approach to connectedness, the author reports significant transmission of uncertainty across the various areas of American policies. Events that affect the political landscape generally, like the election of a new president, appear to have a lasting impact on the EPU network.

Guo et al. (2022) examine the connectedness of policy uncertainty indices between the USA and China in the following categories; MPU, FPU and trade policy uncertainty using monthly data sampled between 2000 and 2021. In their findings, Guo et al. (2022) report that US MPU is a net receiver of uncertainty spillover for considerable periods. They attribute the scenario to adaptation responses by the Federal Bank whenever there is a sudden shift in uncertainty at home or abroad. Nong (2021) investigates the cross-country cross-category connectedness of EPU between China and the USA in the context of the trade war between the two countries. Three categories of EPU were used: MPU, FPU and TPU. Cross-country connectedness is found to be less significant than within-country connectedness. Nong (2021) reports that cross-country connectedness is dominated by TPU spillovers between the countries, whereas within-country connectedness is dominated by spillovers between monetary and fiscal policy. Further findings show that in China, monetary policy is the main source of policy uncertainty, whereas fiscal policy is the main source in the USA. The inherent cultural, political, environmental and economic divergences between the two nations are largely responsible for the dissimilar within-country results between the two countries. Overall, there is a USA-to-China spillover of policy shocks. Jiang et al. (2019) examine the connectedness of categorical policy uncertainties (MPU, FPU, TPU) between China and the USA and report that China’s FPU is the dominant transmitter of uncertainty within the network.

Gabauer and Gupta (2018) investigate internal and external categorical EPU connectedness between the USA and Japan using a novel extension of the TVP-VAR connectedness approach of Antonakakis et al. (2020). The study reports that MPU is the most influential in the network, followed by FPU, TPU and uncertainties associated with the currency market. Our study expands on Gabauer and Gupta (2018) in several ways. We include four countries, all with categorical uncertainty indices, a departure from most previous literature which mostly use two countries in trying to untangle categorical EPU connectedness. Our sample period includes a recent novel event, the COVID-19 pandemic. It would be interesting to examine the spillover of categorical policy uncertainty in the presence of black swan events like COVID-19. We also use Stenfors et al.’s (2022) novel concept of conditional connectedness that allows us to examine the transmission mechanism of all categories of uncertainty indices (EPU, MPU and FPU) given a specific country, or all countries (USA, Japan, Greece and South Korea) given a certain uncertainty category. This novel method ameliorates the omitted variable bias, which according to Antonakakis et al. (2019) lead to incorrectly estimated spillovers. Our approach differs from previous studies on the connectedness of categorical EPU within and across countries in that we conjecture that since FPU and MPU are extracted from aggregate EPU, aggregate EPU will transmit uncertainty spillovers to both. The question we then want to answer is which of the two categories receives the most uncertainty spillover. We hypothesise that the category which receives the less uncertainty spillover between the two is the most influential.

Our results can be summarised as follows. As expected, we find that within-category dynamic total connectedness is higher for EPU, followed by FPU and MPU respectively. Conditional country dynamic total connectedness is highest for Japan followed by the USA, South Korea and Greece respectively. In terms of the net directional connectedness, aggregate EPU transmits considerably higher uncertainty spillovers to FPU than MPU, while the USA and Japan are mostly transmitters of EPU and South Korea and Greece are mostly receivers. In a nutshell, we cautiously conclude that FPU is more influential than MPU and that Japan and the USA are more influential in transmitting policy uncertainties in the network.

Our study proceeds as follows: Section 2 looks at data and methods, Section 3 presents the results from the econometric models, Section 4 discusses the findings, while Section 5 concludes.

2. Data and methods

2.1 Data

We use the aggregate economic policy indices (EPU) and two categorical economic policy indices (FPU and MPU) for four countries (USA, Japan, Greece and South Korea). We get all our data from https://www.policyuncertainty.com/index.html. The choice of the countries is motivated by the fact that these are the only countries with categorical policy uncertainty data from our data source. We specifically choose EPU, MPU and FPU because these are available across all the sampled countries. Though EPU has been reported to affect various macroeconomic variables, our study is more inclined towards financial markets, and in our view, these categories suffice to explain the financial market dynamics.

For the USA, we use a variety of sub-indices based on news data are included in the categorical data (Baker et al., 2016). These are derived using data from the Access World News database, which contains articles from more than 2,000 US newspapers. A set of categorical policy terms for each sub-index, in addition to the economic, uncertainty and policy terms are also used. The MPU sub-index, for instance, would include articles that meet the criteria for being labelled as EPU and also contain the term “federal reserve”. For South Korea, we depend on Cho and Kim's (2023) derivation of categorical EPU for the country. Several policy uncertainty indices for South Korea are created by Cho and Kim (2023) based on articles from thirteen newspapers. The Korea Press Foundation's “Bigkinds” digital archives, which house items from January 1990 to December 2021, serve as their primary source. Cho and Kim (2023) use a greater variety of keywords and more newspapers (13 as opposed to 6) than Baker et al. (2016). For Greece, the study uses EPU indices created by Hardouvelis et al. (2018) who used articles from four Greek newspapers. Their procedures are in accordance with Baker et al. (2016). Finally, we use EPU indices for Japan created by Arbatli et al. (2017) who looked at articles from four major Japanese newspapers. To ensure stationarity, the percentage changes in the aggregate EPU and its two constituents for all the countries are used. All in all, motivated by the availability of data, our sample consists of monthly data for all the categories and countries between 1 March 1998 and 1 December 2021.

2.2 Methods

2.2.1 Connectedness approach.

According to Zhang et al. (2021), the term “spillover effect” refers to a fluctuation in the financial markets that spreads from one market (asset) to another. The concept of variable connectedness measured by the spillover index has evolved through time and has been mostly formalised by Diebold and Yilmaz (2009) using vector autoregressive models in the broad tradition of Engle et al. (1990). They focus on variance decompositions and show how it is possible to combine market-wide spillover effects, encapsulating a wealth of data into a single measure of spillover. The variance decompositions connected to an N-variable VAR are used to create its measure. Some shortcomings of the seminal Diebold and Yilmaz's (2009) spillover index have led to refinements of the method and have culminated in various variants of connectedness measures. To evaluate the parameters in the DY model, a rolling window size must be set up, otherwise, the estimation of the rolling window length at the start of the sample would be lost. TVP-VAR models have been developed to ameliorate the weakness of using the rolling window approach to estimating spillover indices. The study therefore, uses these variants of TVP-VAR models. The study adopts the definition of the generalised forecast error variance decomposition (GFEVD) with a forecast horizon of H from Koop et al. (1996) and Pesaran and Shin (1998):

(1) Θijg(H)=((τ)jj1h=0H1(eiΩh(τ)(τ)ej)2h=0H1(eiΩh(τ)(τ)Ωh(τ)ej)

A zero vector with unity at the ith location is indicated by the notation ei. The normalisation of the elements is indicated in the decomposition matrix as follows:

(2) Θ˜ijg(H)=Θijg(H)j=1kΘijg(H)with j=1kΘ˜ijg.1 and ij=1kΘijg(H)=1,

The following definitions of GFEVD-based spillover measures are taken from Diebold and Yilmaz (2012):

(3) TOj,t=i=1,ijkΘ˜ij,tg(H)
(4) FROMj,t=i=1,ijkΘ˜ji,tg(H)
(5) NETj,t=TOj,tFROMj,t
(6) TCIt=i,j=1,ijkΘ˜ji,tg(H)k1

Variable js impact on variable i is shown by the symbol TOj. FROM is a representation of how i affects j. The negative (positive) value of NET indicates the net recipient (transmitter) of spillover. The average degree of total connectivity is represented by TCI.

2.2.2 TVP-VAR estimations.

Our analysis begins with an estimation of a TVP-VAR framework [3] as follows:

(7)           zt=Btzt1+ututN(0,St)vec(Bt)=vec(Bt1)+vtvtN(0,Rt)
where:

zt, zt−1 and ut are k × 1 dimensional vectors, representing all IRS series in t, t − 1, and the error term, respectively. Bt and St are k × k dimensional time-varying parameter and variance-covariance matrices and vec(Bt), vt are k2 × 1 dimensional vectors and Rt is a k2 × k2 dimensional parameter variance-covariance matrix.

2.2.3 Aggregated connectedness approach.

Using this approach, we use the decomposition approach of Gabauer and Gupta (2018). The study uses the aggregated spillovers between three categories or between four countries to provide a snapshot of the general propagation mechanism. Specifically, we want to establish how much of the spillovers are transmitted between categories disregarding countries or how much of the spillovers are transmitted between countries disregarding categories. We first aggregate the GVED for d groups – 3 in terms of categories or 4 when it comes to countries. Cmn,ta=ikmjkn˜ij,tg(H) stands for the aggregated impact group n has on group m, where km and kn represent disjoint index sets. Cmm,ta=ikmjkm˜ij,tg(H) is a special case of the previous measure and represents the initial spillovers of group m (all spillovers within the same category/country). We then estimate the group-specific spillovers as follows:

(8) Cm.,ta=n=1,mndCnm,ta
(9) Cm.,ta=n=1,mndCnm,ta
(10) Cm,ta=n=1,mndCm.,taCm.,ta
(11) Cta=dd1m=1dCm.,tadd1m=1dCm.,ta
where Cm.,ta reflects total group-specific connectedness to others, Cm.,ta reflects group-specific connectedness from others, Cm,ta is the net total group-specific connectedness and Cta is the total group-specific connectedness index.

2.2.4 Conditional connectedness approach.

The study uses the novel conditional connectedness approach of Stenfors et al. (2022), which gives peculiar insights when examining the transmission mechanism of all countries given a specific EPU category or of all EPU categories given a specific country. This method improves the interpretability of the transmission dynamics as it includes the overall network in estimations, thereby ameliorating the effects of the omitted variable bias. The conditional connectedness approach is estimated as follows:

(12) Cij,t|m=Cij,tjkmCij,t
(13) Ci.,t|m=jkm,jiCji,t|m
(14) Ci.,t|m=jkm,jiCij,t|m
(15) Ci,t|m=Ci.,t|mCi.,t|m
(16) Ct|m=n(km)n(km)1i=1n(km)Ci.,t|mn(km)n(km)1i=1n(km)Ci.,t|m
where n(km) stands for the cardinality of set km, Ci.,t|m shows the conditional total group-specific connectedness to others, Ci.,t|m shows the conditional group-specific connectedness from others, Ci,t shows the conditional net total group-specific connectedness and Ct is the conditional total group-specific connectedness index.

3. Results

3.1 Summary statistics

The summary statistics of the variables are presented in Table 1. All the variables are positively skewed and are not normally distributed. There is also evidence that all series are stationary (Elliott et al., 1996) and exhibit autocorrelation and ARCH/GARCH errors (Fisher and Gallagher, 2012).The summary statistics motivate modelling the interdependence of the variables using a TVP-VAR with heteroscedastic variance-covariances adopted in the study.

3.2 Empirical results

3.2.1 Conditional connectedness.

We start our presentation of findings with the spillovers for each category separately across different countries. The static results of the conditional category-specific connectedness are presented in Table 2.

The total connectedness index (TCI) in Table 2 shows whether, on average, comovements within a network of different countries are high or low. The main diagonal of each category represents the idiosyncratic shocks while the off-diagonal elements represent the interaction across different countries. The TCI values in Table 2 are 24.18%, 14.53% and 18.34% for EPU, FPU and MPU, respectively. The values are moderate but do not show high connectedness using category-specific connectedness. It can also be noticed that Greece is the net receiver of all categorical uncertainty spillovers while South Korea is a net receiver for MPU. USA and Japan are net transmitters of EPU, MPU and FPU spillovers, with the USA being more influential than Japan across all the categories of EPU. The results presented in Table 2 are static and only show the average total connectedness across different EPU categories. It would be more enlightening to understand the evolution of the relationships in a time-space as circumstances are always changing and are likely to affect uncertainty spillovers. Figure 1 shows the results from a TVP-VAR approach that demonstrates the evolution of categorical connectedness across time.

As expected, Figure 1 shows that the conditional category dynamic connectedness across the four countries used in the study is not static but dynamic. Notably, for most of the periods, EPU connectedness is higher than the MPU and FPU subcategories. Comparing the two categories of EPU, FPU seems to be more connected than MPU except, for the period between 2003 and 2006 where the latter has higher connectedness than FPU. It can also be noted from Figure 1 that the connectedness of aggregate EPU and its categories spike in response to significant global events. We first notice spikes in policy uncertainty across all categories between 2000 and 2001, a period that coincides with the dot-com bubble burst as well as the 9–11 September attacks. Spikes in policy uncertainty across all categories can also be unambiguously observed in 2008, a period coinciding with the global financial crisis. Finally, though less pronounced than the previous spikes, the period around 2020 experienced a spike in policy uncertainty, again across all the categories.

Figure 2 shows the conditional category net total directional connectedness. Several points can be noted on the transmitters and receivers of category-specific spillovers. Firstly, Greece is a net receiver of category-specific spillovers across all the categories of EPU. The effect is more pronounced with aggregate EPU. USA is mostly a transmitter of all types of EPU except in the aftermath of 2020, where it becomes a net receiver across all the categories. Japan and South Korea exhibit a role-shifting pattern of being transmitters and receivers of categorical EPU spillovers across time. Interestingly, South Korea and Japan are net transmitters of aggregate EPU in the same period that the USA is a net receiver of spillovers (the period after 2020). This could be attributed to the uncertainty caused by COVID-19 which originated in Asia hence Asian countries are primary transmitters of uncertainty during the period.

Table 3 shows the results on the static conditional country connectedness. The TCI values (72.71%, 70.52%, 49.96% and 63.18% for the USA, Japan, Greece and South Korea, respectively) are higher than conditional category connectedness TCI values.

Table 3 shows that EPU, MPU and FPU are most connected in the USA, followed by Japan and Korea, with Greece only having an average connectedness of 49.96%. In terms of net transmissions, EPU is a net transmitter of uncertainty in all the countries, while for USA and Korea, FPU and MPU are net receivers of uncertainty spillovers. In Japan and Greece, FPU and MPU are net transmitters and receivers of uncertainty spillovers, respectively. Interestingly, in Greece, FPU (3.66%) is a more influential transmitter of uncertainty spillover than EPU (1.74%). The dynamic evolution of conditional country-specific total connectedness of policy uncertainties is shown in Figure 3.

As shown in Figure 3, within-country connectedness is relatively high for all countries, with Japan, the USA, Korea and Greece having the highest TCI values across time respectively. It can also be noticed that connectedness is sensitive to significant global events as seen from the spikes in connectedness around the dot-com bubble burst, the 9–11 September attacks, the global financial crisis and the COVID-19 pandemic. Figure 4 shows the conditional country net total directional connectedness. In all the countries, EPU is a net transmitter of uncertainties across time and the effect is more pronounced for Japan. This is expected since EPU is an aggregate measure from which the other categories of EPU are extracted. In all the countries, MPU is mostly a net receiver of uncertainty spillovers and the effect is more pronounced around 2000 for Greece, Japan and Korea while for the USA the effect is more pronounced around the Global Financial Crisis. Japan is the only country in which FPU unambiguously transmits uncertainty spillovers across the sample period. For the rest of the other countries, values for net total directional connectedness are either close to zero or negative in most of the periods. A pattern that can be observed in USA, Greece and Korea is that MPU receives significantly higher uncertainty spillovers compared to FPU. From these results, it can cautiously be concluded that between the two categories of EPU, FPU is more influential than MPU.

3.2.2 Aggregate dimensions.

In our final analysis, we focus on the significance of both linkages across categories and countries. In line with Stenfors et al. (2022), we provide aggregate connectedness results by allowing for additional auxiliary networks where we concentrate on either the aggregate country dimension or the aggregate category dimension.

In terms of the average results, Table 4 shows strong comovements within the network as demonstrated by the TCI of 69.01%. As expected again, the EPU plays the net transmitting role while the two categories of EPU are both receivers of uncertainty spillovers. Comparing the categories of EPU, it can be noticed that it is the MPU that receives more uncertainty spillovers (−5.6%) compared to FPU (−2.78%). Figure 5 shows the aggregate category net total directional connectedness.

As shown in Figure 5, EPU is a net transmitter of uncertainties as expected. MPU is a net transmitter of spillovers in the period before 2000 before it switches to being a net receiver of spillovers across the remainder of the period. Notably, it can also be observed that across time, MPU receives higher uncertainty spillovers compared to FPU, showing that between the two categories, the latter is more influential. Findings from the aggregate country dimension are presented in Table 5 and the aggregate country net total directional connectedness is in Figure 6.

The aggregate country dimension connectedness results displayed in Table 5 show that connectedness is moderate (25.28%) and that Greece is a net receiver of policy uncertainty while Japan, the USA and Korea are all net transmitters. The dynamic aggregate country-dimension connectedness in Figure 6 shows that across the sample period, Greece is a consistent net receiver of aggregate uncertainty spillovers, while Japan and USA are consistent net transmitters of aggregate spillovers. Korea, on the hand, is mostly a net receiver of aggregate connectedness in the first half of the sample period, with a role-shifting pattern seen in the second half of the sample period, where the country becomes a net transmitter of policy uncertainty.

Also, the receiving and transmitting role of policy uncertainty is amplified during significant global events like the 9–11 attacks and the Global Financial Crisis. In a study that examined the connectedness of Greek EPU and European EPU, Antonakakis et al. (2019) report that Greek EPU dominates European EPU. Our results however reveal that in a network of four countries from different three different geographic regions, Greek is actually a net receiver of EPU spillovers. This could be driven by the different cultural, environmental and other macro-differences among these countries. It is therefore important to include as many countries as possible from different cultural, environmental, political and economic dispositions to comprehensively understand the transmission dynamics of EPU.

4. Discussion

The different results from the various methodological considerations reveal several important facts. Firstly, the USA and Japan are mostly net transmitters of EPU in the system, while South Korea and Greece are mostly net receivers. This is in line with the literature that shows that smaller and open economies are mostly receivers of policy uncertainty while larger economies are usually net transmitters (Cerda et al., 2018). Whereas in this study, we report the USA as the net transmitter of MPU, the latter is a net receiver of MPU for considerable periods. In other studies, Greece has also been reported as the net transmitter of EPU rather than a net recipient. Using a network of European countries, Antonakakis et al. (2019) show that Greece is a net transmitter of EPU between 1998 and 2018. Greece cannot have flexible monetary and fiscal policies due to its membership in the Eurozone. This could explain why it is mostly a receiver of policy uncertainty spillovers in the network involving itself and other non-Euro-zone economies. Another significant finding from the study is the spiking of policy uncertainty connectedness in the presence of significant global events like the Global Financial Crisis of 2008 and the COVID-19 pandemic of 2020. This is in line with Song et al.’s (2022) suggestion that agents are more likely to react to policies during times of crisis. Cui and Zou (2020) also report amplified connectedness of policy uncertainty during crisis periods.

5. Conclusion

The study sought to examine the dynamic connectedness between EPU, FPU and MPU in four countries, namely, the USA, Japan Greece and South Korea. We use the conditional connectedness approach of Stenfors et al. (2022) to understand the cross-category/cross-country evolution of uncertainty connectedness. This approach alleviates the omitted variable bias by including an inclusive network. The TVP-VAR approach used is beneficial as it excludes outliers that can potentially bias the results and ameliorates the bias caused by arbitrarily choosing rolling windows. The study also use the aggregate connectedness approach of Gabauer and Gupta (2018) to understand the aggregated spillovers between the three categories of EPU or between the four countries. The results show that, as expected, using conditional-country connectedness, EPU is a net transmitter of policy uncertainties in all the countries. While FPU and MPU are mostly receivers of policy uncertainties, the latter receives comparatively higher spillovers compared to the former in all the countries. In terms of category-specific connectedness, Greece seems to be a net receiver of all category spillovers, while the USA seems to be a transmitter of categorical spillovers before COVID-19. It can also be observed that both category-specific and country-specific connectedness spikes in the presence of significant global events. Also find that conditional country specific connectedness is higher compared to conditional category-specific connectedness.

Our study has various policy implications. Firstly, as FPU seems to dominate MPU, we echo the sentiment of Thiem (2018) who argues that the complexity of the EPU system can be greatly reduced by neutralising fiscal policy as the primary source and recipient of uncertainty spillovers. The research findings presented in this paper suggest that, to the extent that policymakers have control over the conception and outcome of their main political initiatives, they also have control over the structure of the EPU network. As a result, they have an added motivation to improve communication and management and create more sensible policy suggestions. It is not sufficient to merely use aggregate EPU when assessing spillover effects or making decisions since different policy categories dominate the EPU contagion at different times. Governments can implement more focused responses by identifying the different categories of external hazards. For the international investor invested in Greece or South Korea, monitoring policy uncertainty emanating from the USA and Japan is important especially during crisis periods to ameliorate the uncertainty spilling from these countries on portfolio returns. Future studies could use an extended list of countries and/or EPU categories to fully understand the transmission dynamics of policy uncertainties. Future studies could also include an extended list of categorical EPU indices such as Trade Policy Uncertainty and Exchange Rate Uncertainty.

Figures

Conditional category dynamic connectedness

Figure 1.

Conditional category dynamic connectedness

Conditional category net total directional connectedness

Figure 2.

Conditional category net total directional connectedness

Conditional country dynamic total connectedness

Figure 3.

Conditional country dynamic total connectedness

Conditional category net total directional connectedness

Figure 4.

Conditional category net total directional connectedness

Aggregate category net total directional connectedness

Figure 5.

Aggregate category net total directional connectedness

Aggregate country net total directional connectedness

Figure 6.

Aggregate country net total directional connectedness

Summary statistics

USA Japan Greece Korea
Statistic EPU FPU MPU EPU FPU MPU EPU FPU MPU EPU FPU MPU
Mean 0.040 0.060 0.145 0.016 0.025 0.078 0.028 0.051 0.116 0.011 0.032 0.051
Variance 0.103*** 0.140*** 0.483*** 0.039*** 0.061*** 0.200*** 0.059*** 0.129*** 0.381*** 0.024*** 0.077*** 0.114***
Skewness 2.603*** 1.343*** 2.613*** 0.634*** 0.984*** 1.664*** 0.769*** 1.611*** 2.900*** 0.575*** 1.616*** 0.961***
Ex.Kurtosis 12.862*** 2.791*** 10.631*** 1.230*** 2.020*** 3.950*** 1.148*** 4.869*** 13.219*** 0.524* 4.429*** 1.153***
JB 2294.44*** 178.75*** 1672.15*** 37.187*** 94.813*** 317.83*** 43.89*** 406.32*** 2483.11*** 19.04*** 358.11*** 59.83***
ERS −4.334*** −3.221*** −2.857*** −4.949*** −3.457*** −1.028 −8.714*** −2.186** −3.254*** −3.569*** −5.074*** −5.728***
Q(20) 14.173 18.022** 22.131*** 27.972*** 28.930*** 36.289*** 41.601*** 49.686*** 33.465*** 31.180*** 20.739** 50.558***
Q2(20) 1.722 10.761 2.956 9.475 8.243 5.498 9.256 12.651 16.403* 4.271 4.455 8.968
Notes:

Skewness: D’Agostino (1970) test; Kurtosis: Anscombe and Glynn (1983) test; JB: Jarque and Bera (1980) normality test; ERS: Elliott, Rothenberg and Stock (1996) unit-root test; (20) and Q2 (20): Fisher and Gallagher (2012) weighted portmanteau tests; ***, **, * represents statistical significance at the 1%, 5% and 10% level of significance respectively

Conditional category-specific connectedness table

USA Japan Greece Korea From
EPU
USA 79.25 4.97 2.37 13.41 20.75
Japan 5.18 87.33 0.96 6.53 12.67
Greece 5.59 5.25 83.55 5.62 16.45
Korea 13.1 6.31 3.24 77.35 22.65
To 23.87 16.53 6.57 25.56 TCI
Net 3.12 3.86 −9.88 2.91 24.18
FPU
USA 88.78 2.9 0.75 7.57 11.22
Japan 2.96 89.69 1.21 6.14 10.31
Greece 3.16 2.3 92.46 2.08 7.54
Korea 6.7 6.72 1.08 85.5 14.5
To 12.82 11.92 3.04 15.8 TCI
NET 1.59 1.61 −4.5 1.3 14.53
MPU
USA 86.97 6.61 1.28 5.14 13.03
Japan 6.37 85.05 1.62 6.95 14.95
Greece 3.64 4.86 89.23 2.27 10.77
Korea 5.1 7.09 4.1 83.71 16.29
To 15.11 18.56 6.99 14.36 55.03
Net 2.09 3.61 −3.78 −1.92 18.34
Note:

Results are based on a TVP-VAR (0.99,0.99) model with a lag length of order 1 (BIC) and a 10-step ahead forecast

Conditional country-specific connectedness

EPU FPU MPU FROM
USA
EPU 45.65 25.13 29.22 54.35
FPU 30.77 55.51 13.73 44.49
MPU 33.45 13.14 53.41 46.59
TO 64.21 38.27 42.94 TCI
NET 9.87 −6.22 −3.65 72.71
Japan
EPU 46.05 36.27 17.68 53.95
FPU 38.74 50.04 11.22 49.96
MPU 23.19 13.94 62.87 37.13
TO 61.93 50.21 28.9 TCI
NET 7.98 0.25 −8.22 70.52
Greece
EPU 60.42 27.3 12.28 39.58
FPU 26.23 65.55 8.22 34.45
MPU 15.1 10.8 74.1 25.9
TO 41.32 38.11 20.5 TCI
NET 1.74 3.66 −5.4 49.96
Korea
EPU 53.75 25.11 21.14 46.25
FPU 26.66 58.21 15.13 41.79
MPU 22.59 15.72 61.69 38.31
TO 49.25 40.83 36.27 TCI
NET 3 −0.96 −2.04 63.18
Note:

Results are based on a TVP-VAR (0.99, 0.99) model with a lag length of order 1 (BIC) and a 10-step ahead forecast

Category connectedness

EPU FPU MPU FROM
EPU 49.58 28.79 21.63 50.42
FPU 32.87 53.38 13.75 46.62
MPU 25.93 15.04 59.03 40.97
TO 58.8 43.84 35.38 TCI
NET 8.38 −2.78 −5.6 69.01
Note:

Results are based on a TVP-VAR (0.99, 0.99) model with a lag length of order 1 (BIC) and a 10-step ahead forecast

Aggregate category net total directional connectedness

USA Japan Greece Korea From
USA 80.18 5.12 2.48 12.22 19.82
Japan 5.04 85.18 1.72 8.07 14.82
Greece 7.03 5.45 82.71 4.8 17.29
Korea 12.45 8.32 3.13 76.1 23.9
To 24.52 18.9 7.32 25.09 TCI
Net 4.7 4.08 −9.97 1.2 25.28
Note:

Results are based on a TVP-VAR (0.99, 0.99) model with a lag length of order 1 (BIC) and a 10-step ahead forecast

Notes

1.

In the rest of the study, we use EPU to refer to aggregate EPU where it is not qualified

2.

Henceforth called the DY framework

3.

The length of the TVP-VAR is 1 in line with Bayesian Information Criterion (BIC)

References

Abaidoo, R. (2018), “Policy uncertainty and dynamics of international trade”, Journal of Financial Economic Policy, Vol. 11 No. 1, pp. 101-120, doi: 10.1108/JFEP-02-2018-0034.

Albulescu, C.T. (2021), “COVID-19 and the United States financial markets’ volatility”, Finance Research Letters, Vol. 38, p. 101699, doi: 10.1016/j.frl.2020.101699.

Anscombe, F.J. and Glynn, W.J. (1983), “Distribution of the kurtosis statistic b2 for normal samples”, Biometrika, Vol. 70 No. 1, pp. 227-234, doi: 10.1093/biomet/70.1.227.

Antonakakis, N., Chatziantoniou, I. and Gabauer, D. (2020), “Refined measures of dynamic connectedness based on time-varying parameter vector autoregressions”, Journal of Risk and Financial Management, Vol. 13 No. 4, p. 84, doi: 10.3390/jrfm13040084.

Antonakakis, N., Gabauer, D. and Gupta, R. (2019), “Greek economic policy uncertainty: does it matter for Europe? Evidence from a dynamic connectedness decomposition approach”, Physica A: Statistical Mechanics and Its Applications, Vol. 535, p. 122280, doi: 10.1016/j.physa.2019.122280.

Arbatli, E., Davis, S., Ito, A. and Miake, N. (2017), “Policy uncertainty in Japan”, NBER Working Paper 23411, National Bureau of Economic Research, available at: https://econpapers.repec.org/paper/nbrnberwo/23411.htm (accessed 24 October 2022).

Aye, G.C. (2021), “Effects of fiscal and monetary policy uncertainty on economic activity in South Africa”, (March), doi: 10.47654/V25Y2021I1P167-187.

Baker, S.R., Bloom, N. and Davis, S.J. (2016), “Measuring economic policy uncertainty*”, The Quarterly Journal of Economics, Vol. 131 No. 4, pp. 1593-1636, doi: 10.1093/qje/qjw024.

Baruník, J. and Křehlík, T. (2018), “Measuring the frequency dynamics of financial connectedness and systemic risk*”, Journal of Financial Econometrics, Vol. 16 No. 2, pp. 271-296, doi: 10.1093/jjfinec/nby001.

Bernanke, B.S. (1983), “Irreversibility, uncertainty, and cyclical investment*”, The Quarterly Journal of Economics, Vol. 98 No. 1, pp. 85-106, doi: 10.2307/1885568.

Cerda, R., Silva, Á. and Valente, J.T. (2018), “Impact of economic uncertainty in a small open economy: the case of Chile”, Applied Economics, Vol. 50 No. 26, pp. 2894-2908, doi: 10.1080/00036846.2017.1412076.

Chen, L., Du, Z. and Hu, Z. (2020), “Impact of economic policy uncertainty on exchange rate volatility of China”, Finance Research Letters, Vol. 32, p. 101266, doi: 10.1016/j.frl.2019.08.014.

Cho, D. and Kim, H. (2023), “Macroeconomic effects of uncertainty shocks: evidence from Korea”, Journal of Asian Economics, Vol. 84, p. 101571, doi: 10.1016/j.asieco.2022.101571.

Cui, J. and Zou, H. (2020), “Connectedness among economic policy uncertainties: evidence from the time and frequency domain perspectives”, Journal of Systems Science and Information, Vol. 8 No. 5, pp. 401-433, doi: 10.21078/JSSI-2020-401-33.

D’Agostino, R.B. (1970), “Transformation to normality of the null distribution of g1”, Biometrika, Vol. 57 No. 3, pp. 679-681, doi: 10.2307/2334794.

Debata, B. and Mahakud, J. (2018), “Economic policy uncertainty and stock market liquidity: does financial crisis make any difference?”, Journal of Financial Economic Policy, Vol. 10 No. 1, pp. 112-135, doi: 10.1108/JFEP-09-2017-0088.

Diebold, F.X. and Yilmaz, K. (2009), “Measuring financial asset return and volatility spillovers, with application to global equity markets*”, The Economic Journal, Vol. 119 No. 534, pp. 158-171, doi: 10.1111/j.1468-0297.2008.02208.x.

Diebold, F.X. and Yilmaz, K. (2012), “Better to give than to receive: predictive directional measurement of volatility spillovers”, International Journal of Forecasting, Vol. 28 No. 1, pp. 57-66, doi: 10.1016/j.ijforecast.2011.02.006.

Elliott, G., Rothenberg, T.J. and Stock, J.H. (1996), “Efficient tests for an autoregressive unit root”, Econometrica, Vol. 64 No. 4, pp. 813-836, doi: 10.2307/2171846.

Engle, R.F., Ito, T. and Lin, W.-L. (1990), “Meteor showers or heat waves? Heteroskedastic intra-daily volatility in the foreign exchange market”, Econometrica, Vol. 58 No. 3, pp. 525-542, doi: 10.2307/2938189.

Fisher, T.J. and Gallagher, C.M. (2012), “New weighted portmanteau statistics for time series goodness of fit testing”, Journal of the American Statistical Association, Vol. 107 No. 498, pp. 777-787, doi: 10.1080/01621459.2012.688465.

Gabauer, D. and Gupta, R. (2018), “On the transmission mechanism of country-specific and international economic uncertainty spillovers: evidence from a TVP-VAR connectedness decomposition approach”, Economics Letters, Vol. 171, pp. 63-71, doi: 10.1016/j.econlet.2018.07.007.

Guo, J., Li, Y. and Shao, Q. (2022), “Cross-category spillover effects of economic policy uncertainty between China and the US: time and frequency evidence”, Journal of Asian Economics, Vol. 80, p. 101462, doi: 10.1016/j.asieco.2022.101462.

Hardouvelis, G.A., Karalas, G., Karanastasis, D. and Samartzis, P. (2018), “Economic policy uncertainty, political uncertainty and the greek economic crisis (April 3, 2018)”, doi: 10.1016/j.asieco.2022.101571.

Jarque, C.M. and Bera, A.K. (1980), “Efficient tests for normality, homoscedasticity and serial independence of regression residuals”, Economics Letters, Vol. 6 No. 3, pp. 255-259, doi: 10.1016/0165-1765(80)90024-5.

Jiang, Y., Zhu, Z., Tian, G. and Nie, H. (2019), “Determinants of within and cross-country economic policy uncertainty spillovers: evidence from US and China”, Finance Research Letters, Vol. 31, doi: 10.1016/j.frl.2019.08.004.

Kang, W. and Ratti, R.A. (2013), “Oil shocks, policy uncertainty and stock market return”, Journal of International Financial Markets, Institutions and Money, Vol. 26, pp. 305-318, doi: 10.1016/j.intfin.2013.07.001.

Klößner, S. and Sekkel, R. (2014), “International spillovers of policy uncertainty”, Economics Letters, Vol. 124 No. 3, pp. 508-512, doi: 10.1016/j.econlet.2014.07.015.

Koop, G., Pesaran, M.H. and Potter, S.M. (1996), “Impulse response analysis in nonlinear multivariate models”, Journal of Econometrics, Vol. 74 No. 1, pp. 119-147, doi: 10.1016/0304-4076(95)01753-4.

Neuenkirch, M. (2012), “Managing financial market expectations: the role of Central bank transparency and Central bank communication”, European Journal of Political Economy, Vol. 28 No. 1, pp. 1-13.

Nong, H. (2021), “Have cross-category spillovers of economic policy uncertainty changed during the US–China trade war?”, Journal of Asian Economics, Vol. 74, p. 101312, doi: 10.1016/j.asieco.2021.101312.

Paule-Vianez, J., Lobão, J., Gómez-Martínez, R. and Prado-Román, C. (2021), “Momentum strategies in times of economic policy uncertainty”, Journal of Financial Economic Policy, Vol. 13 No. 3, pp. 285-300, doi: 10.1108/JFEP-10-2019-0198.

Pesaran, H.H. and Shin, Y. (1998), “Generalized impulse response analysis in linear multivariate models”, Economics Letters, Vol. 58 No. 1, pp. 17-29, doi: 10.1016/S0165-1765(97)00214-0.

Roma, C.M., da, S., Louzada, L.C., Roma, P.M., da, S., Goto, H. and Souma, W. (2020), “Earnings management, policy uncertainty and firm life cycle stages: evidence from publicly traded companies in the USA and Brazil”, Journal of Financial Economic Policy, Vol. 13 No. 3, pp. 371-390, doi: 10.1108/JFEP-02-2020-0031.

Song, L., Tian, G. and Jiang, Y. (2022), “Connectedness of commodity, exchange rate and categorical economic policy uncertainties – evidence from China”, The North American Journal of Economics and Finance, Vol. 60, p. 101656, doi: 10.1016/j.najef.2022.101656.

Stenfors, A., Chatziantoniou, I. and Gabauer, D. (2022), “Independent policy, dependent outcomes: a game of cross-country dominoes across European yield curves”, Journal of International Financial Markets, Institutions and Money, Vol. 81, p. 101658, doi: 10.1016/j.intfin.2022.101658.

Thiem, C. (2018), “Cross-category spillovers of economic policy uncertainty”, Working Paper 744, Ruhr Economic Papers, doi: 10.4419/86788864.

Zhang, H., Chen, J. and Shao, L. (2021), “Dynamic spillovers between energy and stock markets and their implications in the context of COVID-19”, International Review of Financial Analysis, Vol. 77, p. 101828, doi: 10.1016/j.irfa.2021.101828.

Corresponding author

Kingstone Nyakurukwa can be contacted at: knyakurukwa@gmail.com

Related articles