*Theoretical Review*

Even though the prevailing literature provides a sufficient indication of the relevant impact of the Russia–Ukraine 2014 war on the economy, it is uncommon how this ongoing war will affect the efficiency of the commodity market. The existing situation provides credible descriptions of the ongoing conflict between Russia and Ukraine from the past, but fresh evidence is missing. Specifically, several studies are unable to supplement the theoretical background of such conflicts, and thus, a theoretical explanation is missing in the literature. Thus, the current analysis argues the testable hypotheses that fully encompass the role of energy markets and other energy resources, e.g., crude oil in the Russia–Ukraine war. The scholarly evidence on this interesting phenomenon is missing, and the literature has not ascertained the direct role of this conflict on energy markets in both countries (Van de Graaf and Colgan 2017). Belyi (2016) explained some limitations of resource measurements in his study. However, Stulberg (2017) has argued that energy markets and energy act as a tactical curb for Russia, Ukraine, and the European Union. Lee (2017) reveals that the conflict between Ukraine and Russia was aroused due to the historical conflict of gas. Similarly, extracting some more understanding from the analysis of Colgan (2013), it can be further identified that four fundamental paths are playing a fundamental role in the ongoing Russia–Ukraine war. These resources are internal energy markets owned by Ukraine, existing energy resources in Ukraine, Ukraine's ability to confront the Russian energy dominion in the EU market, transit routes of Ukraine's gas, and the dependency of the EU and Ukraine on Russian gas (Colgan 2013).

Moreover, recent studies found a negative impact of the Russia and Ukraine war on the global economy, stock market, energy market, commodity price, and resources (Liadze et al. 2022; Berninger et al. 2022; Deng et al. 2022). Tosun and Eshraghi (2022) found that investors have imposed a significant penalty on the remaining firms following the invasion. The review of Mbah and Wasum (2022) revealed that the global economy has begun to feel the impact of this crisis. Inflation, which is already ravaging most global economies, is steadily rising due to the sharp increase in oil, natural gas, and food price shown within a few days of this crisis. Thus, the world economy is experiencing a negative impact on household consumption, increased uncertainty, unpredictable stock swings, supply chain disruptions, bulging utility bills, decreased investment due to political risks, and economic growth impediments. Yousaf et al. (2022), based on a regional analysis, outlined that the European and Asian regions are significantly and adversely affected by this event. Chatziantoniou et al. (2022), in their research, also proved a strong impact of the 2014 war

and other collapses in recent years; more specifically, oil and the Canadian market from G7 are transmitting strong volatility shocks.

#### **3. Data and Methodology**

To understand the spillover effects of before (1 September 2021–23 February 2022) and during (24 February–24 March 2022)<sup>3</sup> the Russian invasion of Ukraine, we use five major commodity spot prices, namely crude oil (OIL), natural gas (N.GAS), platinum (XPTUSD), silver (XAGUSD), and gold (XAUUSD), and we use the G7 (Canada, France, Germany, Italy, Japan, UK, and US) and BRIC (Brazil, Russia, India, and China) MSCI market indices for the period from 1 September 2021 to 24 March 2022. The chosen countries stand for major advanced and developing economies, affecting global development with their high degrees of commodity needs. Moreover, the data were collected from the Bloomberg database system.

As per Table 1, all the commodities are yielding positive average returns. Except for Canada, all other countries are experiencing a negative average return. Natural gas and crude oil are the most volatile commodities, and Russia has shown the highest volatility followed by the UK and Italy. Here, we may undoubtedly observe the direct impact of the Russian invasion on commodities as well as markets<sup>4</sup> . Here, in Table 1, other than platinum and natural gas, all other commodities including all the sample markets are having negative skewness, which shows that the tail of the distribution is left-skewed and longer or fatter towards the left. Gold, silver, and platinum are out of commodities, and Brazil, the US, and Japan are nearing the standard value of Kurtosis, i.e., 3, which depicts the mesokurtic shape of returns in this distribution. All returns series are stationary at a 1% significance level as per the unit root test of the ADF test (Dickey and Fuller 1979), and the Philips–Perron test (Phillips and Perron 1988).



Note: The above table illustrates the descriptive statistics for five commodities, G7 and BRIC markets (gold, silver, platinum, WTI Crude Oil, natural gas, Canada, France, Germany, Italy, Japan, UK, US, Brazil, Russia, India, and China). The period was selected daily from 1 September 2021 to 15 March 2022. Moreover, Std. Dev., JB, ADF, and PP represent standard deviations, Jarque-Bera, Augmented Dickey and Fuller, and Phillip and Perron, respectively, with superiors signifying \*\*\* *p* < 0.01.

Further, from Figure 1, clear spikes are detected at the end of February and March during the invasion time. Here, all the commodities are presenting positive peaks, while gold, platinum, and crude oil have experienced a greater intensity of volatility (Dodd et al. 2022; Costola and Lorusso 2022). Conversely, all the markets exhibit a downfall, i.e.,

*J. Risk Financial Manag.* **2022**, *15*, x FOR PEER REVIEW 7 of 20

**Brazil** −0.001 0.017 −0.308 3.173 2369.90 \*\*\* −12.32 \*\*\* −12.35 \*\*\* **Russia** −0.009 0.065 −3.918 28.792 4208.21 \*\*\* −18.93 \*\*\* −18.65 \*\*\* **India** −0.001 0.016 −2.902 20.200 1908.44 \*\*\* −11.29 \*\*\* −11.29 \*\*\* **China** −0.002 0.019 −1.629 14.153 781.84 \*\*\* −15.21 \*\*\* −15.22 \*\*\*

#### negative volatility has greater impacts than positive shocks supported by many past studies (Dimitriou et al. 2013; Boungou and Yatié 2022; Boubaker et al. 2022). March 2022. Moreover, Std. Dev., JB, ADF, and PP represent standard deviations, Jarque-Bera, Augmented Dickey and Fuller, and Phillip and Perron, respectively, with superiors signifying \*\*\* *p* < 0.01.

Note: The above table illustrates the descriptive statistics for five commodities, G7 and BRIC markets (gold, silver, platinum, WTI Crude Oil, natural gas, Canada, France, Germany, Italy, Japan, UK, US, Brazil, Russia, India, and China). The period was selected daily from 1 September 2021 to 15

**Figure 1.** Evolution of five commodities, G7, and BRIC indices from 9 January 2021 to 15 March **Figure 1.** Evolution of five commodities, G7, and BRIC indices from 9 January 2021 to 15 March 2022.

2022. To examine the return spillovers between the five major commodities, G7 and BRIC markets in a time-varying manner, we utilized the TVP-VAR method of Koop and Korobilis (2014) and integrated it using the connectedness method of Diebold and Yilmaz (2014). This particular system enables the variations to differ in time through a Kalman filter evaluation, which depends on the decay elements. By doing this, the TVP-VAR method eliminates the concern of the frequently randomly selected rolling window size, which might cause quite unpredictable or squashed parameters and a lack of important observations (Antonakakis et al. 2018, 2020; Gabauer and Gupta 2018; Korobilis and Yilmaz 2018). This version also provides unique qualities to acknowledge prospective struc-To examine the return spillovers between the five major commodities, G7 and BRIC markets in a time-varying manner, we utilized the TVP-VAR method of Koop and Korobilis (2014) and integrated it using the connectedness method of Diebold and Yilmaz (2014). This particular system enables the variations to differ in time through a Kalman filter evaluation, which depends on the decay elements. By doing this, the TVP-VAR method eliminates the concern of the frequently randomly selected rolling window size, which might cause quite unpredictable or squashed parameters and a lack of important observations (Antonakakis et al. 2018, 2020; Gabauer and Gupta 2018; Korobilis and Yilmaz 2018). This version also provides unique qualities to acknowledge prospective structural breaks and offers considerable factors to acknowledge the connection amongst the factors.

tural breaks and offers considerable factors to acknowledge the connection amongst the factors. Based upon the Bayesian information criterion (BIC), an autoregressive parameter Based upon the Bayesian information criterion (BIC), an autoregressive parameter vector method with time-varying (TVP-VAR) by Antonakakis et al. (2020) is built on the subsequent formula:

$$y\_t = A\_t Z\_{t-1} + \varepsilon\_t \quad \varepsilon\_t \sim N(0, \Sigma\_t) \tag{1}$$

$$\text{vec}\left(A\_{t}\right) = \text{vec}\left(A\_{t-1}\right) + \xi\_{t} \quad \xi\_{t} \sim N(0, \Xi\_{t}) \tag{2}$$

where *y<sup>t</sup>* , *Zt*−<sup>1</sup> and *ε<sup>t</sup>* are the *K* × 1 dimensional vector, and *A<sup>t</sup>* and Σ*<sup>t</sup>* are the *K* × *K* dimensional matrices. *vec* (*At*) and *ξ<sup>t</sup>* are *K* <sup>2</sup> <sup>×</sup> <sup>1</sup> dimensional vectors, whereas <sup>Ξ</sup>*<sup>t</sup>* is a *K* <sup>2</sup> <sup>×</sup> *<sup>K</sup>* 2 dimensional matrix. As the dynamic connectedness approach of Diebold and Yilmaz (2012, 2014) rests on the Generalised Forecast Error Variance Decomposition (GFEVD) of

subsequent formula:

(Koop et al. 1996; Pesaran and Shin 1998), it is required to transform the TVP-VAR to its TVP-VMA representation by the Wold representation theorem:

$$y\_t = \sum\_{h=0}^{\infty} A\_{h,t'} \varepsilon\_{t-i} \text{ where } A\_0 = I\_K.$$

The *H*-step ahead GFEVD models the impact a shock in series *j* has on series *i*. This can be formulated as follows:

$$\theta\_{ij,t}^{\mathcal{S}}(H) = \frac{\sum\_{h=0}^{H-1} \left(e\_i^{\prime} A\_{ht} \Sigma\_t e\_j\right)^2}{\left(e\_j^{\prime} \Sigma\_t e\_j\right) \sum\_{h=0}^{H-1} \left(e\_i A\_t \mathcal{S}\_t A\_t^{\prime} e\_i\right)}\tag{3}$$

$$\widehat{\theta}\_{ij,t}^{\mathcal{S}}(H) = \frac{\theta\_{ij,t}^{\mathcal{S}}(H)}{\sum\_{k=1}^{K} \theta\_{ij,t}^{\mathcal{S}}(H)} \tag{4}$$

where *e<sup>i</sup>* is a the *K* × 1 dimensional zero vector with unity on its *i*th position. As *θ g ij*,*t* (*H*) stands for the unscaled GFEVD (∑ *K j*=1 *ζ g ij*,*t* (*H*) 6= 1), Diebold and Yilmaz (2009, 2012, 2014) suggested to normalize it by dividing *θ g ij*,*t* (*H*) by the row sums to obtain the scaled GFEVD, *θ*e *g ij*,*t* (*H*).

The scalable GFEVD is at the core of the connectivity approach and facilitates calculating the total directional connectivity to (from) all indexes from (to) index *i*. While the total directional connectivity TO describes the effect that index *i* has on all the others, the total directional connectivity OT describes the impact that all indexes have on index *i*. These connectivity steps can be calculated by:

$$\mathcal{C}\_{i \to j, t}^{\mathcal{S}}(H) = \sum\_{j=1, i \neq j}^{K} \widetilde{\theta}\_{ji, t}^{\mathcal{S}}(H) \tag{5}$$

$$\mathcal{C}\_{i \gets j, t}^{\mathcal{S}}(H) = \sum\_{j=1, i \neq j}^{K} \widetilde{\theta}\_{ij, t}^{\mathcal{S}}(H) \tag{6}$$

Computing the difference between the TO and the FROM total directional connectedness results in the net total directional connectedness of series *i*:

$$\mathbf{C}\_{i,t}^{\mathcal{S}}(H) = \mathbf{C}\_{i \to j,t}^{\mathcal{S}}\ (H) - \mathbf{C}\_{i \gets j,t}^{\mathcal{S}}\ (H) \tag{7}$$

#### **4. Results and Discussion**

This study was conducted on five commodities, G7, and BRIC countries before and during the Russia–Ukraine war. During the invasion crisis, a drastic rise in the prices of commodities, a dramatic fall in the prices of securities, and a huge setback in trade and cross-border investments, more specifically in G-7 and BRIC economies (Wang et al. 2022; Saâdaoui et al. 2022; Orhan 2022) has occurred. This has led to high volatility around the world, especially from the invasion crisis (February 2022-on going). We used daily prices and yield data for five commodities and twelve markets (most developed and developing economies across the world). The data were collected from the Bloomberg database, by applying the formula: *ri*,*<sup>t</sup>* = *ln*(*pi*,*t*) − *ln*(*pi*,*t*−1), daily return was calculated.

#### *4.1. The Connectedness Network Spillovers*

This Russia–Ukraine war has shattered economic activities, trade patterns, market returns and commodities supply chains. We applied the network connectedness of the TVP-VAR method suggested by Koop and Korobilis (2014), which is an advanced version of the traditional Diebold and Yilmaz (2012, 2014) method and estimate for the return spillovers amongst the sample commodities and markets for the period 1 September 2021 to 24 March 2022. Invasion effects can be observed from the results of the invasion on the returns connectedness on commodities and on all sample markets.

From Figure 2a, it can be asserted that prior to the occurrence of the invasion crisis, platinum and natural gas were net recipients of spillovers, and the remaining commodities were net transmitters. It is evident that there is strong connectedness between gold and silver, as both commodities massively influence each other. This description relating to gold and silver has also been stated by (Balli et al. 2019; Naeem et al. 2022; Mbah and Wasum 2022) in their studies. Conversely, the US, China, Japan, and Brazil are the net transmitters with comparatively low intensity, and the rest are recipients. It is quite apparent in the case of capital markets that the UK and other European markets are the most connected markets due to a member of regional economic integration (EU) in the sample countries transmitting the risk/return to each other among European countries. Canada is one of the largest transmitters in the network and is connected to the US, UK, Italy, Germany, and France. The UK is the largest receiver of the spillovers due to major EU countries in the sample data. Before the crisis, Russia, the US, India, China, Japan, and Brazil reflected a lesser connectedness pattern.

Subsequently, an opposite picture is displayed in Figure 2b, where a nest of connections has been presented not only among commodities and capital markets but also within each other, which reflect the consequent effects of the crisis already proven by (Wen et al. 2020a; Bouri et al. 2021a, 2021b; Umar et al. 2022a) in the past, such that commodities were also treated as an alternative investment, more particularly gold and silver. During the invasion crisis, gold and silver are net transmitters, and crude oil, platinum, and natural gas are net recipients. Conversely, most of the capital markets are net transmitters, as they are most affected by the crisis, but only the US, Brazil, China, and Canada are the recipient(s). Conclusively, the ongoing invasion has enormous consequences for sample countries, and it has affected the overall economic positioning of all the sample markets. From the literature, the studies of (Mazur et al. 2020; Bedowska-Sojka et al. 2022; Federle et al. 2022) have also asserted similar effects in the past.

Additionally, a nest is formed among the commodities and markets reflecting high intensity of volatility spillover because risk is being transmitted among them during this GPC. During war, gold and silver among commodities and Japan from markets changed their status from net transmitters to net receivers (Wang et al. 2022). Conversely, natural gas, platinum, and Canada turns net transmitters during the Russian invasion. An interesting observation can be seen that commodities were hardly connected with markets during pre-war time, but huge spillover connectedness is detected during the war (Wen et al. 2020a; Bouri et al. 2021a; Umar et al. 2022a).

**Figure 2.** Network connectedness spillovers between the five commodities, G7, and BRIC markets. Additionally, within the network, the size of the node indicates the magnitude of the contribution of every index to the connectivity of the system, while the colour indicates the origin of the connectivity. The size of the node indicates the level of overflow, and the colour determines whether the market is a net sender (green) or a recipient (pink) of spillover. The finite directional layout algorithm determines the position of the vertices, with the number of vectors determining the route of the vertices. The width of the arrow indicates the strength of the multiple gradients, and the colour determines the direction of the gradient from the strongest (red) to the weakest (black). Note: The outcomes are constructed on a first-order TVP-VAR model with a first-order delay length and a 20 level generalized forecast error variance within the estimates. (**a**) Pre-Russian invasion of Ukraine. (**b**) During Russian invasion of Ukraine. **Figure 2.** Network connectedness spillovers between the five commodities, G7, and BRIC markets. Additionally, within the network, the size of the node indicates the magnitude of the contribution of every index to the connectivity of the system, while the colour indicates the origin of the connectivity. The size of the node indicates the level of overflow, and the colour determines whether the market is a net sender (green) or a recipient (pink) of spillover. The finite directional layout algorithm determines the position of the vertices, with the number of vectors determining the route of the vertices. The width of the arrow indicates the strength of the multiple gradients, and the colour determines the direction of the gradient from the strongest (red) to the weakest (black). Note: The outcomes are constructed on a first-order TVP-VAR model with a first-order delay length and a 20-level generalized forecast error variance within the estimates. (**a**) Pre-Russian invasion of Ukraine. (**b**) During Russian invasion of Ukraine.

#### *4.2. Averaged Total Returns Spillovers 4.2. Averaged Total Returns Spillovers*

To clarify the effect of ongoing GPC, we have also presented the total time-varying (averaged total returns) spillovers between the five commodities and all the sample countries. In Figure 3, it is shown that before the start of war, the spread of COVID-19 was settling down. The spillover effect was decreasing from its peak level of 86% during the second wave of COVID-19 in the month of September 2021 to around 57% in the month of January 2022. However, this spillover augmented in February due to the sudden start of border tensions between the two companion counterparts. After this, a strong spike in spillover effect was observed that crossed the level of 65%. However, this increasing level stopped and settled at 60%, as the war force was limited and peace talks between the two countries were opened. This again supports the findings of (Adams et al. 2015), which suggest that return spillover collectively increased among all the commodities and markets during war crises (Boungou and Yatié 2022; Chatziantoniou et al. 2022; Umar et al. 2022b). In the process of such uncertain events, even limited diversification opportunities were available due to a high degree of spillovers among all markets and commodities (Wen et al. 2020a; Jiang et al. 2020; Naeem et al. 2022). To clarify the effect of ongoing GPC, we have also presented the total time-varying (averaged total returns) spillovers between the five commodities and all the sample countries. In Figure 3, it is shown that before the start of war, the spread of COVID-19 was settling down. The spillover effect was decreasing from its peak level of 86% during the second wave of COVID-19 in the month of September 2021 to around 57% in the month of January 2022. However, this spillover augmented in February due to the sudden start of border tensions between the two companion counterparts. After this, a strong spike in spillover effect was observed that crossed the level of 65%. However, this increasing level stopped and settled at 60%, as the war force was limited and peace talks between the two countries were opened. This again supports the findings of (Adams et al. 2015), which suggest that return spillover collectively increased among all the commodities and markets during war crises (Boungou and Yatié 2022; Chatziantoniou et al. 2022: Umar et al. 2022b). In the process of such uncertain events, even limited diversification opportunities were available due to a high degree of spillovers among all markets and commodities (Wen et al. 2020a; Jiang et al. 2020; Naeem et al. 2022).

Averaged total return of spillovers between five commodities, G7, and BRIC markets.

**Figure 3.** Total time-varying spillovers between five commodities, G7, and BRIC indices. Note: See Figure 2. **Figure 3.** Total time-varying spillovers between five commodities, G7, and BRIC indices. Note: See Figure 2.

#### *4.3. Net Total, "To", and "From" Return Spillovers 4.3. Net Total, "To", and "From" Return Spillovers*

To better understand the spillovers, more specifically during critical periods, we analysed the time-varying behaviour of interconnectedness between commodities and stock markets. Consequently, we also applied the total return spillovers (TO, FROM, NET) as exhibited in Figures 4–6 from all commodities and markets to each commodity and market, respectively. In Figures 4 and 5, total dynamic spillovers to/from each series are displayed and are bidirectional. To better understand the spillovers, more specifically during critical periods, we analysed the time-varying behaviour of interconnectedness between commodities and stock markets. Consequently, we also applied the total return spillovers (TO, FROM, NET) as exhibited in Figures 4–6 from all commodities and markets to each commodity and market, respectively. In Figures 4 and 5, total dynamic spillovers to/from each series are displayed and are bidirectional.

Figure 4 shows the spillover transferred to other commodities and markets, where except for natural gas, all other commodities showed a substantial return spillover to other commodities and markets. Platinum, silver, and gold have shown strong spillover variation during the months of February and March even before the invasion started because Russia is one of the largest exporters of these commodities in the world markets <sup>5</sup> . Conversely, almost every market has transmitted return spillover to other markets, and some Figure 4 shows the spillover transferred to other commodities and markets, where except for natural gas, all other commodities showed a substantial return spillover to other commodities and markets. Platinum, silver, and gold have shown strong spillover variation during the months of February and March even before the invasion started because Russia is one of the largest exporters of these commodities in the world markets <sup>5</sup> . Conversely, almost every market has transmitted return spillover to other markets, and some have reflected spillover effects before the war as well, but post-war peaked spikes can be seen in

each market. Canada seems to be exceptional, as it shows a continuously rising spillover effect since September 2021 due to a slowdown in the economy, but the spillover was further aggravated during the event (Sher 2020). Another important observation is that G7 (except Canada) markets were largely impacted by this war (Federle et al. 2022; Umar et al. 2022b). The US, India, China, and Japan are the largest transmitters to commodities and other markets of the study. This is proven because the US and Japan are one of the largest economies, while India and China are the principal emerging economies in the world. spillover effect since September 2021 due to a slowdown in the economy, but the spillover was further aggravated during the event (Sher 2020). Another important observation is that G7 (except Canada) markets were largely impacted by this war (Federle et al. 2022; Umar et al. 2022b). The US, India, China, and Japan are the largest transmitters to commodities and other markets of the study. This is proven because the US and Japan are one of the largest economies, while India and China are the principal emerging economies in the world.

have reflected spillover effects before the war as well, but post-war peaked spikes can be seen in each market. Canada seems to be exceptional, as it shows a continuously rising

*J. Risk Financial Manag.* **2022**, *15*, x FOR PEER REVIEW 12 of 20

**Figure 4.** Total return spillovers "TO" others. Note: See Figure 2. **Figure 4.** Total return spillovers "TO" others. Note: See Figure 2.

From Figure 5, quite a different image is observed, as WTI crude oil is a prominent recipient of return spillover because EU countries are consuming almost 40% crude oil from Russia (Schiffling and Valantasis Kanellos 2022). Next, platinum, gold, silver, and natural gas (less intensity) are also receiving return spillover from other commodities and markets, but gold and natural gas are experiencing comparatively less spillover effects. In the case of capital markets, other than Canada, all other markets show huge spikes of return spillover from other commodities and markets. Importantly, all European countries were experiencing (receiving) spillover effects not only before the war but also during the war, as they have strong trade ties with both warring countries (Jiang et al. 2020; Berninger et al. 2022; Adekoya et al. 2022). Regarding the BRIC countries, Russia, China, and India are the key players in return spillovers from the commodities and capital markets. From Figure 5, quite a different image is observed, as WTI crude oil is a prominent recipient of return spillover because EU countries are consuming almost 40% crude oil from Russia (Schiffling and Valantasis Kanellos 2022). Next, platinum, gold, silver, and natural gas (less intensity) are also receiving return spillover from other commodities and markets, but gold and natural gas are experiencing comparatively less spillover effects. In the case of capital markets, other than Canada, all other markets show huge spikes of return spillover from other commodities and markets. Importantly, all European countries were experiencing (receiving) spillover effects not only before the war but also during the war, as they have strong trade ties with both warring countries (Jiang et al. 2020; Berninger et al. 2022; Adekoya et al. 2022). Regarding the BRIC countries, Russia, China, and India are the key players in return spillovers from the commodities and capital markets.

**Figure 5.** Total return spillovers "FROM" others. Note: See Figure 2. **Figure 5.** Total return spillovers "FROM" others. Note: See Figure 2.

Additionally, it is observed from Figure 6 that all the commodities are net recipients of return spillovers throughout the sample period, but the quantum is less in the case of natural gas. Crude oil and gold are the most impacted commodities from this invasion crisis, and it is supported by the outcomes from past studies (Billah et al. 2021; Chatziantoniou et al. 2022). Contrariwise, except for the US, China, Japan, and Brazil, all the remaining countries are net transmitters of return spillovers; here, France, Germany, the UK, Italy, and India show rocket spikes. Similar findings were proven by (Adams et al. 2015; Boungou and Yatié 2022; Yousaf et al. 2022; Chatziantoniou et al. 2022) during the war and pandemic crisis situations. All markets are either net recipients or transmitters post-wartime, but India is the only country that was initially a net transmitter and at the end of March, it turned into a net recipient market. This is because the Indian market recovered from the shock nearly to its pre-war level. It is evidently important for the investors, hedgers, and diversifiers from the world to capitalize on this finding on the line of (Mirzaei et al. 2021; Bedowska-Sojka et al. 2022; Mohamad 2022) for international in-Additionally, it is observed from Figure 6 that all the commodities are net recipients of return spillovers throughout the sample period, but the quantum is less in the case of natural gas. Crude oil and gold are the most impacted commodities from this invasion crisis, and it is supported by the outcomes from past studies (Billah et al. 2021; Chatziantoniou et al. 2022). Contrariwise, except for the US, China, Japan, and Brazil, all the remaining countries are net transmitters of return spillovers; here, France, Germany, the UK, Italy, and India show rocket spikes. Similar findings were proven by (Adams et al. 2015; Boungou and Yatié 2022; Yousaf et al. 2022; Chatziantoniou et al. 2022) during the war and pandemic crisis situations. All markets are either net recipients or transmitters post-wartime, but India is the only country that was initially a net transmitter and at the end of March, it turned into a net recipient market. This is because the Indian market recovered from the shock nearly to its pre-war level. It is evidently important for the investors, hedgers, and diversifiers from the world to capitalize on this finding on the line of (Mirzaei et al. 2021; Bedowska-Sojka et al. 2022; Mohamad 2022) for international investment diversification.

vestment diversification. In past studies, (Yoon et al. 2019; Mensi et al. 2022) have suggested that crisis situations place more emphasis on spillovers, which is somehow matched with this research outcome, i.e., the commodities are total positive transmitters, and at the same time, net total spillover is negative. Hence, all the commodities are net recipients from other commodities and markets. Conversely, our empirical results clearly proved that from the sample G7 and BRIC markets, the US, China, Japan, and Brazil are the net recipients, and the remaining markets have transmitted their losses to other markets and commodities. Thus, special attention should be given to France, Germany, UK, Italy, and India, who have shown rocket spikes (Zhang et al. 2020; Cepoi 2020; Boungou and Yatié 2022; Yousaf et al. In past studies, (Yoon et al. 2019; Mensi et al. 2022) have suggested that crisis situations place more emphasis on spillovers, which is somehow matched with this research outcome, i.e., the commodities are total positive transmitters, and at the same time, net total spillover is negative. Hence, all the commodities are net recipients from other commodities and markets. Conversely, our empirical results clearly proved that from the sample G7 and BRIC markets, the US, China, Japan, and Brazil are the net recipients, and the remaining markets have transmitted their losses to other markets and commodities. Thus, special attention should be given to France, Germany, UK, Italy, and India, who have shown rocket spikes (Zhang et al. 2020; Cepoi 2020; Boungou and Yatié 2022; Yousaf et al. 2022), which has proven to be similar to findings in the research of past crisis situations.

2022), which has proven to be similar to findings in the research of past crisis situations.

**Figure 6.** Net total return spillovers between Five commodities, G7, and BRIC markets. Note: See Figure 2. **Figure 6.** Net total return spillovers between Five commodities, G7, and BRIC markets. Note: See Figure 2.

#### *4.4. Robustness Checks 4.4. Robustness Checks*

In this particular section, we performed a few robustness analyses. Along with the TVP-VAR-based connectedness outcomes, we provide 50-day rolling-window VAR and quantile VAR (QVAR) results. Various window sizes happened to be utilized; nevertheless, the 50-observation rolling window revealed close correlations with the TVP-VAR results and is also utilized as a benchmark model in Diebold and Yilmaz (2009, 2012). Given that a VAR model could be determined as an equation-by-equation ordinary least squares (OLS) style, it is a provisional mean-based method and thus is vulnerable to outliers. Suppose we choose each formula by a quantile regression (or the slightest absolute deviation (LAD) regression), in such a case, we concentrate on the conditional median-based computation and can thus probably eliminate the outlier sensitivity issue of the VAR model. Although the dynamics of all three models appear quite comparable, a deeper look discloses that the TVP-VAR model readjusts quicker than its other options, as stressed in Antonakakis et al. (2020) and Korobilis and Yilmaz (2018). This is essential for the forecast of the interconnectedness and thus the risk of the analysed system. This time delay is not too problematic if we only want to track the evolution during the crises. Nevertheless, the outlier sensitivity issue of the VAR model causes inaccurate results, which are more apparent in the Russia–Ukraine war regime. In this particular section, we performed a few robustness analyses. Along with the TVP-VAR-based connectedness outcomes, we provide 50-day rolling-window VAR and quantile VAR (QVAR) results. Various window sizes happened to be utilized; nevertheless, the 50-observation rolling window revealed close correlations with the TVP-VAR results and is also utilized as a benchmark model in Diebold and Yilmaz (2009, 2012). Given that a VAR model could be determined as an equation-by-equation ordinary least squares (OLS) style, it is a provisional mean-based method and thus is vulnerable to outliers. Suppose we choose each formula by a quantile regression (or the slightest absolute deviation (LAD) regression), in such a case, we concentrate on the conditional median-based computation and can thus probably eliminate the outlier sensitivity issue of the VAR model. Although the dynamics of all three models appear quite comparable, a deeper look discloses that the TVP-VAR model readjusts quicker than its other options, as stressed in Antonakakis et al. (2020) and Korobilis and Yilmaz (2018). This is essential for the forecast of the interconnectedness and thus the risk of the analysed system. This time delay is not too problematic if we only want to track the evolution during the crises. Nevertheless, the outlier sensitivity issue of the VAR model causes inaccurate results, which are more apparent in the Russia–Ukraine war regime.

Figure 7 explains two various sensitivity analyses. Panel A shows the variations in the dynamic total connectedness by readjusting the forecast horizon. We observed that after January 2022, the variations in the measurement enhanced significantly. This could be discussed because the network was more consistent during the Russia–Ukraine war, Figure 7 explains two various sensitivity analyses. Panel A shows the variations in the dynamic total connectedness by readjusting the forecast horizon. We observed that after January 2022, the variations in the measurement enhanced significantly. This could be discussed because the network was more consistent during the Russia–Ukraine war, which showed a boost in its efficiency. Additionally, the variations in the dynamics appeared

to smooth out until the completion of the period, which might lead to the switch of the sample markets back to standard time. of the sample markets back to standard time. Lastly, Panel B shows the variant of the dynamic connectedness when we enabled

which showed a boost in its efficiency. Additionally, the variations in the dynamics appeared to smooth out until the completion of the period, which might lead to the switch

*J. Risk Financial Manag.* **2022**, *15*, x FOR PEER REVIEW 15 of 20

Lastly, Panel B shows the variant of the dynamic connectedness when we enabled the decay factor of the variance–covariance to presume various values. Thus, the decay factor of the VAR coefficient was kept constant at 0.99 because it was unconvincing that the connection throughout variables transforms from one day to another by more than 1%. We discovered that the dot grey area showing the variant of the dynamic connectedness by determining various TVP-VAR requirements did not consist of the dynamic connectedness of the VAR and QVAR values. This marks the time delay issue of the rolling-window models again. The VAR model acted significantly dissimilar to the other two models after January 2022, while the QVAR and the TVP-VAR model shared comparable co-movements. the decay factor of the variance–covariance to presume various values. Thus, the decay factor of the VAR coefficient was kept constant at 0.99 because it was unconvincing that the connection throughout variables transforms from one day to another by more than 1%. We discovered that the dot grey area showing the variant of the dynamic connectedness by determining various TVP-VAR requirements did not consist of the dynamic connectedness of the VAR and QVAR values. This marks the time delay issue of the rollingwindow models again. The VAR model acted significantly dissimilar to the other two models after January 2022, while the QVAR and the TVP-VAR model shared comparable co-movements.

**Figure 7.** Sensitivity analyses. Note: Panel (**A**): different forecast horizons are used [5, 15, 25, 35, 45]. Panel (**B**): κ<sup>1</sup> = [0.95,0.96, 0.97, 0.98, 0.99] and κ<sup>2</sup> = 0.99. Panel (**A**): Forecast Horizon Sensitivity Analysis. Panel (**B**): Decay Factor Sensitivity Analysis. **Figure 7.** Sensitivity analyses. Note: Panel (**A**): different forecast horizons are used [5, 15, 25, 35, 45]. Panel (**B**): κ<sup>1</sup> = [0.95,0.96, 0.97, 0.98, 0.99] and κ<sup>2</sup> = 0.99. Panel (**A**): Forecast Horizon Sensitivity Analysis. Panel (**B**): Decay Factor Sensitivity Analysis.

Our robustness results are also consistent, where we found that after January 2022, the variations in the measurement enhanced significantly and the network was more consistent during the Russia–Ukraine war, which shows a boost in its efficiency. Furthermore, the variations in the dynamics appeared to smooth out until the completion of the period, which might lead to the switch of the sample markets back to standard time. The decay Our robustness results are also consistent, where we found that after January 2022, the variations in the measurement enhanced significantly and the network was more consistent during the Russia–Ukraine war, which shows a boost in its efficiency. Furthermore, the variations in the dynamics appeared to smooth out until the completion of the period, which might lead to the switch of the sample markets back to standard time. The decay factor of the VAR coefficient was kept constant at 0.99 because it was unconvincing that the connection throughout the variables transforms from one day to another by more than 1%.

#### **5. Conclusions, Policy Implications, and Limitations of the Study**

This research investigated the effects of the Russian invasion crisis on the dynamic connectedness between five commodities, G7, and BRIC (leading stock) markets. This study contributed many dimensions to the literature on the spillovers of returns and volatility among sample commodities and markets during GPC caused by the Russian attack on Ukraine. More specifically, return spillovers and volatility behaviour were dissimilar in neighbouring markets (EU) and non-neighbouring markets. This study found that due to this invasion crisis, a very strong connectedness among all commodities and markets (G7 and BRIC) exists. Furthermore, the findings display that gold and silver are the receivers from the rest of the commodities and all the sample markets, whereas platinum, natural gas, silver, and crude oil are the transmitters of shocks during this invasion crisis. Except for the US, Canada, China, and Brazil (recipient), all other countries are net transmitters, where European countries have shown large intensity. Some recent studies found in the literature have also supported the current conclusions of this study, such as (Zhang et al. 2020; Cepoi 2020; Boungou and Yatié 2022; Wang et al. 2022; Yousaf et al. 2022; Chatziantoniou et al. 2022). These studies unveil the phenomenon regarding high market contagion in phases of financial crises in the wake of a huge gain in connectedness in several commodities and financial markets. Particularly, during such a war crisis, global uncertainty has increased and influenced the time-varying connectedness patterns between the commodities and capital markets.

Furthermore, the time-varying net connectedness results express strong responsiveness behaviours among all commodities and capital markets, more specifically among EU markets. This study has policy implications that could be beneficial to commodities and stock investor decisions about investments and hedging in such tumultuous situations. Policymakers, institutional investors, bankers, and international organizations are the potential users to make policy decisions. Geopolitical risk level and connectedness amongst sample commodities and markets could be the guiding force for policymakers to understand the level of systematic risk, in light of these links between commodities and their effect on financial markets, and they could be utilized to prepare strategies to diminish the effects of return spillovers between commodities and stock markets in such crises.

This study was conducted in during a specific period and concluded in a short time, which carries some limitations and will set the path for future research. Due to the paucity of time and dynamicity of the environment, this study has some limitations. First, from the BRICS combination, this study drops South Africa because BRIC countries are the top GDP contributor countries among these major emerging economies, while the South African economy (market) is the least integrated with the rest of the world in terms of trade, investments, markets, and commodities flow<sup>6</sup> (Waheeduzzaman 2011; Wei et al. 2020; Billah et al. 2021). Second, this study also left out Ukraine and Gulf markets which are the main sources of the commodities, more specifically oil and natural gas. Future research can target these research gaps to give a more robust understanding of this geopolitical crisis. Moreover, further studies can be conducted on sectoral indexes for a wide-ranging investigation of the dynamics of sectoral changes and their risk and returns. However, this study was conducted immediately after the start of the war, and the results are showing short-term consequences; future research might be conducted by taking long-term data sets post-war, which will be useful for diversifiers and hedgers post-war.

**Author Contributions:** All authors contributed equally to this study. All authors have read and agreed to the published version of the manuscript.

**Funding:** We have received no funding or any other financial support for the conduct of this research.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data that support the findings of this study is available on request.

**Conflicts of Interest:** All the authors declare that we have no potential conflict regarding the conduct of research that may interrupt the publication process.

## **Notes**


#### **References**


Naeem, Muhammad Abubakr, Linh Pham, Arunachalam Senthilkumar, and Sitara Karim. 2022. Oil shocks and BRIC markets: Evidence from extreme quantile approach. *Energy Economics* 108: 105932. [CrossRef]

Orhan, Ebru. 2022. The Effects of the Russia-Ukraine War on Global Trade. *Journal of International Trade, Logistics and Law* 8: 141.

Papathanasiou, Spyros, Dimitrios Vasiliou, Anastasios Magoutas, and Drosos Koutsokostas. 2021. Do hedge and merger arbitrage funds actually hedge? A time-varying volatility spillover approach. *Finance Research Letters* 44: 102088. [CrossRef]

