The impact of the North American shale gas technology on the US’ energy security: the case of natural gas

ABSTRACT

Energy security is sensitive to the behavioural characteristics of resource trade, especially for the US as the world’s largest energy-consuming economy. This paper applies the Markov switching model and dynamic network connectedness measures of Diebold-Yilmaz to explore the impact of the North American shale gas technology on behavioural regimes and spillover effects of natural gas trade cycles, regarding the US’ energy security. The findings indicate asymmetric and time-varying behaviour of the US’ energy security pre- and post-the revolution. Specifically, it seems that overall interaction between substitution- and scale effect of the shale gas revolution through the gravity theory develops the US’ energy security as shocks occur in the energy-trading process. Therefore, the alternating analysis of behavioural regimes and spillover effects, through potential and actual trade links in world natural gas trade networks, is essential to develop availability, affordability, accessibility as well as acceptability dimensions of energy security.

KEYWORDS:

• Energy security
• asymmetric behaviour
• natural gas trade
• Markov switching model
• spillover effects

JEL CLASSIFICATION:

• Q47
• D81
• Q37

Introduction

The concept of energy security covers a wide range of aspects (Yergin 2006), from classic definition, i.e. reliable and affordable flow of oil supply (Yergin 1988; Colglazier and Deese 1983) to contemporary dimensions e.g. accessibility and environmental acceptability (Goldthau 2011). As a comprehensive definition, energy security refers to immediate physical availability, price affordability, transportation and transmission accessibility as well as environmental, political, and social acceptability dimensions of energy resources in the economies (Sutrisno, Nomaler, and Alkemade 2021; APERC 2007).¹

Particularly, in terms of immediate physical availability, the energy source is available if it is abundant enough to go on an important recoverable resource, while the economic dimension of energy security is explained by the affordability of energy resource acquisition. The accessibility dimension of energy security refers to transportation and transmission barriers, e.g. geopolitical factors, long-term sales contracts and massive infrastructure investments among others. From the aspect of environmental acceptability, the energy security reflexes the economy’s success to switch from a carbon intensive- to non-carbon-based fuel portfolio, to lower potential environmental degradation.

The role of energy security on resource- and non-resource sectors, capital formation, technology improvements and economic growth of the energy-exporting countries is inevitable, since they are vulnerable to the external market shocks (Nepal and Paija 2019; Griffiths 2017; Bilgili et al. 2016). On the other hand, as the economy is dependent on the imported-primary energy sources to cover its primary energy demand, there is a limited possibility to meet its energy consumption through domestic supply sources, which leads to a higher risk of the country’s energy supply security.

Hence, both energy importing- and exporting countries require to adopt dynamic energy trade strategies and policies and, therefore, support their energy securities (Sutrisno, Nomaler, and Alkemade 2021; Chalvatzis and Ioannidis 2017; Vivoda 2014; Cohen, Joutz, and Loungani 2011). Even for the United States (the US’) with world’s lowest energy security risk score, highest rank of such economy among 25 large energy-consuming countries makes it necessary to monitor the regional and global energy trade structures to remain less vulnerable in response to the market shocks of energy resources (the US’ chamber of commerce, 2020).

Natural gas, as a form of clean energy, has become an important energy resource in light of energy supply security (Liu et al. 2020). The global natural gas trade has experienced continuous growth rate higher than natural gas demand, contributing to incremental energy supply security. The US’ natural gas production, particularly from geological or tight oil formations (primarily deep shales), is increased through the combination of hydraulic fracturing and horizontal drilling technologies that is called ‘Shale Technology’. The exploitation of shale gas reserves is hardly recognised commercial outside of the North America (Auping et al. 2016). The US and other countries with technically recoverable shale gas resources are different through the institutional characteristics, especially regarding the large-scale extraction process of shale gas reserves (Tian et al. 2014; Kuuskraa, Stevens, and Moodhe 2013).

From the aspect of outcomes, the benefits of the shale gas technology for the economy are classified into explicit market impacts, e.g. raise in consumer surplus due to lower gas prices, increase in producer surplus and local as well as regional economic benefits, and implicit effects including air quality standards and climate change improvements as a result of $\mathrm{C}{\mathrm{O}}_2$ emissions reductions². Especially, the shale gas technology declines natural gas production cost by reducing the $\mathrm{C}{\mathrm{O}}_2$ separation costs through potential economic and technical infrastructure, which leads to lower natural gas prices. Furthermore, the intermediate technology of shale gas revolution mitigates the short-term environmental concerns of the US’ economy since the price reduction of natural gas may decrease $\mathrm{C}{\mathrm{O}}_2$ emissions (Acemoglu et al. 2019). Accordingly, the North-American shale gas technology is recognised as a potential affecting factor to analyse the short-term and long-term behavioural characteristics of the US’ energy environment.

It is noted that, energy trade decisions of individual economies have resulted in regional and global energy trade networks (Sutrisno, Nomaler, and Alkemade 2021; Shirazi et al. 2019) since the shale gas technology has not transferred across the globe (Melikoglu 2014). As excess global energy demand is increased, larger volumes of energy trade and more energy network connectedness are needed to raise energy security (Zhong et al. 2017; Kern and Rogge 2016; Cherp, Jewell, and Goldthau 2011; Csereklyei and Stern 2015; Institute for 21st Century Energy 2015). Also, the structure and performance of natural gas trade flows (sum of export and import) have been developed through energy trade systems, which needs paying attention to various factors determining the energy security level, including energy strategies (An et al. 2014). Moreover, the energy security is sensitive to energy trade shocks, e.g. import- and export shocks and non-trade shocks regarding energy reserves, geopolitics, including foreign policies and transport disruptions (Zhong et al. 2014), production capacities (Guan et al. 2016) as well as production disruptions and price shocks (Shepard and Pratson 2020). Consequently, and based on the existing dynamic substitution of natural gas in resource- and non-resource sectors (Serletis, Timilsina, and Vasetsky 2011; Renou-Maissant 1999), the rule of comparative advantage and high degree of specialisation of the economies (Antweiler, Copeland, and Taylor 2001; Hummels, Ishii, and Yi 2001), natural gas shows a considerable impact on global energy trade strategies (Ji, Zhang, and Fan 2014), economic- growth and development as well as military- and political strategies (Wu and Chen 2019).

We aim to address that the North American shale gas revolution is expected to be as a considerable affecting factor to determine the behavioural characteristics of natural gas import and export by country of the US’ economy that would lead to energy security developments, e.g. more natural gas availability, price affordability, accessibility and acceptability, through the higher expected volumes of natural gas trade flows and hence help to turn the US’ economy into a natural gas net exporter and develop the US’ energy security.

Based on EIA³ monthly energy review (2021) and throughout the US’ natural gas trade network, the major export destinations of the US’ natural gas are Canada, Japan and Mexico and, therefore, the rest of the destinations are denoted as ‘other countries’. Also, the main import sources of the US’ natural gas import are classified as Algeria, Canada, Egypt, Mexico, Nigeria, Qatar and Trinidad and Tobago and hence, ‘other countries’ for the rest of import sources. In the following, the map of the US’ natural gas trade by country is presented in Figure 1.

Figure 1. Map of the US’ natural gas trade by country, based on EIA monthly energy review (2021): (a) the US, (b) Canada, (c) Mexico, (d) Trinidad & Tobago, (e) Algeria, (f) Nigeria, (g) Egypt, (h) Qatar, (i) Japan.

The first classification of recent studies, regarding availability and accessibility dimensions of energy security, focuses on the impact of energy sources’ regional and international trade networks on energy security (Tuchinda et al. 2021; Peng et al. 2021; Shirazi et al. 2020; Shepard and Pratson 2020; Sutrisno, Nomaler, and Alkemade 2021; Rodríguez-Fernandez, Belen Fernandez Carvajal, and Manuel Ruiz-Gomez 2020; Dong et al. 2020; Shirazi et al. 2019; Maltby 2013 among others) and concludes that energy security significantly depends on reliable trade relationships throughout global trade networks of both renewables and non-renewables.

The second group of articles surveys determining the risks around energy security, e.g. energy supply, environment, technology, geopolitical and economic factors, of individual economies and regions (Hasanov et al. 2020; Karatayev and Hall 2020; Augutis et al. 2020; Lin and Yousaf Raza 2020; San-Akca, Sever, and Yilmaz 2020; Balbás Egea and Eguren Egiguren 2019; Zeng, Streimikiene, and Balezentis 2017; Cherp and Jewell 2014) and finds that energy resource diversification, renewables’ development, citizen commitment, the mobilisation of technological and economic resources and finally, a model of efficiency, generation and distribution have constructive roles in energy security enhancement.

The third category of literatures analyses the performance of energy security level based on indicators (Barik, Jaiswal, and Chandra Das 2021; Gong, Wang, and Lin 2021; Li et al. 2020; Augutis et al. 2020; Kosai and Unesaki 2020; Gasser 2020; Yuan and Luo 2019; Lobova et al. 2019; Sarangi et al. 2019; Li and Chang 2019; Le and Nguyen 2019; Fang, Shi, and Yu 2018; Chalvatzis and Ioannidis 2017; Bilgili et al. 2016; Kisel et al. 2016; Narula 2014; Stirling 2010; Kruyt et al. 2009; Scheepers et al. 2007) and finds that strategic management, control and storage of energy supply, higher reserves of energy sources, clean energy development, optimisation of the structure of terminal energy consumption and energy efficiency improvement increase the energy security level in the countries under consideration.

The fourth sort of papers investigates the use of potential opportunities to develop energy security (Gyanwali et al. 2021; Khare et al. 2021; Yong et al. 2021; Bilgili, Koçak, and Bulut 2020; Rajavuori and Huhta 2020; Bekhrad, Aslani, and Mazzuca-Sobczuk 2020; Lavidas 2019; Nikolaidis, Chatzis, and Poullikkas 2019; Azzuni and Breyer 2018; Meyer 2003) and exhibits the positive effect of investment screening projects on energy security enhancement that is applicable through wave energy, energy hub security region, cross-border transactions in energy infrastructures, energy technologies, e.g. storage technologies and shale development, and data-intensive technologies such as the digitalisation of the energy sector.

Also, from the aspect of energy security dilemma, the comparison between the transition towards renewable energy sources or prioritised fossil fuels as reliable supplies is done (Taherahmadi, Noorollahi, and Panahi 2021; Mabea 2020; Esperanza Mata Perez, Scholten, and Smith Stegen 2019; Novikau 2019; Lu et al. 2019; Gillessen et al. 2019; Nakarmi, Mishra, and Banerjee 2016; Aslani et al. 2013; Afzal, Mohibullah, and Kumar Sharma 2010). They conclude that the focus on renewables lowers the import dependence of the economy, while reliable supplies can mitigate the volatility and costs of energy environment.

Accordingly, the behavioural regimes and spillover effects of natural gas trade cycles are important uncertainty signs that should be considered through the trade networks. So, the shale gas revolution would cause higher volumes of natural gas trade and, therefore, more energy security is expected.

Hence, and to fill in the gap among the literatures, the availability, affordability, accessibility and acceptability aspects of the US’ energy security are analysed through both import and export dimensions of natural gas market. To this end, first the Hodrick and Prescott (1997) filter is applied to extract the stationary cyclical movements (short-term fluctuations) of natural gas import and export series by country of the US’ economy, as suggested by Ewing and Thompson (2007). Secondly, the impact of the North American shale gas technology on the movement regimes for country-based natural gas trade of the US’ economy is investigated, using the Markov switching model. Then, from the perspective of the dynamic generalised forecast error variance decomposition, the network connectedness measures of Diebold and Yilmaz (2016) are utilised to analyse the changes in the dynamic spillover effects between natural gas import and export cycle series by country pre- and post- the North American shale gas revolution.

Consequently, four behavioural indices, e.g. the speed of ‘downward’ regime (decrease state), typical (dominant) regime, degree of integration and uncertainty contribution⁴, are utilised in this paper to determine the dynamic behavioural characteristics of the US’ energy security in response to the North American shale gas technology. Then, the applicable and comprehensive natural gas trade strategies are suggested as important factors affecting the structure of energy conservation and vulnerability reduction to increase energy security and sustainable economic development.

Therefore, to understand the potential asymmetric and time-varying performance of energy security regarding the US’ natural gas trade network and its partners at different stages of economic development, the following research questions are investigated:

• How the behavioural regimes (i.e. types of regimes, typical regime and speed of ‘downward’ regime) of the US’ natural gas trade cycles are explained pre- and post-the North American shale gas revolution?

• What are the sensitivity levels (network structure) of the natural gas import and export cycles by country of the US’ economy in response to trade cycle shocks of the major partners? Do the network structures follow a time-varying pattern?

• How influential are the suggested shocks in the short-term fluctuations of natural gas trade networks in response to the shale gas revolution? How their degree of integration changes pre- and post-revolution?

The findings of this paper support the asymmetric behavioural regimes for the cycles of the US’ natural gas trade before and after the North American shale gas revolution. Also, the time-varying contribution of the aforementioned import and export cycles in shock transmission throughout the US’ natural gas trade network is found in response to the revolution. Finally, it seems that the US’ energy security is developed in response to the shale gas revolution, as shocks occur in the energy-trading process. The results may lead to a structural framework that is supposed to enhance the US’ and world energy security and promote sustainable economic development.

Materials and methods

Materials

The natural gas trade (i.e. export and import) data by country of the US’ economy is utilised in this study since energy security is sensitive to the behavioural characteristics of resource trade (Sutrisno, Nomaler, and Alkemade 2021; Chalvatzis and Ioannidis 2017; Vivoda 2014; Chiang, Liu, and Wen 2013; Cohen, Joutz, and Loungani 2011). Specifically, the US’ energy security is affected by the shale gas technology, through the substitution- and scale effect (Acemoglu et al. 2019). Based on the substitution effect, the shale gas technological process facilitates the replacement of oil, coal and green energies (e.g. renewables and nuclear) by natural gas in energy mix that supports the physical availability, price affordability and accessibility dimensions of energy security. Also, the high-carbon substitution effect (coal- and oil replacement by natural gas) reduces $\mathrm{C}{\mathrm{O}}_2$ emissions. In contrast, the low-carbon energy substitution effect (natural gas-green energy source substitution effect) leads to higher $\mathrm{C}{\mathrm{O}}_2$emissions. It is supposed that the overall substitution effect may potentially reduce $\mathrm{C}{\mathrm{O}}_2$ emissions from energy consumption, since the high-carbon substitution effect overcomes the low-carbon substitution effect, which develops the environmental, political, and social acceptability dimensions of energy security. On the other hand, the physical availability, price affordability and accessibility dimensions of energy security are also improved by the scale effect of the shale gas technology since the scale effect results in price reduction of energy sources, which generally increases the energy consumption and, therefore, leads to more $\mathrm{C}{\mathrm{O}}_2$ emissions and then less acceptability of the resource is expected. Consequently, and from the aspect of $\mathrm{C}{\mathrm{O}}_2$ emissions, the shale gas revolution may enhance the US’ energy security if the overall substitution effect is negative and large enough to compensate the scale effect, especially when the marginal emissions of $\mathrm{C}{\mathrm{O}}_2$ from coal and oil are greater than those of natural gas. Moreover, the long-term impact of the shale gas technology is shown by the postponement of switching from fossil fuel consumption towards green innovation, i.e. towards the technology innovation of the used green energy as input for clean energy production. In other words, the environmental, political, and social acceptability dimensions of the US’ energy security will be treated as the economy is getting trapped in fossil fuels (a permanent increase in $\mathrm{C}{\mathrm{O}}_2$ emissions from energy consumption), whereas the availability, affordability and accessibility aspects of the US’ energy security will be improved.

From the other aspect, the energy security, regarding natural gas trade spillovers, is affected through two distinct channels (Poulakidas and Joutz 2009; Adland and Cullinane 2006; Alizadeh and Nomikos 2004). The first channel is that shocks in natural gas trade flows have a direct impact on GDP per capita of country $i$, $j$.as well as the whole trade network that leads to uncertainty transmission across the natural gas trade network, based on the gravity theory⁵. The second channel is that shocks in natural gas prices due to changes in their trade flows have direct impacts on shipping companies’ transportation costs and then on energy security through the gravity theory. Bunker fuel price is correlated with natural gas and crude oil price. Notteboom and Vernimmen (2009) found that bunker prices continuously fluctuate due to market forces. As fuel cost is the main part of transport expenditure in the tanker freight market, tanker operators would look forward to a higher freight rate to compensate for the increased transportation cost when facing an upward trend in natural gas price. Therefore, the natural gas trade shocks and spillover effects related to import and export are shifted between trade participants (Meng et al. 2018). It is believed that increasing reliance on import exacerbates a country’s sensitivity to the effects of natural gas shocks as risks occur in the energy-importing process (Biresselioglu, Yelkenci, and Oz 2015; Biresselioglu, Demir, and Kandemir 2012; Lacasse and Plourde 1995). Specifically, natural gas importers mention about uncertainties regarding the secured supply process, whereas exporters mention that the main concern is highly reliable energy exports or durable demand from buyers that are needed for the domestic economic development.

Therefore, it is expected that overall interaction between substitution and scale effects of the North American shale gas revolution through the gravity theory leads to more natural gas availability, price affordability, accessibility and acceptability through the higher volumes of natural gas trade, and hence helps to turn the US’ economy into a natural gas net exporter and develop the US’ energy security.

Accordingly, the import data in billion cubic feet for each source, e.g. Algeria, Canada, Egypt, Mexico, Nigeria, Qatar, Trinidad and Tobago and other countries as well as export data for each destination, e.g. Canada, Japan, Mexico and other countries are collected from the US’ Energy Information Administration/Monthly Energy Review 2021 for the period Jan 1973–Dec 2019. Moreover, the EIA monthly energy review (2021) shows that the US’ natural gas import from Algeria is zero after the shale gas revolution. Specifically, the cyclical movements for logarithmic series of weighted⁶ monthly data of natural gas trade by country of the US’ economy are suggested in this paper by taking into account changes in quantity of natural gas import and export caused by other partners to obtain distinct behavioural regimes and spillover effects of the US’ natural gas import and export cycles. To investigate the impact of the North American shale gas revolution on behavioural characteristics of the US’ country-based natural gas trade cycles, the time period under consideration is divided using the break point in 2006 as the beginning of the revolution (Aruga 2016; Wakamatsu and Aruga 2013). Also, the North American natural gas market is found to have overlapped with multiple structural breaks during the 2008 global financial crisis (Geng, Ji, and Fan 2016; Westgaard et al. 2011). Therefore, the impact of the 2008 global financial crisis is eliminated to follow the specific impact of the North American shale gas revolution on natural gas trade cycles of the US’ economy without bias. Consequently, the period between 1 January 1973 and 31 December 2005 is mentioned as pre- the North American shale gas revolution and the period from 1 September 2009 to the end of December 2019 is considered as post- the North American shale gas revolution (Geng, Ji, and Fan 2016; Aruga 2016).

Methods

As the pre-requisite of the Markov switching autoregressive and network connectedness methodologies, this study applies the Hodrick and Prescott (1997) filter to calculate stationary cyclical movements (short-term fluctuations) of natural gas import and export series from a long-term trend.

Then, the Markov switching autoregressive model of regime heteroscedasticity is used for asymmetric analysis to explore the distinct behavioural regimes for cycles of natural gas import and export of the US’ economy, in response to the North American shale gas revolution. This model allows the cyclical movements of natural gas trade series to switch across different states, taking into account any changes in the short-term fluctuations over the mentioned time periods. Then, it can explore the trade regimes of natural gas import and export of the US’ economy pre- and post-the North American shale gas revolution, and then can reveal whether the shale gas revolution has caused the changes of the US’ energy security, based on their dominant trade regime differences (Geng, Ji, and Fan 2016).

Finally, the dynamic network connectedness measures of Diebold and Yilmaz (2016) are utilised to identify the dynamic impact of the shale gas revolution on the spillover effects between cycles of country-based natural gas trade network. In a dynamic rolling window analysis, the potential changes of the natural gas import and export spillover behaviour, and, therefore, energy security development of the US’ economy can be detected before and after the North American shale gas revolution, based on the uncertainty transmission (generalised forecast error variance decomposition) throughout the trade network (Mensi et al. 2019; Xu et al. 2019; Giuseppe Caloiaa, Cipollini, and Muzzioli 2018).

Markov switching autoregressive model of regime heteroscedasticity

In accordance with Hamilton (1989), the Markov regime-switching model is presented as follows: (1) $Y_t-{\mu }_{S_t}=\sum _{i=1}^m{\phi }_i\left(Y_{t-i}-{\mu }_{S_{t-i}}\right)+{\delta }_{S_t}{ϵ}_t$(1) where $Y_t$ denotes the cycles of country-based natural gas import and export of the US’ economy, $\mu$ and $\delta$ are considered as the mean and standard deviation of $Y_t$, respectively. Moreover, $S_t$ is a discrete variable $\left(S_t\in \left\{1,2,\dots ,k\right\}\right)$ which exhibits the cycles of natural gas import and export in different regimes. It is also noted that $\mu$ and $\delta$ are dependent on the regime $S_t$ at time $t$. Furthermore, ${\phi }_i$ is known as the parameter of model and ${ϵ}_t$ is a random variable with $i.i.d\sim N\left(0,1\right)$. In addition, the minimum Akaike Information Criterion (AIC) and the statistically significant P-values of the estimated coefficients are suggested to determine the number of states.

Based on Hamilton (1990), the unobservable discrete-time as well as state of Markov process is used to simulate $S_t$. Hence, the matrix of transition probabilities is shown by (2) $P=\left[\begin{array}{ccc}{P}_{11}& \cdots & P_{k1}\\ ⋮& \ddots & ⋮\\ P_{1k}& \cdots & P_{kk}\end{array}\right]$(2) where $P_{ij}=Pr\left[S_t=j|S_{t-1}=i\right]$ with $P_{i1}+P_{i2}+\dots +P_{ik}=1$ for all $i$. Hamilton (1990) proposes the maximum likelihood estimation method to estimate the aforementioned parameters. Moreover, the value of $S_t$ equals $j$ as ${ϵ}_t$ is $i.i.d\sim N\left(0,1\right)$ and, therefore, the conditional probability density function of variable $Y_t$ is (3) $f(Y_t|S_t=j,I_{t-1};\theta )={\displaystyle \frac{1}{\sqrt{2\pi {\sigma }_j}}\mathrm{e}\mathrm{x}\mathrm{p}\left[-{\displaystyle \frac{{\left(Y_t-{\mu }_j\right)}^2}{2{\sigma }_j^2}}\right]}$(3) where $I_{t-1}$ presents all the information that are captured until $t-1$ time. Accordingly, $\theta =\left({\mu }_1,{\mu }_2,\dots ,{\mu }_k;{\sigma }_1,{\sigma }_2,\dots ,{\sigma }_k\right)$ denotes the parameter vector of the model which can be estimated. Furthermore, as $I_{t-1}$ is conditional then the probability $f\left(S_t=j|I_{t-1};\theta \right)$ is known. Therefore, the probability density of variable $Y_t$ can be written as (4) $\begin{array}{rl}& F\left(S_t=j|I_{t-1};\theta \right)=P\left(S_t=1|I_{t-1};\theta \right)F\left(Y_t|S_t=1,I_{t-1};\theta \right)+P\left(S_t=2|I_{t-1};\theta \right)\\ & F\left(Y_t|S_t=2,I_{t-1};\theta \right)+\dots +P\left(S_t=k|I_{t-1};\theta \right)F\left(Y_t|S_t=k,I_{t-1};\theta \right)\end{array}$(4)

Moreover, the total log likelihood function for the observation period is (5) $\mathrm{l}\mathrm{n}F\left(\theta \right)={\displaystyle \frac{1}{n}\sum _{t=1}^n\mathrm{l}\mathrm{n}F\left(Y_t|I_{t-1};\theta \right)}$(5)

Also, the maximisation of the log likelihood function is considered to estimate the model coefficients.

Then, filtering probability of $S_t$ is denoted as follows: (6) $P\left(S_t=j|I_t;\theta \right)={\displaystyle \frac{F\left(S_t=j|I_{t-1};\theta \right)F\left(Y_t|S_t=j,I_{t-1};\theta \right)}{F\left(R_t|I_t;\theta \right)}}$(6)

The smooth probability suggests that the sample information of all the time points is used to determine the probability at different states. Hence, the state at each particular moment for the specified samples is considered, based on the smoothing method proposed by Kim (1994). Accordingly, the smooth probabilities of the model are determined by the following equation: (7) $\left(S_t=j|I_t;\theta \right)=\sum _{i=1}^{k}P\left(S_t=j,S_{t+1}=i|I_T;\theta \right)=P\left(S_t=j|I_t;\theta \right).\sum _{i=1}^k{\displaystyle \frac{P_{ji}\times P(S_{t+1}=i|I_T;\theta )}{P(S_{t+1}=i|I_t;\theta )}}$(7)

Finally, the expected duration of determined regimes is captured from the transition probability $P_{jj}$. Specifically, the expected duration of regime $j$ is (8) $D_{jj}={\displaystyle \frac{1}{\left(1-P_{jj}\right)}}$(8)

Therefore, and to clarify how behavioural indices regarding the Markov regime-switching model work, more energy security is assessed after the revolution as the speed of ‘downward’ regime or decrease state for the US’ natural gas import (export) increases (decreases) as well as the import (export) model exhibits the existence of typical ‘downward’ regime (‘upward’ regime or increase state).

Network connectedness measures of Diebold and Yilmaz (2016)

This paper implements the Diebold-Yilmaz measures (2016) to calculate the dynamic connectedness and degree of integration to analyse the spillover effects and interdependence of natural gas import and export cycles by country in the US’ natural gas trade networks. This methodology clarifies the direction and magnitude of the shock effect in the short-term fluctuations of the US’ natural gas trade networks, based on the variance decomposition matrix of a vector-autoregressive (VAR) model. The VAR ($p$) model can be presented as (9) $S_t=\sum _{i=1}^p{\mathrm{\Psi }}_i{S}_{t-i}+e_t,$(9) where $S_t$: the vector of cycle series of the US’ country-based natural gas import and export, ${\mathrm{\Psi }}_i:$ the companion matrix containing the coefficients. The moving average representation is $X_t={\sum }_{i=0}^{\mathrm{\infty }}{A}_i{e}_{t-i}$. $A_i$ is the recursive form of $A_i={\mathrm{\Psi }}_1{A}_{i-1}+{\mathrm{\Psi }}_2{A}_{i-2}+\dots +{\mathrm{\Psi }}_p{A}_{i-p}$, where $A_0$ is the identity matrix and $A_i=0\mathrm{f}\mathrm{o}\mathrm{r}i>0$. The moving average coefficients denoted by $X_t$ are of utmost importance in understanding the dynamic links between variables. The moving average divides the H-step-ahead forecast error variances of each variable into parts attributable to the network shocks. Specifically, there are hundreds of moving average coefficients to interpret, but variance decomposition framework transforms them to a readable format⁷.

Based on Koop, Pesaran, and Potter (1996) and Pesaran and Shin (1998), this paper follows the generalised forecast error variance decomposition approach to estimate the parameters of the network and calculate variance decompositions, when the shocks happen across the networks. As a consequence, there is no concern regarding the ordering of variables throughout the network.

So, the pairwise spillover effect for H-step ahead or variable $j$’s contribution to the H-period ahead generalised error variance of the variable $i$ determined by (10) $d_{ij}\left(H\right)={\displaystyle \frac{{\sigma }_{jj}^{-1}{\sum }_{h=0}^{H-1}{\left({\stackrel{\hat{\mathrm{\prime }}}{e}}_i{A}_h{\sum }_j^e\right)}^2}{{\sum }_{h=0}^{H-1}({\stackrel{\hat{\mathrm{\prime }}}{e}}_i{A}_h(\sum _{}^{A_h}\hat{\mathrm{\prime }}{e}_j)}}$(10) where ${\sigma }_{jj}$ is the standard deviation of errors (${ϵ}_j$), ∑ is the variance matrix of ${ϵ}_j$, and $e_i$, $e_j$ are the selection column vectors with $i$th element unity and zeros elsewhere, respectively. $d_{ij}\left(H\right)$ is the fraction of the H-step-ahead error variance of $i$ from shocks to $j$. In the generalised variance decomposition matrix, diagonals and off-diagonals present own- (variable $i$ to itself) and pairwise contributions (variable $i$ to variable $j$), respectively. However, row sums in generalised variance decomposition matrices are not necessarily equal to 1 and thus each entry is normalised by the row sum as follows: (11) $\widetilde{d_{ij}}\left(H\right)={\displaystyle \frac{d_{ij}\left(H\right)}{{\sum }_{j-1}^N{d}_{ij}\left(H\right)}}$(11)

Diebold and Yilmaz (2016) define the pairwise directional connectedness from $j$ to $i$ using$C_{i\leftarrow j}\left(H\right)=\widetilde{d_{ij}}\left(H\right)$. Accordingly, the total directional spillover effect is calculated to determine the coordination measures across the suggested cycles. Hence, the net pairwise directional connectedness from $j$ to $i$ is identified as follows: (12) ${\tilde{d}}_{ij}^{\mathrm{n}\mathrm{e}\mathrm{t}}\left(H\right)=C_{i\leftarrow j}\left(H\right)-C_{j\leftarrow i}\left(H\right)$(12)

Also, the directional spillover effect from others to $i$ is expressed by (13) $C_{i\leftarrow o}\left(H\right)={\displaystyle \frac{{\sum }_{\begin{array}{c}j=1\\ j\ne i\end{array}}^N\widetilde{d_{ij}}\left(H\right)}{{\sum }_{i,j=1}^N\widetilde{d_{ij}}\left(H\right)}\times 100={\displaystyle \frac{{\sum }_{\begin{array}{c}j=1\\ j\ne i\end{array}}^N\widetilde{d_{ij}}\left(H\right)}{N}\times 100}}$(13) while the directional spillover effect from $i$ to others is (14) $C_{o\leftarrow i}\left(H\right)={\displaystyle \frac{{\sum }_{\begin{array}{c}i=1\\ j\ne i\end{array}}^N\widetilde{d_{ij}}\left(H\right)}{{\sum }_{i,j=1}^N\widetilde{d_{ij}}\left(H\right)}\times 100={\displaystyle \frac{{\sum }_{\begin{array}{c}i=1\\ j\ne i\end{array}}^N\widetilde{d_{ij}}\left(H\right)}{N}\times 100}}$(14)

Moreover, the net directional spillover effect is considered as (15) $C_i^{\mathrm{n}\mathrm{e}\mathrm{t}}\left(H\right)=C_{o\leftarrow i}\left(H\right)-C_{i\leftarrow o}\left(H\right)$(15)

Finally, the spillover index (average total spillover), $C\left(H\right)$, is calculated to show the degree of integration of the network: (16) $C\left(H\right)={\displaystyle \frac{{\sum }_{\begin{array}{c}i,j=1\\ j\ne i\end{array}}^N{d}_{ij}\left(H\right)}{{\sum }_{i,j=1}^N{d}_{ij}\left(H\right)}\times 100={\displaystyle \frac{{\sum }_{\begin{array}{c}i,j=1\\ j\ne i\end{array}}^N{d}_{ij}\left(H\right)}{N}\times 100}}$(16) where $d_{ii}\left(H\right)$ is the uncertainty contribution of variable $i$ to itself, $d_{ij}\left(H\right)$ is the uncertainty contribution of variable-$j$ to $i$. The spillover index exhibits the degree of integration or the average share of each variable in shock transmission throughout the entire network. A higher degree of integration reduces the time, risks, and costs needed to reach the market and, therefore, more energy security is expected. Also, more total spillover effect (uncertainty contribution) shows more important role of the variable in shock transmission throughout the import and export network that can amplify the improvement or decline energy security process. Moreover, the net directional spillover effect is used to conclude that whether one variable is a net shock-transmitter or recipient across the network. Consequently, energy security is affected through natural gas import behavioural indices as follows: $\mathrm{i}\mathrm{f}\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{S}\mathrm{p}\mathrm{i}\mathrm{l}\mathrm{l}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}\uparrow \mathrm{\&}\mathrm{S}\mathrm{p}\mathrm{e}\mathrm{e}\mathrm{d}\mathrm{o}\mathrm{f}{}^{\mathrm{\prime }}{\mathrm{D}\mathrm{o}\mathrm{w}\mathrm{n}\mathrm{w}\mathrm{a}\mathrm{r}\mathrm{d}}^{\mathrm{\prime }}\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{i}\mathrm{m}\mathrm{e}\uparrow \downarrow \stackrel{\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}\mathrm{s}}{\to }\mathrm{E}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{g}\mathrm{y}\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{y}\uparrow \downarrow$

But, the effect of natural gas export behavioural indices on energy security is summarised as follows: $\mathrm{i}\mathrm{f}\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{S}\mathrm{p}\mathrm{i}\mathrm{l}\mathrm{l}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}\uparrow \mathrm{\&}\mathrm{S}\mathrm{p}\mathrm{e}\mathrm{e}\mathrm{d}\mathrm{o}\mathrm{f}{}^{\mathrm{\prime }}{\mathrm{U}\mathrm{p}\mathrm{w}\mathrm{a}\mathrm{r}\mathrm{d}}^{\mathrm{\prime }}\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{i}\mathrm{m}\mathrm{e}\uparrow \downarrow \stackrel{\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}\mathrm{s}}{\to }\mathrm{E}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{g}\mathrm{y}\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{y}\uparrow \downarrow$

Results and discussion

Results

Results of the Hodrick and Prescott (1997) filter

The Markov switching autoregressive- and network connectedness approaches require that no non-stationary variables are involved. Hence, the unit root tests, based on automatic bandwidth selection of Andrews (1991) and Newey and West (1994), support the conclusion that all cyclical movements (short-term fluctuations) extracted from long-term trend of the suggested series under consideration, are stationary at the 1% significance level.

The cycles of the US’ natural gas import and export by country show different behavioural characteristics such as the magnitude and duration of fluctuations (ups and downs) pre- and post- the North American shale gas revolution (Figures 2 and 3). Therefore, the asymmetric and time-varying analysis of behavioural regimes and spillover effects of the cycles may help the US’ economy to mitigate the natural gas trade fluctuations and the economic vulnerability in the line of energy security.

Figure 2. The US’ natural gas import by country: Actual, Trend & Cycle: (a) Algeria, (b) Canada, (c) Egypt, (d) Mexico, (e) Nigeria, (f) Qatar, (g) other countries.

Figure 3. The US’ natural gas export by country: Actual, Trend & Cycle: (a) Canada, (b) Japan, (c) Mexico, (d) other countries.

Results of the US’ natural gas trade behaviour pre- and post- the North American shale gas revolution

To understand how availability, affordability, accessibility as well as acceptability dimensions of the US’ energy security react in response to the North American shale gas technology, four behavioural indices, e.g. the speed of ‘downward’ regime, typical regime, degree of integration and uncertainty contribution, are analysed.

The US’ natural gas import behaviour

Regimes of the US’ natural gas import and export cycles are explained in accordance with the number of cycle states, which is basically specified on the minimum AIC value criterion as well as the statistically significance of the estimated parameters. Therefore, and based on the sign and the value of estimated parameter, two general ‘upward’ (increase state) and ‘downward’ (decrease state) regimes of the cycles on this paper are classified into the following sub-regimes, e.g. ‘slightly’, ‘moderately’ and ‘sharply’ (Geng, Ji, and Fan 2016; Zhang and Zhang 2015; Artis, Krolzig, and Toro 2004; Ferrara 2003). Especially, the ‘upward’ regime (‘downward’ regime) is assessed as the sign of the estimated parameter is positive (negative) that shows the increase (decrease) state of the specified regime. Also, as the results show more than one ‘upward’ or ‘downward’ regimes, the ‘slightly’, ‘moderately’ and ‘sharply’ regimes are detected based on the descending to ascending sizes or magnitudes of the estimated parameters, respectively. In accordance with Table 1, after the shale gas revolution and based on the monthly expected durations of the specified regimes, the typical regime of import cycles from Canada and Egypt (Mexico) switches from ‘downward’ (‘sharply upward’) to ‘upward’ (‘downward’), while the ‘downward’ (‘upward’) regime of Nigeria and Qatar (Trinidad and Tobago) remains unchanged in comparison with pre- revolution. Moreover, the speed of ‘downward’ regime for import cycles from all countries under consideration, except Nigeria, is increased by the revolution, that is aligned with Acemoglu et al. (2019), Kim, Cho, and Lee (2017), Geng, Ji, and Fan (2016) and Middleton et al. (2012).

Table 1. Results of the US’ natural gas import behaviour pre- and post- the North American shale gas revolution.

Markov regime switching model: the US’ natural gas import – before the revolution
Expected durations (months)	Algeria	Canada	Egypt	Mexico	Nigeria	Qatar	Trinidad &Tobago
‘Downward’	3.58	66.89	9.57	–	6.99	4.36	–
‘Upward’	–	11.65	5.42	–	2.94	–	–
‘Slightly Downward’	–	–	–	–	–	–	–
‘Slightly Upward’	2.53	–	–	18.64	–	1.33	13.18
‘Sharply Downward’	–	–	–	–	–	–	–
‘Sharply Upward’	15.56	–	–	23.00	–	1.98	14.40
Typical Regime	‘Sharply Upward’	‘Downward’	‘Downward’	‘Sharply Upward’	‘Downward’	‘Downward’	‘Sharply Upward’
Markov regime switching model: the US’ natural gas import – after the revolution
Expected durations (months)	Algeria	Canada	Egypt	Mexico	Nigeria	Qatar	Trinidad &Tobago
‘Downward’	–	2.03	1.75	24.37	12.52	3.48	1.16
‘Upward’	–	11.38	3.99	–	2.12	4.05	19.84
‘Slightly Downward’	–	–	–	–	–	–	–
‘Slightly Upward’	–	–	–	6.51	–	–	–
‘Sharply Downward''	–	–	–	–	–	–	–
‘Sharply Upward''	–	–	–	11.28	–	–	–
Typical Regime	–	‘Upward’	‘Upward’	‘Downward’	‘Downward’	Alternative	‘Upward’
Speed of Downward Regime	–	Increase	Increase	Increase	Decrease	Increase	Increase
Network connectedness measures: the US’ natural gas import – before the revolution
S.I: 14.4%	Algeria	Canada	Egypt	Mexico	Nigeria	Qatar	Trinidad &Tobago
Contribution to others	16.8	23.7	0	2.6	0	0	0
Contribution from others	23.2	17.5	0	2.5	0	0	0
Net spillover effect	−6.4 (Recessive)	6.2 (Dominant)	0	0.1 (Passive)	0	0	0
Network connectedness measures: the US’ natural gas import – after the revolution
S.I: 29.5%	Algeria	Canada	Egypt	Mexico	Nigeria	Qatar	Trinidad &Tobago
Contribution to others	0	60.2	19.4	37.1	41.9	18.7	16.5
Contribution from others	0	45.7	25.3	28.8	35.6	24.7	25.1
Net spillover effect	0	14.5 (Dominant)	−5.9 (Recessive)	8.3 (Dominant)	6.3 (Dominant)	−6 (Recessive)	−8.6 (Recessive)

On the other hand, the spillover index (S.I) of natural gas import by country of the US’ economy is doubled after the shale gas revolution (29.5%). Therefore, the shale gas revolution leads to markedly higher-average level of network sensitivity, degree of integration and the importance of focus for reasoning shock transmission through the natural gas import cycles that are called the internal affecting factors defining the network. In accordance with total spillover effects, the cycles of natural gas import from Canada (Algeria) meet the unilateral spillover effect to (from) the import network before the revolution. Moreover, natural gas import cycles from Canada, Algeria and Mexico contribute in shock transmission to the network respectively, while Algeria, Canada and Mexico experience uncertainty contribution levels from others throughout the network. Also, the cycles of natural gas import from Canada, Mexico and Nigeria (Trinidad and Tobago, Qatar and Egypt) are net shock-transmitters (receivers) throughout the import network after the revolution. Furthermore, natural gas import cycles from Canada, Nigeria, Mexico, Egypt, Qatar and Trinidad and Tobago contribute in shock transmission to as well as from the network, respectively. Specifically, the spillover contribution of the US’ import cycles from Canada and Mexico to the network is increased after the revolution. As the other result of shale gas revolution, the US’ import from Algeria is ceased, whereas the US starts to import natural gas from Egypt, Nigeria, Qatar, Trinidad and Tobago and some other gas exporting countries.

Hence, and compatible with the studies of Peng et al. (2021); Shirazi et al. (2020), Sutrisno, Nomaler, and Alkemade (2021), Rodríguez-Fernandez, Belen Fernandez Carvajal, and Manuel Ruiz-Gomez (2020), Dong et al. (2020), Shirazi et al. (2019), Zhong et al. (2017) and Kern and Rogge (2016) among others, more energy network connectedness is needed to raise energy security, since the average share of each cycle in forecast error variance decomposition throughout the US’ import network is positively affected by the shale gas revolution. Moreover, and consistent with Biresselioglu, Yelkenci, and Oz (2015), Biresselioglu, Demir, and Kandemir (2012) and Lacasse and Plourde (1995), increasing reliance on import exacerbates a country’s sensitivity to the effects of natural gas shocks as risks occur in the energy-importing process.

The US’ natural gas export behaviour

Based on the results of monthly expected durations of the specified regimes, the typical ‘upward’ (‘downward’) regime of export cycles to Canada and Mexico (Japan) is not affected by the revolution. But, the speed of ‘downward’ regime for export cycles to Mexico and Japan (Canada) is increased (decreased) after the revolution (Table 2). Therefore, the energy security can be increased through the strategic management, control and storage of energy supply, higher reserves of energy sources, clean energy development, optimisation of the structure of terminal energy consumption and energy efficiency improvement as the outcome of the shale gas revolution (Barik, Jaiswal, and Chandra Das 2021; Gong, Wang, and Lin 2021; Li et al. 2020; Augutis et al. 2020; Lin and Yousaf Raza 2020 among others).

Table 2. Results of the US’ natural gas export behaviour pre- and post- the North American shale gas revolution.

Markov regime switching model: the US’ natural gas export – before the revolution
Expected durations (months)	Canada	Japan	Mexico
‘Downward’	1.93	76.00	5.49
‘Upward’	–	25.50	10.14
‘Slightly Downward’	–	–	–
‘Slightly Upward’	43.46	–	–
‘Sharply Downward’	–	–	–
‘Sharply Upward’	6.20	–	–
Typical Regime	‘Slightly Upward’	‘Downward’	‘Upward’
Markov regime switching model: the US’ natural gas export – after the revolution
Expected durations (months)	Canada	Japan	Mexico
‘Downward’	1.51	8.95	1.96
‘Upward’	6.13	4.57	4.81
Typical Regime	‘Upward’	‘Downward’	‘Upward’
Speed of Downward Regime	Decrease	Increase	Increase
Network connectedness measures: the US’ natural gas export – before the revolution
S.I: 13.6%	Canada	Japan	Mexico
Contribution to others	18.7	15.0	7.0
Contribution from others	11.3	22.5	6.9
Net spillover effect	7.4 (Dominant)	−7.5 (Recessive)	0.1 (Passive)
Network connectedness measures: the US’ natural gas export – after the revolution
S.I: 25.1%	Canada	Japan	Mexico
Contribution to others	40.6	5.9	45.2
Contribution from others	37.6	10.0	38.8
Net spillover effect	3 (Dominant)	−4.1 (Recessive)	6.4 (Dominant)

Like the import behaviour, the degree of integration (S.I) across cycles of the US’ natural gas export by country is doubled after the shale gas revolution (25.1%). Hence, the average share of each cycle in forecast error variance decomposition throughout the export network is positively affected by the shale gas revolution. Moreover, natural gas export cycles to Canada (Japan) show the unilateral spillover effect to (from) the export network, while export cycles to Mexico are passive in shock transmission before the revolution. Additionally, natural gas export cycles to Canada, Japan and Mexico contribute in uncertainty transmission to the network respectively, while Japan, Canada and Mexico experience contribution levels from others throughout the network. Also, the cycles of natural gas export to Mexico and Canada (Japan and some other importing countries) are net shock-transmitters (receivers) throughout the export network after the revolution. Furthermore, natural gas export cycles to Mexico, Canada, and Japan contribute in shock transmission to as well as from the network, respectively. Specifically, the spillover contribution of the US’ export cycles to Canada and Mexico to the network is considerably increased, while the spillover contribution the US’ export cycles to Japan is markedly decreased after the revolution. Consequently, energy security enhancement is applicable through energy hub security region, cross-border transactions in energy infrastructures and energy technologies, e.g. storage technologies and shale development, that is aligned with Yong et al. (2021), Rajavuori and Huhta (2020), Bekhrad, Aslani, and Mazzuca-Sobczuk (2020) and Azzuni and Breyer (2018).

The spillover behaviour of the US’ natural gas import and export

The Hodrick and Prescott (1997) filter is also used to depict and explore the actual, short-term fluctuations and long-term trend of total spillover effects, extracted from the US’ natural gas import and export networks.

Before the revolution, the actual spillover effect exhibits consecutive U-shaped and inverted U-shaped patterns, while an upward trend is detected after the revolution. But, the short-term fluctuations of total spillover are mitigated after the shale gas revolution. Accordingly, the spillover cycles of the US’ natural gas import expose energy security development after the North American shale gas revolution (Figure 4).

Figure 4. Time-varying spillover effect of the US’ natural gas import by country pre- and post- the North American shale gas revolution: (a) pre- the North American shale gas revolution, (b) post- the North American shale gas revolution.

Also, the actual spillover effect of the US’ natural gas export exhibits consecutive U-shaped and inverted U-shaped trend, before the revolution, while the inverted U-shaped and U-shaped trend are detected, respectively, after the revolution. Moreover, the short-term fluctuations of total spillover are markedly mitigated after the shale gas revolution (Figure 5). Therefore, the US’ energy security is developed after the North American shale gas revolution, since the spillover cycles of natural gas export are mitigated.

Figure 5. Time-varying spillover effect of the US’ natural gas export by country pre- and post- the North American shale gas revolution: (a) pre- the North American shale gas revolution, (b) post- the North American shale gas revolution.

Additionally, the US’ energy security seems to be promoted after the shale gas revolution since the typical regime of import cycles switches from ‘sharply upward’ to ‘downward’ and the speed of typical ‘downward’ regime regarding total spillover of natural gas import is increased. Furthermore, the behavioural characteristics (e.g. the speed of the ‘downward’ regime and typical regime) of total spillover of the US’ natural gas export show that the shale gas revolution has no negative impact on the US’ energy security (Table 3).

Table 3. Results of total spillover of the US’ natural gas trade pre- and post- the North American shale gas revolution (Markov regime switching model).

Total spillover of the US’ natural gas import – before the revolution
Expected durations (months)	‘Slowly Upward’: 5.9	‘Moderately Upward’: 6	‘Sharply Upward’: 11	Typical Regime: ‘Sharply Upward’
Total spillover of the US’ natural gas import – after the revolution
Expected durations (months)	‘Downward’: 6.3			‘Upward’: 1.9
Speed of ‘Downward’ regime: increase			Typical regime: ‘Downward’
Total spillover of the US’ natural gas export – before the revolution
Expected durations (months)	‘Downward’: 28.6		‘Upward’: 11.3	Typical regime: ‘Downward’
Total spillover of the US’ natural gas export – after the revolution
Expected durations (months)	‘Downward’: 8			‘Upward’: 7.7
Speed of ‘Downward’ regime: slightly decrease			Typical regime: alternative

Discussion

The behavioural characteristics of the US’ natural gas import from Canada, Egypt, Mexico, Qatar and Trinidad and Tobago are in the line to enhance the US’ energy security. Specifically, they show markedly contribution in forecast error variance decomposition, when the cyclical shocks happen throughout import network. Also, the US becomes less vulnerable on natural gas import from the mentioned countries since the speed of ‘downward’ regime is increased after the revolution. Moreover, the US’ natural gas import from Nigeria indicates considerable contribution to the network. However, the behaviour of the US’ import from Nigeria is not on the way to improve the US’ energy security due to decreasing speed of the ‘downward’ regime after the revolution. Therefore, as the shocks in natural gas import cycles have a direct impact on economic growth, economic stability promotes energy security and vice versa. So, based on the gravity theory, the interaction between substitution and scale effect of the shale gas revolution lowers the US’ sensitivity to the effects of natural gas import shocks as risks occur in the energy-importing process. It can be concluded, since the overall total spillover effect of the US’ natural gas sources as well as speed of ‘downward’ regime is increased after the revolution.

But, the post-revolution behaviour of the US’ natural gas export to Canada supports the US’ energy security. Especially, the natural gas export to Canada is considered as one of the main contributors in forecast error variance decomposition throughout export network that may occur to the US’ reliable energy exports (durable demand). In addition, the speed of ‘downward’ regime is decreased as the result of substitution and scale effects of the revolution. While the behaviour of natural gas export to Mexico may not lead to more the US’ energy security, the opposite results are detected after the revolution. Furthermore, the US’ energy security may not be affected by the increasing speed of the ‘downward’ regime regarding the US’ natural gas export to Japan. Also, the US’ export to Japan does not show a markedly contribution to the US’ natural gas export network that may be due to high transportation cost (Table 2). Hence, the overall greater total spillover effect of the US’ natural gas export is accompanied with higher speed of the ‘upward’ regime that promotes the US’ energy security.

As the other sign of the US’ energy security progress, the shale gas revolution leads to more spillover index (S.I) or degree of integration across the US’ natural gas import and export networks (Tables 1 and 2). Hence, the time, risks and costs that are needed to reach the natural gas market are reduced and, therefore, the US’ energy security is increased after the shale gas revolution. Moreover, it seems that the US has increased natural gas trade flows via swaps trading with Canada and Mexico after the revolution, which improves the US’ energy security. Furthermore, the total spillover effects of behavioural regimes of the US’ natural gas import and export verify that the US’ energy security is progressed after the shale gas revolution (Table 3).

Consequently, the overall interaction between substitution and scale effect of the shale gas revolution (Acemoglu et al. 2019; Kim, Cho, and Lee 2017; Middleton et al. 2012) through the gravity theory (Biresselioglu, Yelkenci, and Oz 2015; Biresselioglu, Demir, and Kandemir 2012; Lacasse and Plourde 1995) develops the US’ energy security as shocks occur in the energy-trading process. Specifically, the substitution effect of the shale gas technology facilitates the replacement of coal, oil and green energy sources (e.g. nuclear power and renewable energy resources) by natural gas in energy mix that supports the physical availability, price affordability and accessibility dimensions of the US’ energy security. Furthermore, the intermediate technology of shale gas revolution develops the environmental, political, and social acceptability dimensions of the US’ energy security since the price reduction of natural gas decreases the $\mathrm{C}{\mathrm{O}}_2$ emissions. Also, the availability, affordability and accessibility aspects of the US’ energy security are supported through higher natural gas trade flows, caused by the scale effect of the shale revolution (Sutrisno, Nomaler, and Alkemade 2021; APERC 2007).

Finally, to validate the robustness of the results, the Markov switching specifications, e.g. Log likelihood and Durbin-Watson statistics as well as probabilities and expected durations, are compared through different non-switching regressors and a number of regimes and, therefore, the most reliable econometric results are presented. Also, different rolling window lengths and forecast horizons are used in dynamic network connectedness measures and it is found that the results are robust since the differences in the spillover effects are narrow.

Notwithstanding the potential energy security improvements, including more natural gas availability, price affordability, accessibility and acceptability, as the benefits of the shale gas technology, the technological advancements of the shale gas innovation significantly intensify the shale gas production that cause undesirable explicit market effects and environmental concerns, e.g. methane emissions as the by-products of shale gas and marine pollution due to high water intensity of hydraulic fracturing (Bilgili et al. 2016; Wang et al. 2014), local anomalies and habitat destruction (Mason, Muehlenbachs, and Olmstead 2015), which are considered as the limitations of this article and hence, suggested to analyse by further investigations.

Conclusion

Despite oversupply in the current state of natural gas markets, new energy security concerns are created since natural gas markets are transforming from regional to global and interdependent market.

While the energy security has explicit and implicit impacts on the economy, the US’ natural gas trade behaviour shapes geopolitics and affects the global energy security. The role of the North American shale gas revolution on energy security is known as a necessary condition to overcome the barriers on the way of vulnerability reduction and promotion sustainable economic development for the US as the world’s largest energy-consuming economy. To this end, a comprehensive analysis is aimed to examine asymmetric and time-varying behavioural characteristics of the US’ natural gas trade.

Accordingly, first the Markov regime switching model is applied to determine potential asymmetric movements may happen by the cycles of the US’ natural gas import and export using monthly data before and after the US’ shale gas revolution. Then, the network connectedness measures of Diebold and Yilmaz (2016) are used to follow the potential time-varying impact of the US’ shale gas revolution on contribution of natural gas trade cycles in shock transmission throughout the US’ trade network.

The findings support the asymmetric behavioural regimes for the cycles of the US’ natural gas trade before and after the North American shale gas revolution. Also, the time-varying contribution of aforementioned import and export cycles in shock transmission throughout the US’ natural gas trade network is found, in response to the revolution. Consequently, based on the research findings and to target the structure of energy conservation and vulnerability reduction through natural gas trade for expanding sustainable economic development of the US’ economy, the most important policy implications are as follows:

• Alternating investigation of behavioural regimes and spillover effects through potential and actual trade links in world natural gas trade networks is essential to assess energy security since the impacts of shale gas revolution cause higher volumes of natural gas trade

• Periodic analysis of the related internal factors defining the trade networks, i.e. preference ordering of cooperation and competition-based trade strategies, diversification of import sources and export destinations

• Periodic analysis of the related external affecting factors defining the trade networks, i.e. natural gas derivatives’ market development, energy reserves, production capacities and geopolitics

• Keep existing natural gas import sources to cover domestic demand and use the advantage of scale effect to export to Japan and Korea as the US’ traditional alliances with approximately 60% of Asian LNG demand and non-traditional emerging importers in the Asia, e.g. Bangladesh, China, India and Pakistan that have most of the future growth

• Analysis of the institutional characteristics and intermediate technology of the US’ shale gas production to facilitate utilisation of the desirable explicit and implicit energy security outcomes, especially for China, Argentina, Algeria, Canada, Mexico, Australia, South Africa, Russia and Brazil that have technically recoverable shale gas reserves, but haven’t markedly started to extract the resource due to the technological constraints

Also, the following areas of interest in the future studies are expected to be valuable for the US’ and world’s energy security improvement:

• Study the environmental concerns, e.g. methane emissions and marine pollution, of the shale gas technology

• Focus on the North-Western Europe as a potential natural gas market may enhance the US’ energy security since domestic productions’ rapid decline and increase in import requirements are accompanied with growing natural gas demand fluctuations of the region

• The development in co-operation between henry hub and NBP⁸ that may improve the security of supply along major natural gas trade routes

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Asia Pacific Energy Research Centre.

2 - See Mason, Muehlenbachs, and Olmstead (2015), for more details.

3 Energy Information Administration (2021).

4 For details, see the estimation methods.

5 Gravity Theory:$T_{ij}={\alpha }_0\left(Y_i{Y}_j/D_{ij}\right)$, $T_{ij}$ is the trade flow from country i to country j, $Y_i\mathrm{\&}{Y}_j$ are country i’s GDP and country j's GDP, respectively, $D_{ij}$ is their distance, and ${\alpha }_0$ is the constant. For details, see Tsurumi, Managi, and Hibiki (2015).

6 The US’ natural gas import and export series by each country are weighted by dividing the total natural gas import and export, respectively.

7 For more details, see Hamilton (1989) and Diebold and Yilmaz (2016).

8 National Balancing Point (UK).