International Natural Gas Market Integration.

• Author: Li, Raymond

1. INTRODUCTION

   1.1. Motivation and General Introduction

   This paper examines the degree of integration and its evolution for natural gas trade around the world. Historically, natural gas pricing around the globe has been rather divergent. It has been the view of many analysts that North America constitutes a relatively highly competitive market structure, while Europe is better represented as more oligopolistic (with considerable State participation at all stages of the supply chain), and Asia is only slowly evolving from a structure of several bilateral monopolies, with Japan the original and dominant player in the region. North American natural gas trade is conducted primarily through a highly integrated transportation pipeline grid reaching into nearly all corners of the United States, and containing substantial links into Canada and Mexico. Through displacement and backhaul mechanisms, natural gas from nearly anywhere on the continent can find a market in any part of the continent. Neither Europe nor Asia yet has this degree of market integration nor the degree of gas-on-gas competition that it supports.

   Our research motivation is to examine whether or not these geographically distant and physically disconnected (in terms of pipeline linkages) regions of the world are integrated or tending toward an integrated system in which the market prices paid in each region are related to those observed and paid in the other regions.

   1.2. Literature Review

   The literature on the integration of natural gas markets has covered regional market integration in North America and Europe, and inter-regional integration among North American, European and Asian markets. The Law of One Price is the backbone of these market integration analyses. Following the Federal Energy Regulatory Commission (FERC) reforms in the 1980s and early 1990s, a number of studies have examined the behavior of North American natural gas price trends and their inter-relations. For example, De Vany and Walls (1993), King and Cuc (1996) and Serletis (1997) apply bivariate cointegration tests to detect pairwise long-run common stochastic trends in the prices as evidence of market integration. In a more recent study, Cuddington andWang (2006) apply autoregressive models to bilateral price gaps to examine the stationarity property of the price gaps. Impulse response functions are also obtained to show the half-life of the price gaps. This approach is similar to the cointegration analysis in principle. As discussed by King and Cuc (1996), a shortcoming of the cointegration approach is that it implicitly assumes a fixed structural relationship between the variables over time and hence cannot take account of dynamic structural changes in the market. In order to overcome this shortcoming, King and Cuc (1996) supplement the cointegration findings by the Kalman filter. The Kalman filter is applied to each price pair and the time-varying market integration ([beta]) coefficients are plotted as indicators of the strength of the pricing relationships over time. In general, the empirical results of the North American studies suggest that the North American markets have been more integrated since the FERC reforms but full integration has not come about yet.

   On the European side, Asche et al. (2002) examine the German import prices from the Netherlands, Norway and Russia. Despite the significant price differentials that exist among these prices, the authors find that the differentials are stable and the prices are cointegrated under both the bivariate and multivariate framework. Neumann et al. (2006) apply the Kalman filter to analyze natural gas prices for the National Balancing Point (UK), Zeebrugge (Belgium) and Bunde (Germany). They find near perfect price convergence between the UK and Belgium but no particular pattern of development can be identified in the Belgium and Germany pair. This implies that the Continental European market is not yet integrated.

   At the international level, Siliverstovs et al. (2005) investigate the integration of natural gas markets in Europe, North America and Japan using monthly prices under the principal components and cointegration framework. They find evidence of market integration between Europe and Japan but the North American market appears to be distinct. Neumann (2009) applies the Kalman filter to daily price data from the National Balancing Point (NBP), Zeebrugge and Henry Hub. The time-varying market integration coefficient moves towards one for the NBP-Henry Hub and Zeebrugge-Henry Hub pairs, revealing a rising degree of integration of these markets. Employing the cointegration and Granger causality techniques, Brown and Yucel (2009) supply evidence that the Henry Hub and NBP prices are cointegrated and there is bidirectional causality between these prices. The findings of Neumann (2009) and Brown and Yucel (2009) suggest that shipments of LNG may have created more opportunities for arbitrage and hence tightened the markets in the Atlantic basin.

   The law of one price coupled with the cointegration approach is a popular choice for researchers examining market integration. This popularity is in large part due to the fact that the implication of cointegration (if found amongst the price series) is consistent with the notion of market integration under the law of one price framework, i.e. the prices have a long-run equilibrium relationship to which the prices will return, in case any of the prices temporarily wander off. However, there are cases where the cointegration approach is not appropriate, at least when used on its own. If the markets in question were undergoing the process of integration during the period under study, the cointegration approach may lead one to conclude that the market is not integrated, even if convergence and integration have been accomplished during the latter part of the period. Upon failing to find cointegration without any further information, it is virtually impossible for researchers to know whether the markets are in the progress of integration or not integrated at all. The Kalman filter has been used to estimate the time-varying market integration coefficient and provide further information complementing the cointegration approach. However, the Kalman filter approach, as applied in the market integration literature, relies on a bivariate law of one price formulation. Therefore, when the market integration question involves the interaction of more than two prices, the Kalman filter analysis can only reveal part of the full picture.

   This paper analyzes the integration of the natural gas markets of the U.S., EU, Japan, South Korea and Taiwan using data from January 1997 to May 2011. We are particularly interested in whether natural gas markets have become more integrated with the expansion of the LNG trade and the shale gas developments in the U.S., while taking into consideration the effects of the global financial crisis. The econometric methodology best suited to answer this question is a convergence test developed by Phillips and Sul (2007) to identify convergence groups in the data. Phillips and Sul (2007) provide a formal statistical test for detecting convergence among a group of prices as well as an algorithm for the detection of convergence subgroups in the case where some price series are not converging within a core group. Our choice of econometric techniques for this analysis is guided by the expectation that, if long run relationships exist between the price series, they are unlikely to be stable over the sample period. In this instance cointegration analysis is not suitable. Therefore, in contrast with the previous applied literature in this area, we do not base our analysis on cointegration methodologies.1 We complement our convergence tests with the application of the Kalman filter to estimate the time varying relationships between each pair of the price series. The next section reviews the pricing structure and trade patterns of the international natural gas market. Section 3 describes the econometric methodology. Section 4 presents and discusses the empirical findings. Section 5 concludes.

2. OVERVIEW OF THE INTERNATIONAL NATURAL GAS MARKET

   2.1. Pricing Structure

   In North America, most natural gas is sold under arrangements that include a pricing mechanism tied to the price of natural gas quoted at Henry Hub, Louisiana. Henry Hub (HH) is the locus of numerous transmission pipelines and underground storage facilities, making it a natural hub for pricing. The Henry Hub is also the pricing point that underlies the New York Mercantile Exchange derivatives contracts on natural gas. Further, numerous other pricing locations around the country are based on differentials from HH.

   European natural gas pricing has been largely based on links to oil products like fuel oil, whereby the price for natural gas adjusts to reflect changing market conditions for fuel oil, rather than the gas-on-gas pricing that evolved in North America in the 1980s with price directly tied to supply and demand conditions for natural gas itself. There is some movement in Europe away from oil-based pricing, but the progress appears to be slow. Some exceptions are observed, especially for natural gas trade in the United Kingdom. In the UK, natural gas...