فعل و انفعالات و کشف قیمت در بازارهای لحظه ای گاز طبیعی در آمریکای شمالی

Recent advances in modeling causal flows with time series analysis are used to study relationships among eight North American natural gas spot market prices. Results indicate that the Canadian and US natural gas market is a single highly integrated market. Further results indicate that price discovery tends to reflect both regions of excess demand and supply. Across North America, Malin Hub in Oregon, Chicago Hub, Illinois, Waha, Texas, and Henry Hub, Louisiana region, are the most important markets for price discovery. Opal Hub in Wyoming is an information sink in contemporaneous time, receiving price information but passing on no price information. AECO Hub in Alberta, Canada, receives price signals from several markets and passes on information to Opal and the Oklahoma region.

Historically, the natural gas industry was one of the most highly regulated sectors in North America. Direct US Federal regulation of the natural gas industry was initiated by The 1938 Natural Gas Act in response to market power concerns. This Act gave the Federal Power Commission (FPC), now the Federal Energy Regulatory Commission (FERC), the authority to regulate natural gas interstate commerce (US Department of Energy, 2006). In 1954, the US Supreme Court decision, Phillips Petroleum Co. vs. Wisconsin, forced the FPC to extend price controls to producers. “The impact of the Phillips decision was pervasive and far-reaching, ultimately culminating in the natural gas shortages of the 1970s” (US Department of Energy, 2006). In response to the shortages, The Natural Gas Policy Act of 1978 was passed, starting the deregulation process. Regulation of the natural gas industry in Canada started with the construction of the TransCanada Pipeline and the creation of the Canadian National Energy Board in the 1950s. The TransCanada pipeline changed natural gas from a low-valued “trapped” by-product of oil production to a valuable commodity. The original Canadian natural gas market structure was one of a single buyer, transporter, and seller of natural gas (Booth, 2003). Beginning in 1985, the Canadian government began regulatory reforms that changed the environment of the Canadian natural gas industry (Wilson, 1997; Serletis and Rangel-Ruiz, 2004). These reforms put an end to natural gas price regulation and changed integrated monopolies into separate marketing, transmission, and distribution service companies (unbundling of services) (North America Energy Working Group, 2002). Deregulation in the natural gas industry has been a long continually evolving process. One goal of deregulation is to increase market efficiency through enhancing competition. An outcome has been the development of market centers and hubs, which serve as natural gas spot markets. These market centers and hubs (henceforth markets) are located at the intersection of major pipeline systems and within major producing regions (US Department of Energy, 2004). If deregulation is achieving its goal, the North American markets should be highly integrated. Market integration is one important indicator of market competitiveness and allocative efficiency. Evidence on market integration, however, is conflicting. Cuddington and Wang (2006) and King and Cuc (1996), for example, discuss an east–west split in North American gas markets, whereas Serletis (1997) did not find such a split. The objectives of the current study are to characterize the dynamic integration among major natural gas spot markets in North America and to investigate each individual market's role in price discovery using data removed from the major deregulation policies. The current study differs from the growing literature on natural gas markets in several ways. Most important, this study is the first study to describe the structure of dynamic interactions among North America natural gas spot markets by combining recent advances in modeling causal flows with time series analysis. A vector error correction model provides the basis for empirical results on the interdependencies among the natural gas markets. In addition to the usual innovation accounting analysis, contemporaneous causal flows based on directed acyclic graphs are presented. The study provides a dynamic picture of daily information flow among eight North American natural gas spot markets for the recent past (1998–2007). The eight markets were chosen to provide geographical diversity, while accounting for data availability. Previous studies have not considered such a widespread geographical dispersion simultaneously among many markets. The current study does not analyze the spot markets using bivariate approaches, but rather all eight markets are included in the analysis in a more general multivariate framework. Another important difference between some of the previous literature and the current study is that the current study uses daily data instead of weekly or monthly data used by many previous studies. Taylor (2001) concludes that temporal aggregation problems arise when using monthly data to attempt to identify rapid price adjustments. Data used for analysis need to reflect institutional context and logistical framework of the markets being examined. Further, Brinkmann and Rabinovitch (1995) note that a problem with daily natural gas prices in the years immediately following the development of the spot markets was the lack of price fluctuations because of infrequent trading. Of particular relevance is that the data used in the current study are removed from the early years of the spot market development. More recent data reflect the continued reactions to evolving regulations and market conditions by participants, thereby giving a clearer picture if the markets are integrated. New regulations, such as FERC's Order 637, which is an attempt to update gas pipeline operations and increase the level of transparency (US Department of Energy, 2006), continue to define the gas market. Further, changes in total US pipeline capacity are reflected in the more recent data. Interstate pipeline capacity increased by 17% in the 7-year period 1991–1997 (6% between 1994 and 1997), but increased 19% in the 5-year period 1998–2004 included in our price data. The year with the largest increase is 2002 (Tobin, 2006). In addition, Moss (2007) reports that the Herfindahl–Hirschman Index for the US natural gas pipeline industry increased by 10% between 1992 and 1997, but increased by 33% between 1997 and 2001 to over 1800, indicating a highly concentrated industry. The more recent data also reflect changes in the Canadian market such as transportation reform in the early 1990s (Booth, 2003) and increases in exports to the US (Vollman, 1996). Finally, daily aggregate weather effects are included as exogenous variables in the current study, which are absent in previous time series analysis of natural gas markets.

The eight natural gas spot market prices are tied together with five long-run co-integration relationships. Exclusion and exogeneity tests show that all eight markets are present in and adjust toward the long-run co-integration relationships. These results provide evidence that the eight markets in North America are integrated. There are seasonal differences in the long-run relationships because the exogenous variables, cooling and heating degree-days (CDD and HDD), are not excluded from the co-integrating vectors. Besides industrial use, natural gas is used for heating and electricity generation in the winter and primarily for electricity generation in the summer, and therefore seasonality is plausible. Contemporaneous causal flows, forecast error variance decompositions, and impulse response functions all do not indicate a strong east–west split among North America natural gas spot markets as suggested by the discussions in King and Cuc (1996) and Cuddington and Wang (2006). Further as noted above, the markets are tied together by five long-run co-integration relationships, implying that new price information in one market is transmitted to other markets through arbitrage activities in the long run. These results further suggest that there is no east–west split. Serletis (1997) also argues that an east–west split does not exist. The results suggest a higher level of integration between the markets in more recent data. One plausible explanation for the differences in the studies is the time frame of analyses. King and Cuc (1996) and Cuddington and Wang (2006) include data from the period soon after the start of the major deregulation of the natural gas industry. Serletis (1997) and our analysis uses data farther removed from the start of deregulation. Park (2005) using the same methodology as used here on the same natural gas markets, but data from 1998 to 2002, found fewer contemporaneous price flows than found in the current study using data from 1998 to 2007. Taking these studies’ results into account may indicate that the natural gas market has developed into a single integrated market in North America since deregulation. Deregulation policies appear to have worked. A single integrated natural gas market indicates natural gas flows to the highest valued user. This inference is in line with some previous studies which indicate that integration has increased over time (DeVany and Walls, 1993; Doane and Spulber, 1994). Another reason for differences in the results is the methodology used. Here, all eight markets are included simultaneously in the analysis instead of using bivariate analysis. Contemporaneous time causal flows reflect the major natural gas transportation corridors in North America (Fig. 5). But, the relationship between causal flows, and therefore price information, and directional flow of natural gas is mixed. To clarify, consider the contemporaneous price information flows from Chicago to AECO; here information flows from an excess demand market toward an excess supply market. In other cases, for example, Henry Hub to Ellisburg-Leidy, the causal or information flow matches up to the direction of gas transportation, excess supply to demand regions. Inference from these observations is the economically pleasing but not highly surprising result that both supply and demand are important in price discovery. Although the general inferences from the results do not depend on the causal relationship assumed, the importance of the Chicago market relative to the other four major players, AECO, Malin, Waha, and Henry Hub, is affected by the assumed relationship. A plausible reason to assume the Chicago to Malin Chain relationship is as follows. The US gas market is divided into six major market areas: Central, Midwest, Northeast, Southeast, Southwest, and Western (US Department of Energy, 1998). The Midwest area, which includes Chicago, has the lowest ratio of natural gas production to consumption among the six market areas. This ratio implies that the Midwest is dependent on the import of gas from the other areas to meet its demand (US Department of Energy, 1998). The Midwest importing the most natural gas is a possible explanation for why Chicago is the most important market. Still another possible reason for the importance of the Chicago market is as follows. Natural gas supplies from southern Oklahoma and western Texas tend to be the marginal supplier for both the eastern and western markets. Natural gas from Oklahoma and Texas can easily be transported east or west. Gas from Texas and Oklahoma is sold to the market with the highest price (Serletis, 1997). Chicago being an excess demand market may be bidding up the price of natural gas in Oklahoma and Texas, making Chicago an important market for price discovery in North America. One possible explanation for the difference in responses by the other markets to shocks in Ellisburg-Leidy price is that the Ellisburg-Leidy market area, along with the Midwest region, has relatively larger underground storage capacities than the other regions (US Department of Energy, 2002). Gas withdrawals from storage facilities can help mitigate the price shocks in the other markets for short periods. This statement is supported by the observation that natural gas storage withdrawals account for a significant proportion of the supply necessary to meet total demand during the heating season, particularly in the East Consuming Region (Herbert et al., 1997). Opal's smaller influence may be attributed to the relatively smaller gas pipeline capacity leaving this region over most of the study period. Henry Hub has received considerable attention in the literature. It is the most active and publicized market center in North America and has an extensive receipt and delivery capability (US Department of Energy, 2003). More than 180 customers regularly conduct business at Henry Hub through 14 pipeline systems and storage facilities. Henry Hub is also the delivery point for NYMEX futures (US Department of Energy, 2003). Further, Henry Hub accounts for a large portion of gas transportation toward the East Consuming Region. Results suggest that Henry Hub plays an important role in natural gas spot market price discovery. There are additional issues that are not addressed in this study but may provide insights into the pricing of natural gas in North America. Inclusion of Mexican market(s) and a southern California market would allow for a more complete picture of the North American gas market to emerge. Data limitations did not allow for the inclusion of these markets in this study. The only exogenous factor considered in this study was temperature. Factors such as variations in demand, storage capacity, future markets, and types of end use of natural gas should be considered in future studies. The potential for structural change in the price data exists as markets develop, the use of natural gas shifts toward electricity, and additional pipeline capacity is added to the system. Rudimentary analysis of the data indicates a potential structural shift occurring around 2003–2004. This coincides with an increase in movement of natural gas to Mexico. Additional analysis on structural change is warranted. How is the price affected by storage capacity and amount of natural gas in storage? Another important consideration is the role that the future market plays in price discovery. The future market is active in the natural gas industry. Considering that electricity power plants accounted for about 27% of the US natural gas consumption in 2002 (US Department of Energy, 2004), spot markets from both electricity and natural gas industries should be included in future studies to examine the interdependencies of these two energy markets. Increased attention to pipeline capacity may provide further insights to differences in market integration over time. Finally, contract arrangements and their influence on spot market transactions and prices may be a fruitful avenue of future research.