Вы находитесь на странице: 1из 10

Energy Economics 33 (2011) 217226

Contents lists available at ScienceDirect

Energy Economics
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e n e c o

Natural gas prices, LNG transport costs, and the dynamics of LNG imports
Don Maxwell a,1,2, Zhen Zhu a,b,
a b

Department of Economics, University of Central Oklahoma, Edmond, OK 73034, United States C.H. Guernsey and Company, 5555 N. Grand Blvd, Oklahoma City, United States

a r t i c l e

i n f o

a b s t r a c t
According to the U.S. Energy Information Administration, LNG is projected to become a much larger share of U.S. natural gas consumption, rising from current levels of around 2.5% of total natural gas consumption to 12.4% by 2030. Because natural gas and LNG are substitutes, natural gas prices are expected to be an important determinant of LNG imports. Furthermore, an increasing share of LNG is traded under short-term contracts with spot shipments being diverted to markets offering the highest returns (netbacks). Relative natural gas prices as well as LNG transportation costs are important determinants of LNG netbacks. This paper examines the empirical relationship between U.S. LNG imports, the Henry Hub price of natural gas relative to U.K. and Asia gas prices, and a proxy for LNG transportation costs using monthly data from 1997 to 2007. Granger causality tests, error variance decomposition, and impulse response analyses using a VAR model are employed to establish Granger-causality as well as the dynamics of natural gas prices and LNG transportation innovations on LNG imports. 2010 Elsevier B.V. All rights reserved.

Article history: Received 17 November 2008 Received in revised form 18 June 2010 Accepted 25 June 2010 Available online 3 July 2010 JEL classication: Q41 Keywords: LNG Demand analysis Natural gas price

1. Introduction Prior to the late 1990s, imported LNG was an insignicant component of the U.S. natural gas mix. As recently as 1998, net imports of LNG accounted for less than one-tenth of one percent of total U.S. domestic consumption (Energy Information Administration, 2007a). The U.S. had sufcient domestic production as well as supplemental pipeline imports from Canada necessary to meet demand. Energy consumption in general has risen, with natural gas representing an increasing share of that consumption. Technological innovations in electric generation, together with a desire to reduce carbon emissions, have increased the demand for natural gas. With diminishing yields from U.S. and Canadian gas elds, the gap between domestic U.S. gas consumption and production has widened, causing the U.S. to increasingly rely on LNG imports to augment domestic natural gas supplies and pipeline imports. According to the Energy Information Administration (2008a, b, c) LNG net imports accounted for about 1.3% of U.S. domestic natural gas consumption with that percentage projected to rise to 3.3% by 2035 (EIA, 2009). Because of the growing demand gap in the U.S. and growing natural gas demand around the world, there has been a worldwide wave of investment in LNG production facilities, tankers, import
Corresponding author. Professor of Economics, University of Central Oklahoma and a consulting economist at C.H. Guernsey and Company in Oklahoma City, United States. Tel.: +1 405 974 5287. E-mail addresses: dmaxwell@uco.edu (D. Maxwell), zzhu@uco.edu (Z. Zhu). 1 Tel.: + 1 405 974 5287. 2 Don Maxwell is an Emeritus Professor of Economics at University of Central Oklahoma. 0140-9883/$ see front matter 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.eneco.2010.06.012

terminals, and gasication facilities. In the U.S. alone, as of 2006, there were 10 proposed LNG gasication terminals that had received FERC approval and many more awaiting approval or in planning (Natural Gas Intelligence, 2008). Investments such as these have become attractive because of high natural gas prices and falling LNG production costs, especially liquefaction and shipping costs. Industry trends away from long-term to short-term contracts and the increasing number of spot LNG cargoes have made LNG shipments more responsive to natural gas prices. Higher gas prices provide producers incentives to increase the quantity of LNG supplied. Those higher prices together with falling production costs fostered an expansion of the LNG industry. In short, LNG has become a competitive substitute for both domestically produced natural gas and Canadian pipeline imports for the United States. This paper examines the empirical relationship between U.S. LNG imports and U.S. natural gas consumption as well as the impacts of two other important determinants of LNG supply: U.S. natural gas spot prices relative to European and Asian gas prices and LNG shipping costs. Although there are numerous articles and reports describing the LNG industry (Ofce of Fossil Energy, 2005; EIA, 2007a; Gaul and Platt, 2007; Center for Energy Economics, 2007, Victor and Hayes 2006), there are few academic articles focusing on the LNG market and its trends.3 And there appear to be no studies that empirically model the magnitude, signicance, and the timing of interactions of these variables with an exception of Hayes (2006).

3 See, for example, Brito and Hartley (2007), Jensen (2003), Karabetsou and Tzannatos (2003), Kim (1996).

218

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226

Because some LNG industry data are proprietary, especially for LNG value train costs and netbacks (an industry term for producer net revenues), and unavailable in the public domain as time series, the scope of this paper is primarily limited to time-series data that are available from the U.S. Energy Information Administration. After a brief review of the LNG market, focusing on demand and supply determinants of U.S. LNG imports, this paper describes the data, methodology, empirical results and presents conclusions. 2. Demand for imported LNG U.S. demand for LNG is driven by long-term and seasonal trends in domestic natural gas consumption and production. Economic and population growth increase residential and commercial demand for heating, cooling, and electric power as well as the demand for industrial goods produced with or that use natural gas as feedstock supplies. Increasing home size, an increasing desire to reduce carbon emissions, and efforts to use compressed natural gas as an alternative fuel source in transportation have further increased the demand for natural gas. Electric power generation and industrial demand for natural gas account for 58% of total gas consumption (EIA, 2008a). Because of technological innovations in electric generation, the higher btu content of natural gas, lower plant construction costs, and a reduced carbon footprint, natural gas has become an attractive fuel source for the electric power industry (Ofce of Fossil Energy, 2005). As a consequence, the share of total natural gas consumption by the electric power industry increased from 18.4% in 1994 to 29% in 2007 (EIA, 2008a). While coal is still the most important fuel source, accounting for 50% of electric power generation, natural gas red electricity has increased its share of total electricity production from 12.5% in 1994 to 20.2% in 2007 (EIA, 2008b). In the short-run, natural gas consumption is seasonal, rising during cold weather months as the number of heating days increase and during the summer for gas-powered electric production used for cooling. Because consumption is seasonal, but production is not, utilities and electric power companies often inventory natural gas during off-peak periods in order to meet seasonal demand. Even though the dominant source of the inventory is the underground gas storage, LNG is used to meet the demand in the seasonal/short term swing.

U.S. dry gas production has not kept pace with consumption, creating a need to import natural gas. Although the U.S. both exports and imports natural gas, net imports of natural gas, pipeline and LNG, in 2008 made up about 12.7% of total gas consumption. The EIA forecasts that this percentage will fall to 10.9% by 2015 and to 5.9% by 2035 (EIA, 2009). About 85% of natural gas imports come from Canada. Although the number of Canadian drilled wells has increased yearafter-year, Canadian annual gas production has leveled off. From 1995 to 2005, the annual number of drilled Canadian gas wells increased by 379%, yet dry gas production peaked around 2002 (EIA, 2007a). As shown in Fig. 1, the gap between U.S. domestic gas consumption and dry gas production has increased from 6.75 BCF per day in 1994 to 10.4 BCF in 2007 (EIA, 2008a). Canadian gas exports lled this gap until around 1999 (EIA, 2007b) when the U.S. began to rely more heavily on imported LNG. The EIA predicts that LNG will make up an increasing share of U.S. natural gas net imports, reaching 57% of natural gas imports by 2035, as well as a larger share of total natural gas supplies. This study uses U.S. natural gas consumption as a general demand proxy for imported LNG. Of course the demand for LNG as an alternative source of natural gas depends not only on the forces driving natural gas demand, but also depends on domestic production as well as pipeline imports. Furthermore, it depends on the price of natural gas in major consumption countries in Europe and Asia (Hayes, 2007a, b). As will be discussed below, most LNG contracts in the U.S. are short-term contracts. And most contracted LNG shipments use the Henry Hub Spot price of natural gas as a benchmark. Ultimately, end users of natural gas pay the same price regardless of whether that gas is domestically produced or imported. Buyers, however, will naturally seek out the most competitively priced source of gas. Importers of LNG, primarily as few large oil companies such as Shell, BP, Statoil, BG Group, Suez, and Excelerate Energy (EIA, 2007a), contract for LNG shipments if those shipments can be received at U.S. terminals at prices that are competitive with Henry Hub prices. While average LNG import prices move closely with Henry Hub prices, at times the Henry Hub price rises above or falls below LNG prices. The volume of LNG imports increases as Henry Hub prices rise relative to LNG prices. As shown in Fig. 2, for most years following 1999, Henry Hub spot prices exceeded LNG prices, likely contributing to an increase in demand for imported LNG.

Fig. 1. U.S. LNG imports.

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226

219

Fig. 2. U.S. LNG import prices.

So in addition to a demand proxy, U.S. natural gas consumption, this study also examines the empirical relationship between the Henry Hub spot price and the volume of LNG imports. Both U.S. natural gas consumption and the Henry Hub spot price are expected to be positively related to LNG imports. 3. LNG supply and pricing The U.S. has been involved in LNG trade since the 1969, exporting Alaskan LNG to Japan. But LNG imports did not become important in the U.S. until the late 1990s. Although LNG receiving terminals were built in the U.S. during the 1970s, rising energy prices spurred domestic exploration, production, and building gas reserves during the 1980s, resulting in two of the facilities being shut down (Ofce of Fossil Energy, 2005). It wasn't until domestic production and pipeline imports of natural gas began to level off and natural gas prices began to rise that there was a renewed interest in LNG. Importing LNG became more lucrative as Henry Hub spot prices began to rise in the late 1990s. LNG producers around the world, 14 countries as of 2007 (Gaul and Platt, 2007), are in areas of the world with stranded natural gas reserves, but limited end markets. For instance, 58.4% of the world's gas reserves are found in Qatar, Russia, and Iran, but they account for only 19.4% of world natural gas consumption (Ofce of Fossil Energy, 2005). In 2006, the U.S. imported LNG from Trinidad and Tobago, Egypt, Algeria, and Nigeria with Trinidad and Tobago accounting for 67% of total LNG imports (EIA, 2008c). Liquefaction of abundant gas reserves is a way for these countries to exploit their resources. The entire process from exploration and production of feedstock supplies to destination terminal regasication and delivery to the pipeline system is known as the LNG value chain. It consist of four components, 1) exploration for and production of natural gas, 2) liquefaction, 3) shipping, and 4) storage and regasication. For LNG producers to have incentives to export to the U.S., they must receive returns sufcient to cover costs, including a desired return to equity. For a typical LNG value chain, exploration and production of feedstock supplies represent 15 to 20% of total capital costs, liquefaction comprises 30-45% of costs, shipping accounts for another 10 to 30%, and gasication and storage account for the remainder, 15 to 25% (Ofce of Fossil Energy, 2005). LNG contract prices are determined based on whether LNG is priced Free-on-board (F.O.B.) or Ex-ship. Traditionally, LNG has been priced Ex-ship, where contract prices reect downstream prices

(Henry Hub for the U.S.) less gasication and other destination terminal costs and shipping, including insurance (Holmes, 2007). The remainder (netback to producer) must be sufcient to cover all costs associated with acquiring feedstock supplies, liquefaction and export terminal costs plus yield a sufcient return to equity. F.O.B. LNG contract prices are prices of LNG delivered to the tanker at the export terminal. Here shipping and insurance are the responsibility of the buyer. Contracts are increasingly of the F.O.B. type. Such contracts give buyers greater exibility regarding shipping costs and the ability to exploit prot opportunities through arbitrage. But there may be several destination markets competing for the same spot shipments of LNG. Spot shipments tend to go where netbacks are the highest. For instance, rising U.S. natural gas spot prices during the rst half of the 2000s increased netbacks and the number of shipments arriving at U.S. terminals, with the number of shipments peaking in 2004. But in 2006, other Atlantic Basin countries such as Spain, France, and the United Kingdom outbid the U.S. for LNG (EIA, 2007a; Gaul and Platt, 2007, Hayes, 2007a, b). According the to EIA, netbacks to Trinidad and Tobago, the main U.S. supplier, from destinations like Spain and the U.K. ($9.17 and $5.32 MMbtu, respectively) were much higher in 2006 than the $3.71 netbacks from terminals in the U.S (EIA, 2007c). But in 2007, U.S. LNG imports increased to new highs as netbacks from U.S. terminals to Trinidad and Tobago increased relative to those from terminals in Belgium and Spain (ScottMadden Inc., 2007). Downstream prices, such as Henry Hub prices, are positively correlated with producer netbacks. Others things equal, an increase in the Henry Hub spot price would be expected to increase netbacks and therefore the number of spot cargoes sent to the U.S. But other components of the LNG value chain also impact netbacks directly or increase returns to equity. For instance, various technological innovations in liquefaction caused a signicant decline in liquefaction capital costs from $600 per ton of capacity in the late 1980s to $200 by the early 2000s. Shipping costs also fell dramatically. Competition between shipyards reduced the construction cost of LNG tankers from about $280 million for a 138,000 cu. meter ship in 1995 to $150 to $160 million by the mid-2000s (Ofce of Fossil Energy, 2005). Larger tankers enjoy economies of scale. Advances in propulsion systems reduced operating cost. Decreases in liquefaction costs increase producer returns. Decreases in shipping costs increase netbacks and returns to equity for producers that deliver LNG on an Ex-ship basis. For F.O.B. contracts, lower shipping costs increase prot margins for buyers.

220

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226

Higher gas prices and falling costs have spurred a wave of investment in LNG facilities worldwide in such areas as Russia, Yemen, Equatorial Guinea, and Norway (Research Reports International, 2006). As of March 2007, liquefaction capacity serving the Pacic and Atlantic basins was 170 million tons per year with 91 million tons of capacity under construction and 285 million tons of capacity in the planning stages. Shipping has experienced similar growth. As of March 2007, the LNG tanker eet was comprised of 224 tankers. 145 more were awaiting construction, up considerably from the 45 tankers ordered the previous year (Center for Energy Economics, 2007). A complete LNG value chain takes years to complete and $ 9 to $10 billion to build (Ofce of Fossil Energy, 2005). So given the volatility of energy markets, investors face considerable risks. Traditionally, this risk was mitigated with long-term contracts of 20 years or more. Producers built dedicated facilities to produce LNG for specic customers and transported that LNG in tankers purchased specically for that contract. Buyers guaranteed minimum purchases. Because of these long-term agreements, lenders were willing to nance dedicated LNG projects. But today the trend is toward shorter and less restrictive contracts, such as F.O.B. contracts. Today, 12% of worldwide contracts are short term, but that number is expected to rise to 15 to 20% (Ofce of Fossil Energy, 2005). And there is reason to believe that this shift may occur rapidly (Brito and Hartley, 2007). Short-term

contracts are even more prevalent in the U.S., representing 80% of U.S. trade in LNG in 2003 and 70% in 2004 (Ofce of Fossil Energy, 2005). In addition to U.S. natural gas consumption and Henry Hub spot prices, this paper also focuses on LNG value chain costs and their impact on LNG imports. In particular, we investigate the relationship between ship construction costs and LNG imports. Time-series data on other components of the value chain are generally unavailable, at least in the public domain. So this paper considers the capital cost component of shipping costs. LNG shipping rates are usually expressed on a charter-day basis, similar to average cost pricing. The charter rate is designed to cover operating expenses during the voyage, make a contribution to capital costs (essentially a daily equivalent capital cost and nance expense assuming a useful life of 25 years or so), as well as provide a return to the ship owner. Capital costs, including nance costs, are the largest component of daily charter rates, representing $45 to $55 thousand per day of daily charter rates that average $60 to $100 thousand per day (Ndao, 2004). Overall, shipping costs are sensitive not only to daily charter rates, but also to the number of days in transport the number of nautical miles from point of departure to point of destination. At a given point in time, much of the variation in landed LNG prices to various U.S. import terminals can be attributed largely to differences in shipping distances.

Fig. 3. LNG imports, gas prices, consumption, and tanker cost.

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226

221

U.S. Gas Consumption


U.S Gas Comsumption Demand (Bcf/Day)
100 90 80 70 60 50 40 30 20 10
Jan-1997 Jul-1997 Jan-1998 Jul-1998 Jan-1999 Jul-1999 Jan-2000 Jul-2000 Jan-2001 Jul-2001 Jan-2002 Jul-2002 Jan-2003 Jul-2003 Jan-2004 Jul-2004 Jan-2005 Jul-2005 Jan-2006 Jul-2006 Jan-2007 Jul-2007

300

Tanker Cost

250

Tanker Cost

200

150

100

50

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

Source: Energy Information Administration.


Fig. 3 (continmued).

Other things equal, decreases in tanker prices reduce capital and nance costs, reducing daily charter rates and the cost of shipping LNG. Consequently, falling long-run marginal costs give LNG producers incentives to expand existing facilities as well as invest in new LNG projects. Over time then, falling tanker construction costs increase the supply of LNG. Lower shipping costs enable sellers to competitively price LNG and thus compete with domestic production and pipeline imports. U.S. LNG imports, therefore, are expected to be inversely related to LNG tanker prices. In sum, increases in U.S. natural gas consumption, increases in Henry Hub spot prices relative to other regions of the world, and decreases in tanker prices have likely played a role in rising U.S. LNG imports. Next, this study empirically examines the relationship between these supply and demand factors and U.S. LNG imports, beginning with a description of the data and methodology and followed with the results and summary. 4. Data As noted above, this study analyzes the association between U.S. LNG imports and U.S. natural gas consumption, U.S. natural gas prices relative to U.K. and Asian prices, and tanker construction costs, a

proxy for LNG transportation costs. This paper also investigates the dynamic response of LNG imports to innovations in these variables. Most data are from the U.S. Energy Information Administration.4 As shown in Fig. 3, from 1997 to 2007, U.S. LNG imports increased from initial levels of less than 5000 MMcf to about 100,000 MMcf, a 20-fold increase. Fig. 4 shows that the majority of U.S. imports were from three countries: Algeria, Nigeria, and Trinidad and Tobago with Trinidad being the primary supplier in the 2000s Fig. 3 also shows U.S. natural gas prices, represented here by the Henry Hub spot price. In general, natural gas prices have exhibited an upward trend with prices around $2 MMBtu in the late 1990s, rising gradually to $7 to $8 by 2007. Also evident are several large price spikes. One occurred during the 2000/2001 heating season while another occurred during the 2002/2003 heating season. These periods

4 U.S. natural gas consumption, LNG imports, and Henry Hub spot price data are obtained from tables at http://www.eia.doe.gov/oil_gas/natural_gas/info_glance/natural_gas.html. Brent oil price data is obtained from EIA. Tanker cost data are based on a report LNG Industry Cost Declining at http://www.eia.doe.gov/oiaf/analysispaper/global/lngindustry. html. U.K. natural gas price data is from ICIS Heren and U.S./U.K exchange rate is from Federal Research Bank of Saint Louis FRED database.

222

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226

Fig. 4. US LNG imports from Algeria, Nigeria and Trinidad percentages.

were characterized by abnormally cold winters. Another price spike occurred during the fall of 2005 when hurricanes Katrina and Rita damaged gulf coast production. Gas consumption, in contrast, showed no apparent trend other than typical seasonal variations in demand. As noted above, seasonal natural gas demand is driven mainly by residential and commercial winter heating and summer cooling needs while industrial and transportation demands are more stable. Fig. 3 also shows a substantial decline in tanker construction costs. This is the estimated average price of a 138,000 cubic-meter ship,

having capacity to carry 2.9 Bcf of natural gas. As discussed previously, LNG value train costs consist of production, liquefaction, shipping, and storage and regasication costs. As noted above, value train costs are generally unavailable as time series. This study uses one component of these costs, tanker costs, as a proxy for value train costs. Because only annual data are available and only up to 2003, annual LNG tanker construction cost data are extrapolated into a monthly series and extended to 2007 using an exponential smoothing technique. Because there is no monthly variation within a given year's tanker cost data, some of the richness associated with typical monthly data is lost. And

Fig. 5. U.S./U.K and U.S./Asian gas price ratios.

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226

223

it is possible that the extrapolated values for 20042007 differ from actual values. Fig. 5 depicts the natural gas price in the U.S relative to those in U.K. (representing Western European countries) and Asian countries. The U.K. price is represented by the monthly Heren NPB price index which is similar to the U.S. Henry Hub price. The NPB price has been adjusted using the dollar-pound exchange rate obtained from the Federal Research Bank of Saint Louis FRED database. The Asian gas price is proxied by the Brent oil price since the Asian gas price is closely linked to oil prices. In general, U.S. gas prices have been higher than U.K. prices (about 34% higher) while Asian prices are generally lower (about 17% lower) during the sample period. All variables are expressed in natural logarithms except the natural gas price ratios. In addition, gas consumption is ltered using monthly dummies. The residual series derived from ltering is used as the gas consumption variable in the unit root test. LNG import data is also examined for seasonality by regressing LNG imports on monthly dummies. No seasonality was found. 5. Methodology In order to determine the appropriate impact analysis method, a unit root test is conducted. The unit root test used here is the conventional Dickey-Fuller test. Lags in the regression are determined using the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) criteria. As discussed below, the empirical results suggest that all of the monthly data are stationary, allowing the use of the Granger causality test. The Granger causality test (Granger, 1969) is designed to detect the causal direction between two time series. More precisely, the Granger causality test detects a correlation between the current value of one variable and past values of another variable. Consider a bivariate VAR model with two time series Yt and Xt. Yt = 12 + 11i Yti + 12j Xtj + 12t ;
i=1 j=1 T21 T22 T11 T12

Table 1 Unit root test. Variable LNG imports Gas consumption US./U.K. price ratio U.S./Asian price ratio Tanker cost
a

Laga 2 1 1 0 12

t-valueb 3.60 5.786 4.63 3.97 4.33

The lag length in the Dicky-Fuller unit root test is determined by the Akaike Information criterion and Bayesian Information Criterion. b t-values are obtained with a specication with a constant in the equation. c Gas consumption has been ltered rst by running the regression of the variable on monthly dummies. All variables are in natural logs except for the price ratios. Critical values for the DF unit root test are 4.033, 3.466, and 3.147 at the 1%, 5% and 10% levels respectively.

for t = and 0 otherwise, and is a nxn symmetric positive denite matrix. The VAR(p) model can be written as a vector moving average: Yt = + t + 1 t1 + 2 t2 + 3 t3 + = + Lt ; 5 where L is a lag operator. Thus, the matrix s has the interpretation Yt + s = s : 0 t 6

The row i, column j element of s identies the consequences of a one-unit increase in the jth variable's innovation at date t for the value of the ith variable at time t + s, holding all other innovations at all dates constant. A plot of this element as a function of s is the impulse response function. Also calculated is the forecast error variance decomposition (Hamilton, 1994) from the Mean Square Error (MSE) of the s-periodahead forecast of the Y variables:
  MSE Y t + s j t n  h io n = j = 1 Var ujt aj aj + 1 aj aj 1 + 2 aj aj 2 + + s1 aj aj s1 ;

Xt = 22 + 21i Xti + 22j Ytj + 22t ;


i=1 j=1

7 where u is an element of vector ut and aj is the jth column of the matrix A as in the following relationships: t = Aut = a1u1t + a2u2t + + anunt, and A is a unique lower triangular matrix with 1's along the principal diagonal. A is derived from the real symmetric positive denite variancecovariance matrix () of the VAR model such that = ADA, where D is a unique diagonal matrix with positive entries along the principal diagonal. The resulting variance decomposition shows the percentage of forecast errors explained by shocks to other variables. This study employs a non-structural VAR model consisting of ve variables, Y = [GasConsumption, US/UK Price, US/Asia Price, TankerCost, LNGImports]. The lags used in the VAR are chosen using AIC and BIC criteria. In deriving impulse response functions and variance decomposition, the Cheloski decomposition is used, with the order of the variables as specied in the Y vector. However, the results are not sensitive to variable ordering.
Table 2 Granger causality test. Bi-variate causation LNG Imports Gas Consumption Gas Consumption LNG Imports LNG Imports U.S./U.K.Price U.S./U.K. Price LNG Imports LNG Imports U.S./Asian.Price U.S./Asian Price LNG Imports LNG Imports Tanker Cost Tanker Cost LNG Imports F value (P value) 0.996 0.893 1.90 1.80 1.065 2.767 3.009 2.321 (0.459) (0.556) (0.044) (0.059) (0.398) (0.065) (0.001) (0.024) Causality NO NO YES YES NO YES YES YES

where is the difference operator, T is the lag order, and are parameters for estimation, and t is an error term. To test whether the Granger causality runs from X to Y, the null (H0) hypothesis is: H0 : 12j = 0 j = 1; 2; q: 3

If H0 is rejected, at least one of 12j s is not equal to zero, suggesting that past values of X have signicant linear predicative power on current values of Y. It normally denotes that X Granger causes Y (in our notation, X =N Y). While a Granger causality test may be useful in detecting the direction of causality, it does not reveal specic dynamic relationships between relevant variables. Therefore, this study also examines the impacts of changes in particular variables on the timing, direction, and duration of LNG imports and considers whether those impacts are economically signicant. These questions are answered by impulse response analysis and variance decomposition using vector autoregression (VAR) models. A p-th order vector autoregression, VAR(p), can be written as Yt = c + 1 Yt1 + 2 Yt2 + + p Ytp + t ; 4

where Y is a nx1 vector, c is a nx1 vector of constants, and i is a nxn matrix of autoregressive coefcients for i = 1, 2,, p. The nx1 vector t is a vector generalization of white noise with E(t) = 0, E(t, ) =

224

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226

Fig. 6. Impulse response of LNG imports.

6. Empirical evidence Table 1 below presents the unit root test results. The tests use a time trend and a constant as regressors. In general, the results suggest that all variables are stationary at the 5% signicance level, allowing the use of Granger causality tests and VAR analysis. The Granger causality tests presented in Table 2 are based on regression equations using 12-period lags. The overall conclusions of this study, however, are not affected by lag length. Regarding LNG Imports and Gas Consumption, there is no apparent causality running in either direction. But LNG Imports and the US/UK price ratio are related, with the causality running both directions. This result is consistent with expectations. LNG imports represent the marginal gas supply and are expected to impact prices. On the other hand, LNG Imports can obviously be affected by relative natural gas prices. While LNG imports do not seem to Granger cause U.S to Asian price ratios signicantly, a higher U.S. to Asian gas price ratio seems to cause rising LNG imports. In this sense, higher U.S. natural gas prices attract LNG imports as well as divert spot shipments of LNG to the U.S. For LNG Imports and Tanker Cost, the causality runs both directions. For the case of Tanker Cost Granger causing LNG Imports, as argued previously, a reduction in tanker costs increases LNG imports as LNG shipping costs fall, producer netbacks rise, and F.O.B. buyer prot margins widen. Regarding LNG Imports Granger causing Tanker Cost, a plausible explanation is that as importers and producers of LNG seek to position LNG as a competitive substitute for domestic natural gas production and pipeline imports, they have incentives to reduce value chain costs in general, including shipping costs. And competition among shipyards decreased tanker prices as the volume of LNG trade and new tanker orders increased.

Impulse response and variance decomposition analyses extend and provide support for the Granger causality results. Fig. 6 shows how LNG Imports respond over a 24 month period to shocks in specic variables. Several interesting patterns emerge. First, all variables have statistically signicant impacts on LNG Imports. The dotted lines indicate upper and lower 95% signicance bands. All shocks have the expected directional effects. After a 5-month lag, Gas Consumption has a positive, but small impact on LNG Imports. A higher U.S. to U.K. gas price ratio diverts shipments to the U.S. with a 12 month lag, while a higher U.S. to Asian price ratio has a longer lag at about 45 months. Furthermore, as expected, Tanker Cost is negatively related to LNG Imports, having signicant impacts throughout the lagged periods.

Table 3 Decomposition of variance for LNG import. Step 1 2 3 4 5 6 7 8 9 10 15 20 24 LNG imports 100.0% 87.8% 73.7% 71.2% 71.6% 69.1% 69.2% 67.6% 65.0% 62.8% 57.0% 55.5% 53.9% U.S. demand 0.0% 0.0% 0.0% 0.1% 0.2% 0.2% 0.2% 1.0% 1.1% 1.7% 10.1% 11.3% 11.5% US/UK price Ratio 0.0% 3.9% 10.5% 13.6% 14.1% 14.8% 14.5% 14.8% 15.5% 16.4% 14.8% 15.5% 16.0% US/Asia price Ratio 0.0% 0.9% 0.9% 1.1% 1.0% 1.4% 2.1% 2.9% 4.9% 6.1% 6.4% 6.2% 6.4% Tanker cost 0.0% 7.4% 14.8% 14.0% 13.2% 14.4% 14.0% 13.8% 13.5% 13.0% 11.7% 11.5% 12.1%

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226

225

Fig. 7. Variance decomposition of LNG imports.

Impulse responses also reveal timing effects. The impact of gas consumption seems to be greatest at lags of 5, 6, and 7 months, beyond which the impacts begin to diminish in size.5 Rising U.S. to U.K. price ratios, in contrast, have a more immediate impact on LNG Imports, with the maximum effect occurring at a lag of 2 months. The impact then tapers off, but is still signicant for several more months. Higher U.S. to Asian price ratios have a delayed impact, but the impact lasts much longer than for the other variables. Tanker costs have a persistent impact on LNG Imports lower costs leading to permanently higher LNG imports. Finally, the largest impact comes from a LNG import shock itself. While the impulse responses above suggest signicant dynamic effects, forecast variance decomposition provides evidence regarding the economic signicance of the impacts. Table 3 and Fig. 7 show the impacts on LNG Imports from selected shocks. After a two-month lag, shocks to LNG Imports explain about 90% of the forecast error variance. The impact then diminishes to about 50% at the end of the 24-month period. Shocks to US/UK gas price ratios quickly gain strength, and within a half-year period, the shocks explain about 15% of the forecast error variances. In contrast, the US/Asian gas price ratios do not seem to have as signicant an impact as the U.S./U.K. price ratios, possibly reecting the fact that the shipments to Asian countries are harder to divert. Tanker Cost shocks intensify quickly and reach a maximum of 15% within a few months. Finally, Gas Consumption has very little impact until about a year after the shock when the forecast error variance rises to about 11%. 7. Conclusions Despite sharp increases in U.S. LNG imports and the increasing importance of LNG in the U.S natural gas mix, empirical studies on the subject are rare. This study begins to ll the void by focusing on several key economic LNG demand and supply variables for which time-series data are available. Of the variables considered, U.S. to U.K. and Asian gas price ratios and shipping costs appear to be the most important determinants of U.S. LNG imports. Additionally, it appears that LNG imports respond
5 This effect could be due to either the lagged response to positive winter consumption deviations or the latent seasonal effect, especially if it is related to greater than average winter demand resulting in greater than average summer imports, since the U.S. tends to import more LNG in the summer on average. We would like to thank the anonymous referee for pointing this out.

more rapidly to shocks to gas prices than to shipping costs, the latter of which tends to impact LNG imports gradually. These patterns are consistent with industry trends: an increasing number of spot LNG cargoes, short-term contracts pegged to spot natural gas prices, and a worldwide wave of investment in LNG facilities and shipping. As natural gas red generation continues to replace coal and other forms of electric power generation, gas demand in the U.S. will continue to increase. Demand and supply imbalances will likely continue to cause high and volatile gas prices. But those high prices together with lower LNG value chain costs have made LNG a viable and attractive substitute for domestic and imported pipeline natural gas. As shipping technology improves and transport costs fall, LNG imports are likely to become more responsive to relative natural gas prices, a major signal directing LNG ows. And continued growth of U.S. LNG imports appear to depend upon rising U.S. gas price ratios. In the last several years, increased U.S. domestic production along with reduced domestic demand has caused U.S. gas prices to decline. Incidentally, EIA's projections of LNG imports have been reduced substantially.

Acknowledgment We would like to thank an anonymous referee for the insightful comments and suggestions that helped improve the paper. However, errors remain ours.

References
Brito, D., Hartley, P., 2007. Expectations and the evolving world gas market. The Energy Journal 28 (1), 124. Center for Energy Economics, University of Texas at Austin (2007). Introduction to LNG: An Overview of Liqueed Natural Gas, Its Properties, Organization of the LNG Industry, and Safety Considerations. Retrieved from http://www.beg.utexas.edu/energyecon/ lng/documents/CEE_Introduction_To_LNG_Final.pdf. Energy Information Administration, U.S. Department of Energy (2007a). U.S. Natural Gas Imports and Exports: Issues and Trends 2005 Retrieved from http://tonto.eia. doe.gov/FTProot/features/ngimmpex05.pdf. Energy Information Administration, U.S. Department of Energy (2007b). Natural Gas. In International Energy Outlook (Chapter 4). Retrieved from http:// www.eia.doe.gov/ oiaf/nat_gas.html. Energy Information Administration, U.S. Department of Energy (2007c). Natural Gas Year-in-Review 2006. Retrieved from http://tonto.eia.doe.gov/ftproot/features/ ngyir2006.pdf. Energy Information Administration, U.S. Department of Energy (2008a). U.S. Natural Gas Supply, Consumption, and Inventories. In STEO Table Browser. Retrieved from http://tonto.eia.doe.gov/cfapps/STEO_Query/steotables.cfm.

226

D. Maxwell, Z. Zhu / Energy Economics 33 (2011) 217226 Holmes C. (2007). World LNG Netback Series Begins in this Issue. Retrieved from http://www.ogj.com. Jensen, James T., 2003. The LNG revolution. Energy Journal 24 (2), 145. Karabetsou, C., Tzannatos, E.S., 2003. LNG changes in the context of the expanding market of natural gas. European Research Studies 6 (3-4), 6784. Kim, Suduk, 1996. LNG demand for power generation and optimal LNG inventories. Journal of Economic Research 1 (2), 381398. Natural Gas Intelligence Press, Inc. (2008). North American LNG Import Terminals: Status of Proposed and Existing Facilities. Retrieved from http://intelligencepress. com/features/lng. Ndao M. (2004). The Dynamics of the LNG Shipping Market and the Development of LNG Supply for Western Europe. Retrieved from http://www.sec.ou.edu/mee/ researchpapers07.php. Ofce of Fossil Energy, U.S. Department of Energy (2005). Liqueed Natural Gas: Understanding the Basic Facts. Retrieved from http://www.fossil.energy.gov/ programs/oilgas/publications/lng/LNG_primerupd.pdf. ScottMadden, Inc. (2007). ScottMadden's Energy Industry Update: Highlights of Recent Signicant Events and Emerging Trends in the Energy Industry. Retrieved from http://www.scottmadden.com/pdfs/Energy-Industry-Update-June-2007Executive-Summary.pdf. Victor, Jaffe, Hayes, Mark H., 2006. Natural Gas and Geopolitics: From 1970 to 2040. Cambridge University Press.

Energy Information Administration, U.S. Department of Energy (2008b). U.S. Electricity Generation by Fuel and Sector. In STEO Table Browser. Retrieved from http://tonto. eia.doe.gov/cfapps/STEO_Query/steotables.cfm. Energy Information Administration, U.S. Department of Energy (2008c). U.S. Natural Gas Imports by Country. In Natural Gas Navigator. Retrieved from http://tonto.eia. doe.gov/dnav/ng/ng_move_s1_a.htm. Energy Information Administration, U.S. Department of Energy (2009). Annual Energy Outlook 2010 (Early Release). Retrieved from http://www.eia.doe.gov/oief.aeo. Granger, C., 1969. Investigate causal relations by econometric models and cross spectral methods. Econometrica 37, 424438. Gaul D. & Platt K. (2007). Short-term Energy Outlook Supplement: U.S. LNG ImportsThe Next Wave. Retrieved from http://www.eia.doe.gov/emeu/steo/pub/pdf/ LNG_Jan2007.pdf. Hamilton, J., 1994. Times Series Analysis. Princeton University. Hayes, Mark H., 2006. Flexible LNG Supply, Storage and Price Formation in a Global Natural Gas Market. Stanford University. Dissertation. Hayes Mark H. (2007a), Monthly Gas Trade in the Atlantic Basin circa 2015, PESD Working Paper, Stanford University. http://pesd.stanford.edu/publications/list/ 0/0/1. Hayes Mark H. (2007b), Flexible LNG Supply and Gas Market Integration: A Simulation Approach for Valuing the Market Arbitrage Option, PESD Working Paper, Stanford. http://pesd.stanford.edu/publications/list/0/0/1.

Вам также может понравиться