To access the full paper, download the PDF on the left-hand sidebar.
While much has been made in recent years about the increasing liquidity and size of a spot market for liquefied natural gas (LNG), with some analysts and governments even discussing creation of a centralized trading hub, most LNG is still sold under confidential, bilateral long-term contracts as has been the case since the 1960s. In fact, in 2013, according to data from the International Group of Liquefied Natural Gas Importers (GIIGNL), 73% of all LNG trades took place under long-term contracts (LTCs), which are especially prevalent in Asian markets. Despite the fact that this constitutes an enormous trade, there is very little transparency about how prices are specified, what actual transaction prices are or when pricing terms change.
Using publicly available customs data on sixteen different trade-routes of the largest importers of LNG, I apply sophisticated econometric techniques for detecting structural breaks of an unknown number and date to estimate and characterize the empirical relationship between LNG import prices and crude oil prices. These estimates allow me to make statements about the underlying pricing terms of LNG contracts, as well as when and how the pricing terms are revised. My results complement an existing literature on gas market integration, which considers cross-market convergence of prices without modeling the role of oil-indexed LTCs in determining LNG prices.
It is generally accepted that LTCs set LNG prices equal to an intercept term plus a slope times a crude oil benchmark. It is also known that some Japanese contracts have specified an “S-curve” that moderates the effect of very high or low oil prices on LNG prices. The exact parameters of the contracts, however, are not known. While it might be possible to estimate these using standard cointegration approaches, LTCs specify re-negotiation clauses, and we should expect that pricing terms may change over the course of a 20-year contract. In particular, pricing may be renegotiated when changes in market fundamentals cause the price of LNG inside a contract to diverge from the value outside the contract—the price a seller might receive from a spot sale less the transportation differential or the buyer, from an alternative supplier. Both changes in pricing terms and S-curve behavior that modifies the parameters of linear pricing terms outside of a mid-range are likely to induce a structural break in the empirical pricing relationship. Ignoring these changes not only gives biased and inconsistent estimates for contract parameters but also fails to deal with an important feature of contracts. By allowing for structural breaks of an unknown number and date, I avoid potential bias and can speak to the timing and types of revisions that occur in pricing terms.
My results strongly suggest that LNG is tightly indexed to oil, but terms are considerably more complex and varied than rules of thumb. Japanese contracts appear to have undergone the most revision. Initial breaks may be due to S-Curve behavior above thresholds at either $25 or $39 oil and occur in 2000 and 2004. Subsequently, tight LNG markets during the mid-2000s caused a mismatch between the price of LNG determined by the upper tail of an S-curve and its value as a substitute for oil, which steadily rose in price. Revisions brought LNG prices back to rough thermal parity with oil. It is interesting to note that pricing relationships do not undergo wholesale changes after Fukushima in 2011. This suggests that LTCs function as a form of insurance against shocks. Additionally, the fact that almost all contracts link current LNG prices to past oil prices (usually a weighted average of multiple months) may serve to smooth the magnitudes and timing of pricing shocks. LTCs in South Korea, Taiwan and Spain also appear oil-indexed but have far fewer revisions and likely no S-curve behavior. In particular, Korean and Taiwanese contracts generally set LNG prices to rough thermal parity with oil. Spanish slopes are much shallower, with LNG prices significantly below oil prices in recent years.
These results are the first rigorous characterization of global LNG pricing terms and revisions, and they complement an existing literature on international gas pricing that does not model the underlying data-generating process. This paper should be of interest to both economists and energy firms who are interested in how LNG prices are set in LTCs, under which the majority of the LNG trade is still priced. My results will also be of particular interest to firms entering into LTCs as they negotiate or re-negotiate pricing.
This material may be quoted or reproduced without prior permission, provided appropriate credit is given to the author and Rice University’s Baker Institute for Public Policy. The views expressed herein are those of the individual author(s), and do not necessarily represent the views of Rice University’s Baker Institute for Public Policy.
