Today is the official release event for the MIT study on the Future of Natural Gas. It’s a wide ranging study examining the role of natural gas in meeting future energy demand under carbon dioxide emissions constraints, and recommending appropriate policies–both for the US government and for industry. I was a member of the study group producing the study. I’ll let the full length study speak for itself on the many different issues it addresses. I want to use this blog post to expand on one specific point which is the huge uncertainties we face in charting any path forward as manifested in the fluctuating price of natural gas.
These uncertainties are apparent in the recent “decoupling” of natural gas and oil prices. Historically, actors in the industry have employed various “rules of thumb” to describe the equilibrium natural gas price as a function of the oil price. One is the “10-to-1” rule: the oil price (denominated in dollars per barrel) is expected to be 10 times the natural gas price (denominated in dollars per mmBtu). But since the end of 2008, there has been a remarkable decoupling between the two prices.
Starting in December 2008, the price of crude oil started to recover from its low of $31/bbl, and by September 2009 it had risen to nearly $70/bbl. In contrast, the price of natural gas continued to fall. From its already low level of $5.37/mmBtu in December 2008, it had dropped below $2/mmBtu in September 2009. At the conclusion of 2010, the price of oil reached above $90/bbl, while the price of natural gas had recovered to only $4/mmBtu. Since mid-2010, the ratio of the price of crude oil to the price of natural gas has been in the neighborhood of 20-to-1. This decoupling has been remarked on widely in press reports on energy topics–for example, recently in the Wall Street Journal and the Financial Times.
This isn’t the first time people have discussed the decoupling of natural gas and oil prices. Throughout the 1980s and early 1990s, the United States experienced a so-called “gas bubble”–an excess supply of deliverable gas—that kept natural gas prices low relative to the then prevailing price of crude oil. The situation reversed itself in the late 1990s and early 2000s so that the price of natural gas was regularly above the level one might have predicted based on the historical relationship and the then prevailing price of crude oil. Each time there has been discussion about decoupling and a new era. But somehow years later we find ourselves once again discussing the same old rules of thumb and the decoupling that now marks a new era.
My student, David Ramberg, has a paper (co-authored by me) on The Weak Tie Between Natural Gas and Oil Prices (as of the date of this post, the link is to the original version, with a revised version coming soon). We review a series of studies preceding the recent decoupling which document statistically how changes in the price of oil translate into changes in the price of natural gas — the statistical term of art is cointegration. The preceding studies can be found here, here and here. But if the prices are cointegrated, then what’s up with the recent and past episodes of decoupling? And what is really meant when analysts assert that the two prices have decoupled? “Decoupled” could mean any of the following:
- (i) the prices have temporarily broken away from the usual relationship to which they will later return, or,
- (ii) the prices have permanently broken away from the old relationship and moved into a new relationship, or,
- (iii) the two prices no longer maintain a relationship with one another at all.
Which is it?
The figure above shows the natural gas and oil price series. The two main conclusions from our study can be seen in this figure, although there’s much more statistical work behind them.
First, there is an enormous amount of unexplained volatility in natural gas prices. The raw price series for natural gas—without controlling for cointegration and any explanatory variables—is approximately twice as volatile as the raw oil price series. Hence, any simple formulaic relationship between the price of oil and the price of natural gas will leave a large portion of the short-run movements in the price of natural gas unexplained. The more statistically sophisticated approach of constructing a conditional Error Correction Model, which includes the cointegrating relationship, a set of exogenous explanatory variables, and accounts for the reversion of natural gas prices back to the cointegrating relationship, still leaves a large portion of the volatility in natural gas prices unaccounted for.
Second, while a cointegrating relationship can be documented, it is not stable through time. In the 1989 to 2005 period, the price of natural gas shifted up compared to the price of oil, but since 2006, this process reversed itself. Other researchers attribute the shift up in the former period to the increased demand for natural gas arising from the installation of advanced combined cycle gas turbine (CCGT) power plants with significantly improved heat rates. The shift down in the latter period is widely attributed to the increase in shale gas production due to the expanded use of horizontal drilling and hydraulic fracturing.
Although the two price series have been cointegrated, the confidence intervals for both short and long time horizons are large. In particular, the historical cointegrating relationship may not be a very reliable predictor of the future natural gas price, at least not at longer horizons over which shifts in the underlying technological and macroeconomic forces are unpredictable.
Discussions of the decoupling between natural gas and oil prices ought to be more precise about the time horizon involved. There are likely to be many occasions when the prices have temporarily broken away from the usual relationship to which they will later return. These decouplings can be severe, but they are also not very long lasting – less than one season typically – and the old relationship is reestablished. At the slightly longer horizon of a decade, prices can decouple from one relationship only to recouple in a new relationship. Shifts in the relationship between the two price series are probably driven by shifts in underlying technological and economic factors that are themselves difficult to predict. Finally, there is not yet any evidence that the relationship between the two price series has been severed completely. Indeed, it is hard to imagine that natural gas and oil prices could decouple completely and permanently. While pricing by energy equivalence is not likely to ever hold, the idea of an energy equivalence does express an important fact constraining the degree to which the two prices can become uncoupled. For example, while conversion of gas to liquids may seem expensive now, the technological possibility of conversion does place a cap on the degree to which oil prices can rise relative to natural gas prices.
Or, maybe not. Predicting the future is a dangerous business.