For 50 years, we have tried to develop new fuels to replace in-ground reserves of crude oil and natural gas in the U.S. But now, the expectation of developing shale-gas or “frac-gas” reserves that are orders of magnitude greater than the 200 trillion cubic feet (Tcf) now in place from conventional reservoirs has rewritten the energy-supply outlook.

As General Electric Co. states, “The energy America has been looking for is right here at home: natural gas. New superefficient gas turbines can turn that natural gas into enough electricity to power every home in the country for 70 years. The more energy we find at home, the less we have to buy from abroad. And using more electric cars and trucks helps reduce our dependence on oil imports.” Based on such an expansion of reserves, a large increase in production would follow. For 50 years, the ratio of production to reserves has been about 1 to 10, so that an increase of reserves from 200 to 300 Tcf would expand production from the historic level of 20 to 30 Tcf. To use this additional supply in power generation would require at least a one-third increase in pipeline capacity to move it to power plants. And, new gas-fired plants would have to be built, given that the current capacity to generate power with gas is insufficient to both displace a significant amount of coal and to expand to meet growing demand for industrial, commercial and household electricity. For this to take place in the coming decade, mas- sive capital outlays in electric power would be required, making the new frac gas a game-changing development.

the level of production has been projected in the EIA database to 2020 as declining from 2010.

Industry spokesmen have taken the position that frac-gas reserves discovered through the hydraulic-fracturing process will exceed 800 Tcf. The basis of bias in reported frac-gas reserves is that hundreds, if not thousands, of different geological estimates are alike in being excessively optimistic. But they come from widely different geological and economic conditions.

The seven major shale deposits differ in terms of stage of production from shale formations. The Barnett shale, which is largest in historic production, has leveled off since 2010, while the Haynesville has rapidly increased from almost zero to basically the same level of production in a much shorter time. Three other promising shale formations—the Marcellus in the Northeast, the Fayetteville in Arkansas and the Woodford in Oklahoma—are at early stages in what should be a two-decade-long period of productivity.

It is unlikely that all the various experts who estimate not only near-term production but life of ultimate reserves could systematically overestimate the numbers.

The possibility exists that we can bypass the issue of “excessive optimism” by focusing on prices rather than production quantities. The U.S. Energy Information Administration (EIA) constructs a price index for frac gas that is the same as for dry gas produced and sold in wholesale/industrial retail markets since the Natural Gas Act of 1938. Projections of futures prices are based on demand models that treat future transactions “as if” they replicated that current process of setting spot prices.

EIA future prices are approximations for market clearing prices, so any excess supply that consists of over estimation is eliminated. That is, if excess future production (or reserves) are posted in biased reports to EIA data collectors, the impact on the sources would be reductions in future forecast prices.

Next period, for example, a 10% overstatement of current reserves (based on current production) would generate a 10% to 20% reduction of the next-period price forecast in the model, depending on assumed demand elasticity.

in contrast to gas, coal is not projected to decline from 2011 to 2020.

Finally, if there were a significant gas supply bubble in the forecasts, based on exaggerated reserve projections, as occurred after price deregulation in the early 1980s, there would be a significant price increase embedded in the forecast. Producers, land owners and brokers would only hold the excess newly discovered supplies for future sales at higher prices. In the last bubble, the EIA price index increased by up to 500% in 1981-83.

But to the contrary, the present EIA forecast projects Henry Hub spot prices to be approximately the same as current prices, out to 2015. The gas market is acting as if there is not going to be a shortage—or an excess—of supply relative to demand in the next few years.

Academic estimates

How much replacement of coal by gas is likely in power generation before 2020? This seems to be the core question, given that there are limited other uses for large amounts in chemical or refinery operations, industrial heating or transportation to absorb a large part of the current supply increases.

To see how far into the power-generation infrastructure the gas expansion has to go requires access to building plans for each new gas-fired facility in the country.

But there is an alternative approach for a first assessment of gas-for-coal substitution that does not require detailed plant-by-plant data. Rather than documenting planned responses of hundreds of specific corporate decision-makers, an econometric design can be used. It assumes the aggregate reduction of coal in power plants will follow the pattern set out by shifts of coal to gas in relation to relative gas-price decreases across the industry in the past 35 years.

The first approach to answering the big question is to estimate the cross elasticity of coal demand with respect to gas price. Cross elasticity means shifting from one energy source to another, cheaper one. Here, cross elasticity is defined as the percentage reduction in coal per percentage reduction in gas price.

The common sense of this ratio can be broken down into two further ratios, i.e. (percent change in coal demanded/percent change in gas demanded) x (percent change in gas demanded/ percent change in gas price). The first is the relative change in coal-to-gas volume; the second is the orthodox definition of the demand elasticity for gas alone.

A cross elasticity of 1 would indicate that gas is a perfect substitute for coal. Anything less than 1 is deemed inelastic and therefore, not a perfect substitute.

When volume changes were limited, as in the 1990s, and prices were high, due to small gains annually in reserves, both ratios would be expected to be less than one, and the product less than one.

But with the advent of frac gas, while the physical fuels change ratio could be less than one, the gas demand elasticity would be far in excess of one. The cross elasticity on an annual basis should be larger in the recent years that are not included in the databases of the econometric studies.

Based on data compiled on “the cost of fossil fuel at (all) electricity generating plants” by the EIA, experts at the University of Calgary and the World Bank have estimated numerous equations of average relationships, including the “cross elasticity of coal with respect to gas from the relative price reduction in gas.” Of interest as well are estimates developed by Michio Morishima, a professor who in 1967 developed a formula to measure elasticity of substitution when multiple fuels are in play.

The Calgary equations, based on data for 1973 to 2007, by A. Serletis and others, have a cross elasticity of coal to gas of 0.064. This indicates that for every 10% reduction in gas price, holding coal price constant, there has to be 6/10ths of a percent shift to gas in the production of electric power. With gas prices from 2008 to 2009 falling by more than 50%, and with no prediction of a return to previous high natural gas prices as frac-gas supply increases, the forecast for fuel substitution from coal to gas is little more than 3%.

The econometric estimate for the elasticity of substitution is not much different. Estimated from the same database, the Morishima elasticity of substitution is 0.201, so that the ratio of coal to gas decreases by 2% for every 10% decrease in relative gas price. Because coal has twice the installed capacity of gas over that decade, this slightly larger response leaves gas with only slightly more than 15% of the electric power-generation market.

While we could find no other studies that examined interfuel substitution for electricity generation, we were able to find additional studies that examined interfuel substitution possibilities in total energy demand. Serletis and Shahmoradi calculated the Morishima elasticity of substitutions from coal to gas at between 0.307 and 0.480, based on EIA data from 1996 to 2004.

Clifton T. Jones calculated coal-to-gas cross elasticities of between 0.027 and 0.210, based on U.S. industrial energy consumption data from 1960 to 1992. If used to project the hypothetical sweep of gas into coal markets for power generation, this would indicate that less than 20% of coal-use capacity would move over to natural gas.

Our estimate

We made an additional estimate of our own, using as much recent data as possible, on the shift from coal to gas, which would be affected by the 50% reduction in the price of gas from 2008 to 2009. With EIA coal and gas demand data and price data for the electricity sector from 1990 to 2009, we calculated an ordinary least-squares cross elasticity of coal demand to gas price of 0.20.

While the equation is simple and (most likely) the estimates are not efficient, it is probable that there are not yet enough years of fracgas supply volume to be able to assess the impact on coal utilization. The results in the summary regression table indicate that the cross elasticity approximates 0.20. Therefore, a 50% reduction in gas price would generate a 10% reduction in coal demand.

EIA future prices are approximations for market clearing prices, so that any excess supply consisting of over estimation is eliminated.

A field survey

With such a narrow range of cross elasticity estimates, based on previous price and quantity behavior before frac gas came into the markets, we have to turn to more “micro” and recent data. A representative study comes from Credit Suisse’s September 2010 Electric Utilities report, “Growth From Subtraction.” The report estimates that out of the 340 gigawatts (GW) of coal-plant capacity in the U.S., 50 GW, or 15% of capacity, will be forced to shut down, with an additional 100 GW switching to gas (or requiring significant investment to install emission control technologies to meet anticipated EPA limits on emissions), or an additional 30% of capacity.

The 100 GW of capacity that requires investments is discretionary because the alternative option is to switch to gas generators. Why? The EIA estimates that retrofitting a coal plant may cost up to $700,000 per megawatt (MW) of capacity, versus the cost of a combined-cycle gas turbine plant of $750,000 per MW.

Therefore, coal generators are vulnerable where coal pricing is at a premium to natural gas—easily determined and hedged today on the futures markets. The smaller coal units are most vulnerable, because the comparable cost of reaching environmental compliance is higher, and 70% of the small coal plants (72 GW in aggregate) were built more than 40 years ago and are fully depreciated. As a result, coal generator owners will review the economic implications of investing in emissions-control technology.

Combined-cycle gas turbines are more economic than retrofitting coal plants, with gas prices below $6 per MMBtu, given existing coal prices and the high costs of retrofits.

As dire as the future of small, old coal plants may seem, they are significant contributors to electricity generation, as they are dispatched at 48% of capacity, compared with 63% for the U.S. average. Therefore, it is possible that the implementation of EPA policy will slow down as the deadlines of late 2014 and early 2015 are extended.

Local opposition in Texas from the state electricity regulator, ERCOT, and the local generators has forced the EPA to retreat. If such opposition arises in other areas, one may expect that the EPA will allow these small plants to run until they are decommissioned. The result of business as usual? The 324 million tons per year of coal used to supply the 100 GW of plants with no environmental controls would stay in place. The 157 million tons per year used to supply the 50 GW of small coal plants, which were expected to be shut down and replaced with gas, would continue to run. In total, the 1 billion tons of coal used annually to supply the utility industry would remain practically unchanged.

using EIA coal and gas demand data and price data for the electricity sector from 1990 to 2009, an ordinary least-squares cross elasticity of coal demand to gas price of 0.20 was calculated.

Conclusions

No person or source in energy policy formation appears to have gone behind the expansive statements on the coming surge of frac-gas volumes to determine whether this is the “discovery of the present century” or not. If they had done so, they would have begun survey work that specified when, and to what extent, the new frac-gas supplies would replace coal as the preferred fuel in power plants.

But we have scavenged data and relevant estimates from diverse sources of demand cross elasticities that support an assessment. These various estimates, all based on old data, do not exceed 0.20 except in elasticities of substitution outside our range of analysis; a simple linear regression based on more recent data also had an estimate of this cross elasticity of 0.20.

If the price of gas is reduced by 50% due to the arrival of large volumes of frac gas on the market, the replacement of coal by frac gas, in our judgment, is going to be in the range of 0.20 x 50%, equal to 10%, and is not going to exceed 20%.

Turning to the micro studies of these markets by Credit Suisse, we find that the percentage could be as low as 15% and as high as 30%, based on the opportunity to replace obsolete coal plants with new gas plants that produce lower-cost and cleaner electric power.

The results in the summary regression table indicate that the cross elasticity approximates 0.20. Therefore, a 50% reduction in gas price would generate a 10% reduction in coal demand.

This will require new supplies of frac gas in the range of 1.8 Tcf to 3.7 Tcf per year—and infrastructure to move that new gas to market.

Our judgment call that the limit of gas-for-coal will not exceed 20% is based on the replacement process being determined by federal pollution-control authorities, whoever that might be.

That is not totally consistent with the hyperbole that frac gas is the U.S.’ most important new energy source of this century.

The Yale Graduate Energy Study Group, an informal group, is composed of Robert Ames (vice president, Solazyme), Anthony Corridore (marketing director, Lafarge), Edward Hirs (managing director, Hillhouse Resources LLC, and lecturer at the University of Houston) and Paul MacAvoy (Yale professor emeritus).