1. The severity of the crisis is much greater than I assumed might occur when I first began speaking out on the potential for sharply escalating natural gas prices almost five years ago. Prices have risen even more steeply than I assumed, and expanding supplies to keep pace with increases in demand has proven to be even more difficult. The result has been a steep increase in prices, in order to drive a sufficient number of users out of the market to match supply and demand.
Further, the increases in natural gas prices that have occurred to date could prove to be just the “tip of the iceberg.” By no later than the middle part of the next decade, a massive gap is likely to develop between the amount natural gas required by the U.S. economy and supplies available to the U.S. market. While the size of this supply gap will depend upon many factors, by 2020 it could reach as much as 10 Tcf/year. In Btu equivalent terms, this is nearly twice the amount of oil the U.S. currently imports from the Middle East.
2. An energy supply gap of this magnitude could result in hundreds of billions of dollars per year in needless energy costs and seriously impede the growth of the U.S. economy. While this massive shortfall in supplies will not necessarily result in physical curtailments of supply, at least on a routine basis, the price increases required to balance supply and demand could prove to be brutal. Further, under some plausible scenarios, by the end of the next decade, supplies to electric generators could become subject to periodic curtailments – posing a significant threat to electric supply reliability.
3. This impending crisis can be prevented by adopting a comprehensive national energy strategy to reduce U.S. dependence upon natural gas as a fuel to generate electricity. It is still possible to avoid this crisis. What has been missing to date has been an understanding of the severity of the potential natural gas supply gap facing the U.S. market and a willingness to develop and implement a comprehensive, realistic strategy to address it.
4. Action must be taken now. Given the lead time required to make major changes in the existing U.S. energy infrastructure and the magnitude of the potential supply gap, however, steps must be taken immediately to begin putting an alternative strategy in place. Delay will virtually guarantee steep further increases in natural gas prices in future years, and the permanent loss of additional manufacturing jobs, as industrial users continue to be forced out of the U.S. market, in order to free up additional supplies of natural gas for use to generate electricity – and, increasingly, to produce ethanol and synthetic crude from oil sands in Canada.
EIA’s Long-term Forecasts Significantly Understate
The Natural Gas & Electricity Price & Supply Risks Facing the U.S.
How could this be? How could the U.S. economy be faced with a potential energy supply crisis of this severity without the magnitude of this potential crisis being more widely understood?
While there is no single answer, the most important factor is the long-term forecasts of expected supply, demand and price of natural gas issued by the U.S. Department of Energy’s Energy Information Administration (EIA).
Earlier in the decade, EIA severely over-estimated the supplies of natural gas likely to be available to the U.S. market and drastically under-estimated the likely price. Unfortunately, there is a significant risk that EIA’s current estimates will prove to be just as far off the mark, potentially resulting in far reaching harm to the U.S. economy. It is important, therefore, that policy-makers begin to develop a better understanding as soon as possible of the extent of the price and supply risks faced by the U.S. economy in coming years.
The Reference Case price forecasts issued by EIA over the past 12 months provide a useful starting point for critiquing EIA’s work.
In its 2006 Annual Forecast, for example, issued this past February, EIA predicted that the wellhead price of natural gas in the U.S. would drop below $6.00/mmBtu within 24 months, averaging $6.13/mmcf (equivalent to $5.97/mmBtu) in 2007 and $5.78/mmcf ($5.62/mmBtu) in 2008. See Annual Energy Outlook 2006 (AEO 2006), Year-by-Year Reference Case Tables, Table 13.
These prices are almost $ 2.00/mmBtu below the current NYMEX 12-month strip prices for 2007.
In EIA’s 2006 forecast, however, these two years were the outliers at the high end of EIA’s range of expected future prices – i.e., according to EIA’s forecast, the last two high price years.
After 2008, EIA predicted that prices will begin falling like a rock, declining to $4.46/mmcf ($4.34/mmBtu) by 2016 (in $2004) and remaining below $5.00/mmBtu until 2023, when EIA predicted that the average wellhead price would increase to $5.05/mmBtu in $2004:
EIA’s February 2006 Price Forecast, Annual Energy Outlook 2006 ($/mmBtu)
For EIA to publish a price forecast in February of 2006 predicting that the wellhead price of natural gas would fall below $ 5.00/mmBtu for much of the next decade is little short of stunning.
As a practical matter, since liquidity in the futures market drops off dramatically after the first 24 to 36 months, EIA’s long-term price forecast is the most important long-term price signal available to the market. In addition, EIA’s long-term forecast of supply and demand provides the starting point for most private forecasts of natural gas prices and, indirectly, most electricity price forecasts.
As explained in this and subsequent articles, EIA’s recent forecasts have no hope of proving to be accurate in the real world. Unless and until the errors in EIA’s forecasts are corrected, however, serious mistakes in energy supply planning may be inevitable.
When it issues its 2007 long-term forecast, it appears likely that EIA will back off of its 2006 price forecast – at least to a limited degree. The Early Release version of Annual Energy Outlook 2007 (AEO 2007), issued earlier this month, increases EIA’s forecast of U.S. prices in most years during the period between 2007 and 2020:
Average Lower 48 Wellhead Price, AEO 2007 vs. AEO 2006 (2005$/mmBtu)
Further, this increase in predicted price levels occurs despite: (i) a decrease in expected demand; and (ii) higher expected production from U.S. on-shore wells, without any explanation of why lower demand and increased supply should lead to an increase in predicted price levels.
As discussed below, however, even with this modest bump-up in prices, EIA’s Reference Case price forecast almost certainly is far off the mark – understating significantly likely price levels in many years even in the specific conditions assumed in EIA’s Reference Case scenario, and understating price levels even more drastically in more likely, real world conditions.
EIA Has Demonstrated Repeatedly that It’s Inability to Develop
Reliable Estimates of Future Supply, Demand or Price
During the past several years, EIA’s price forecasts typically have been far off the mark. Cumulatively, for example, its 2002 Reference Case forecast underestimated U.S. natural gas costs over the past 5 years by nearly $350 billion, for natural gas alone:
Taking into account the additional impact on the wholesale price of electricity, the total increase in energy costs undoubtedly is substantially in excess of $ 400 billion – which approaches the cost-to-date of the war in Iraq.
Since 2002, EIA has increased its natural gas price forecast every year. Notably, however, despite this pattern of continuing increases, as recently as January of 2005, EIA was predicting an average wellhead price for this year of $4.55/mmBtu -- almost $2.00/mmBtu below its most recent estimate for 2006 -- and predicting that prices would decline to $3.75/mmBtu by 2008.
EIA’s recent track record should provide a strong cautionary note, therefore, regarding the likely accuracy of its future forecasts – especially since the personnel and underlying methodology EIA uses to develop its forecasts have not undergone any fundamental revisions during this period.
Nonetheless, many private forecasting firms continue to use EIA’s forecasts as the starting point for their analyses and issue long-term price forecasts that (not surprisingly) are in the same general range as EIA’s forecast.
By the Mid to Later Part of the Next Decade, Prices Could Increase To 2-3 Times EIA’s Forecast Level
The flaws in EIA’s most recent price forecast can not be easily dismissed. Instead, they are staggering in scale. They stem from EIA simultaneously:
There also is a fundamental disconnect between EIA’s price forecast and its estimate of future U.S. demand for natural gas. Consumption of natural gas, of course, can not exceed available supplies for any sustained period. Instead, supply and demand must match.
To the extent the amount of gas needed by the U.S. economy significantly exceeds available supplies, therefore (as we believe inevitably will occur over the next decade) prices will need to be driven up -- potentially quite steeply -- to drive out of the market a sufficient number of users so that consumption matches supplies available for use.
EIA asserts in the Early Release version of AEO 2007, however, that despite more than a 40 % increase in total electric generation over the period covered by EIA’s forecast, power sector consumption of natural gas will peak in 2016, after increasing modestly from current levels, and then gradually decline, falling below current levels despite a 10 to 25 % projected decline in the real price of natural gas (depending upon the year). Further, despite this sharp decline in natural gas prices, during this same period, use of natural gas in the industrial sector is expected to increase only modestly from current post-2005 Hurricane lows.
The likelihood of EIA’s assumed scenario materializing, however, in our judgment is essentially zero. Instead, over the next 10 to 20 years, steep increases in the price of natural gas are likely to be required, in order to limit demand; the likelihood that the price of natural gas will decline significantly from currently levels and that demand will nonetheless simultaneously decrease in the power sector and remain largely stagnant overall, despite huge growth in overall energy use in the U.S. economy, defies common sense.
That is exactly the assumption that much be accepted, however, in order for EIA’s most recent long-term natural gas and electricity price forecasts to be treated as credible.
EIA’s Estimate of Future U.S. Demand is Far Off the Mark – Particularly in the Mid-to-Later Part of the Next Decade
The errors in EIA’s natural gas supply forecast will be discussed in Part III of this series. In estimating how much natural gas would be demanded at the price levels EIA forecasts, however, EIA commits at least eight major errors. Specifically, EIA:
1. Drastically understates likely future power sector demand for natural gas – in all likelihood by at least 2.0 to 2.5 trillion cubic feet per year by the mid- to later part of the next decade and potentially by as much as 5.0 to 7.5 Tcf by the 2025 to 2030 time frame;
2. Fails -- at least prior to the Early Release version of AEO 2007 -- to adequately take into account the rapid increase in demand that is occurring for use of natural gas to expand production of ethanol and other bio-fuels (which typically are very natural gas intensive, both for fuels processing and for growing crops);
3. Does not make any provision for the potential impact on natural gas consumption of requirements for reductions in emissions of Greenhouse Gases that might be enacted at either the federal or State level (such as A.B. 32, recently signed into law into California) and/or other new emissions restrictions that might result in increased utilization of gas-fired plants, even though the potential impact of such requirements, if they were ever to be adopted, is huge – potentially increasing demand for natural gas by several Trillion Cubic Feet per year;
4. Makes assumptions that could prove to be overly-optimistic regarding life extensions for aging coal-fired plants -- especially if environmental requirements are tightened;
5. Fails to assess the potential impact on demand for natural gas and likely price levels of natural gas of deviations from “normal” weather;
6. Fails to adequately evaluate the potential impact of lower-than-expected availability of other forms of generation (i.e., nuclear, hydro and coal), due to poor performance, retirements of older units, poor hydro availability, or potential conversion of coal-fired units to natural gas;
7. Examines only a narrow range of scenarios regarding potential future oil prices; and
8. Fails to adequately take into account the potential impact of higher oil prices on demand for natural gas and the price and availability of LNG.
Every one of these failings is significant; each is discussed in detail below.
The combined impact of these flaws, however, is to paint a fundamentally inaccurate picture of likely future supply, demand and pricing of natural gas in the U.S. market – and therefore electricity pricing and reliability as well, given the increasing dependence of the U.S. electricity market on gas-fired generation.
In order to have any hope of developing a realistic U.S. energy strategy, therefore, it is essential that a comprehensive new assessment of the North American natural gas market be undertaken immediately that properly assesses these issues.
Over the Past Several Years, EIA Has Repeatedly Failed to Anticipate Increases in the Use of Natural Gas to Generate Electricity
During the past several years, increased power sector demand for natural gas has been one of the primary factors driving up the price of natural gas, with an aggregate increase of more than 1.25 Tcf/year over the past 3 years.
Within the industry, the causes of this increase are well known. On a weather-adjusted basis, U.S. demand for electricity typically grows every year. Further, since the early ‘70s, no new nuclear plants have been licensed and only a handful of new coal-fired plants have been permitted.
Nonetheless, throughout the ‘80s and most of the ‘90s, as a result of surplus generating capacity that came on line shortly after the oil price shocks in the late ‘70s and improvements in plant performance during the ‘80s and ‘90s, it was possible to satisfy increased U.S. demand for electricity almost entirely by increased generation from existing coal and nuclear plants. Prior to 1998, increases in the use of natural gas to generate electricity were surprisingly small:
Beginning in the late 1990s, however, for the first time in two decades, electric utilities were required to add large amounts of new generating capacity in order to serve continuing increases in demand. At the time, EIA’s natural gas supply forecast predicted that natural gas production could be rapidly expanded, with only modest impact on the cost of natural gas, even over extended time periods.
As recently as four years ago, for example, in 2002 – in one of the costliest errors in U.S. forecasting history – EIA still was confidently predicting that supplies of natural gas delivered to the U.S. market from the lower 48 states and existing fields in Canada could be increased by almost 50%, to 34.1 Tcf by 2020, with the wellhead price of natural gas remaining in the $2.30 to $3.17/mmBtu range or lower throughout the 20-year period covered by EIA’s forecast (in $2000).
We now know, of course, that EIA’s expected 34 Tcf, $3.17/mmBtu scenario has no hope of being achieved. Within less than 12 months after EIA issued its forecast, the wellhead price reached $3.15 and has never fallen below that level since. Despite record mild weather last winter, which has helped to hold down prices through much of this year, EIA’s most recent estimate of the wellhead price for 2006 is $6.49/mmBtu – more than twice the upper end of the 20-year price range it predicted just 4 years ago.
In the interim, however, the power industry has constructed more than 225,000 MW of new generation – virtually all of it gas-fired:
This is enough generation to serve the total current load in Germany, Great Britain and France combined – assuming adequate supplies of natural gas could be found to operate these units.
The net impact of the construction of this new generation has been to fundamentally transform the generating mix in the U.S. – and to permanently eliminate a significant number of U.S. manufacturing jobs.
Prior to the addition of these gas-fired generating units, except in Texas and a few other areas of the country (e.g., along the California coast), natural gas generally was used only in peaking units, many of which were operated less than 100 hours a year.
Now that this massive construction program has been completed, however, more than 40% of all of the generating capacity in the U.S. is gas-fired. Further, gas-fired generating units are now the marginal source of electricity supply in most Regions of the country for an increasing number of hours each year.
This in turn has had the effect of causing the amount of natural gas used to generate electricity to escalate at a startling rate for several consecutive years:
As a result, in July of this year, for the first time, more than 1 Trillion Cubic Feet of natural gas was used to generate electricity in a single 31-day period – an average of 32.3 Bcf/day! In effect, during this period, almost 60% of the natural gas produced by the 500,000 + operating wells in the U.S. was used simply to generate electricity. In August, consumption of natural gas to generate electricity was almost as high.
Further, while temperatures this summer were much hotter-than-normal, 6 of the past 7 summers have been hotter-than-normal. In addition, load is continuing to grow every year. Even if temperatures over the next few summers revert to more normal levels, therefore, summer-month consumption will soon match last summer’s levels.
Winter-month power sector consumption of natural gas also has recently begun to accelerate sharply, and could become an increasingly important factor adding to winter-month demand for natural gas in future years.
The net effect of the addition of this massive amount of new gas-fired generation has been to profoundly change the power industry’s fuel mix, leaving the industry dependent upon gas-fired generation to meet a significant share of incremental demand for electricity even in Regions that historically have relied primarily on coal and nuclear generation. Further, as a practical matter, as a result of the massive number of new gas-fired generators added since 1999, at least for the next 7 to 10 years, the industry has no alternative other than to continue to rely primarily on increased use of gas-fired generating units to meet expected growth in demand for electricity:
U.S. Dependence Upon Gas-Fired Generation Likely to Continue to Grow
EIA, however, appears not to grasp the significance of this shift. Its last several forecasts have failed almost entirely to anticipate the increase in power sector consumption of natural gas that has occurred over the past three years.
Strikingly, EIA’s most recent final Annual Energy Outlook, AEO 2006, issued in February of this year, estimated that, over the 5-year period between 2004 and 2009, power sector consumption of natural gas would increase by a cumulative total of only 40 Bcf nationwide over a 5-year period – viz., from 5.32 Tcf/year in 2004 to 5.36 Tcf/year in 2009. This is an average growth of only 8 Bcf/year (i.e., approximately 0.15 %/year), starting from a year in which summer-time weather was unusually cool, minimizing power sector consumption of natural gas!
EIA’s February 2006 Estimate of 2006-2009 Power Sector Demand for Natural Gas (trillion cubic feet)
At a time when gas-fired generation is the marginal source of supply in virtually every Region of the country, it is difficult to understand how EIA could have convinced itself this was a plausible estimate – especially since the summer of 2004 was one of the mildest in the past 30 years, holding power sector consumption of natural gas in the summer of 2004 significantly below normal levels. In effect, it’s almost as if EIA concluded that this enormous fleet of new generating units – which with expected additions will soon equal total U.S. generating capacity in 1970 – could be run on thin air.
This year alone, power sector consumption of natural gas is likely to exceed EIA’s estimate of likely consumption in 2006 by more than 1 Trillion Cubic Feet (or 20.5%) – i.e., at least 6.22 Tcf, vs. an estimate in Annual Energy Outlook 2006 (AEO 2006) of 5.16 Tcf. This is a level that as recently as this past February EIA forecast would not be reached until 2013 – 7 years from now!
This is not a minor discrepancy; instead, increased demand of this magnitude ordinarily is sufficient to cause a severe price spike. Only extremely mild weather this past January (which held withdrawals from storage in January more than 400 Bcf below normal levels) prevented natural gas prices from reaching double digit levels during the last several months of this year.
While record hot weather this past July and early August and much lower-than-normal power sector use of residual fuel oil contributed to this higher-than-expected consumption, less than ½ of the variance from EIA’s estimate can be explained based upon these factors, in part because the impact of hotter-than-normal weather this summer was partially offset by lower-than-normal power demand for electrostatic heating last winter and by favorable hydro availability in the Pacific Northwest throughout much of the year. Further, EIA projects that consumption of residual fuel oil will remain at significantly lower levels for much of the next decade.
In the Early Release version of AEO 2007, EIA recognizes that its estimates of power sector demand during the period between 2007 and 2012 are untenable, and increases sharply its estimates of power sector consumption of natural gas during this period, compared to AEO 2006:
Power Sector Consumption of Natural Gas, AEO 2007 vs. AEO 2006 (trillion cubic feet)
This drastic revision, just 10 months after EIA issued AEO 2006, should be sufficient, by itself, to raise major issues regarding the reliability of EIA’s estimates. For the period between 2007 and 2012, both AEO 2006 and the Early Release version of AEO 2007 assume a nearly identical generating mix. Further, total demand is nearly the same in both reports. The most significant difference is that natural gas prices are assumed to be higher in AEO 2007 than in AEO 2006, at least for the next few years. It is not immediately apparent, therefore, why use of natural gas to generate electricity is assumed to jump significantly during this period, compared to estimates EIA presented earlier this year.
Even with these increases during the period between 2007 and 2012, however, EIA’s estimate of future power sector consumption of natural gas almost certainly falls far short of the mark.
Notably, even with these revisions, EIA’s estimates of power sector consumption are still up to 2.0 Tcf/year below EIA’s estimates of future power sector consumption of natural gas issued just 2 years ago in Annual Energy Outlook 2005 (AEO 2005) (which in our view are still too low):
Rather than responding to evidence that there is a major flaw in its forecasts that causes it to understate growth in power sector demand for natural gas by thoroughly reviewing its methodology, therefore, and replacing it with an approach that allows EIA to more accurately assess the market, over the past two yeas, EIA has reduced its projection for long-term power sector use of natural gas dramatically.
Further, this reduction is expected future use of natural gas to generate electricity has occurred even though EIA continues to project that: (i) natural gas prices will decline significantly in real terms compared to current levels; and (ii) U.S. electricity demand will grow by more than 40 % during the period covered by its forecast!
In short, despite far higher-than-expected near-term growth in consumption of natural gas, at price levels far above EIA’s forecast levels, longer-term, EIA is projecting that power sector consumption of natural gas is likely to decline, despite a significant decline in the cost of natural gas and continued strong growth in U.S. demand for electricity!
EIA may well believe this is a plausible scenario; it is not entirely clear, however, why others should – at least without a dramatic change in national energy policy compared to the status quo.
Further, even in near-term, the revised forecast of power sector consumption of natural gas presented in the Early Release version of 2007 bears little relationship to what realistically can be expected to occur in the U.S. market.
It is entirely plausible, of course, that after this year’s record hot summer, power sector consumption of natural gas will decline in 2007 -- although even this coming year such a decline is by no means certain to occur.
Between 2006 and 2011, however, EIA’s forecast assumes that net electric generation will increase by approximately 372,000 GWhrs – a cumulative increase of 9.1 % over the next 5 years (including the parasitic load associated with new pollution control equipment).
Remarkably, even though gas-fired generation currently is the marginal source of supply in every Region of the country, EIA assumes that this massive increase in electric generation will have only a modest impact on use of natural gas to generate electricity over this 5-year period.
Specifically, it’s forecast of power sector consumption of natural gas five years from now (i.e., projected consumption of 6.51 Tcf in 2011) allows room for only 300 Bcf of increased natural gas consumption over a period of 5 years, compared to EIA’s most recent estimate of actual 2006 levels in it’s December Short-Term Energy Outlook:
Power Sector Consumption of Natural Gas vs. Actual, AEO 2007 (trillion cubic feet)
This averages out to an increase of just 58 Bcf/year – which is less than the year-over-year increase that occurred in some weeks this past summer.
EIA’s forecast then calls for power sector consumption of natural gas to: (i) increase modestly between 2011 and 2016; (ii) remain relatively flat between 2016 and 2020; and then (iii) begin to rapidly decline starting around 2021:
Power Sector Consumption of Natural Gas, AEO 2007 Projections (trillion cubic feet)
Is this forecast of future U.S. power sector use of natural gas even remotely plausible?
We believe the answer is no. Instead, both near-term and longer-term, EIA’s estimate of future power sector consumption of natural gas has virtually no chance of proving to be valid, even under the increasingly unlikely scenario in which no further emissions restrictions are adopted that create an increased preference for natural gas vs. coal.
Under any scenario in which even moderate restrictions on emissions of greenhouse gases or criteria pollutants are enacted at either the federal or state level, the gap between EIA’s estimate and actual power sector consumption of natural gas is likely to be even more extreme.
By the Middle Part of the Next Decade, Power Sector Consumption of Natural Gas Is Likely to Exceed EIA’s Most Recent Estimates by At Least 1.5 to 2.0 Tcf/year
Even before taking into account the potential impact of changes in environmental requirements, for example (some of which already have begun to be enacted at the State level), it should be clear that by the middle part of the next decade, power sector demand for natural gas in a typical year is likely to exceed EIA’s estimates by at least 1.5 to 2.0 Tcf/year in a typical year. If new emissions restrictions are adopted that increase utilization of natural gas, the increase in power sector consumption of natural gas, compared to EIA’s estimate in it’s the Reference Case forecast in the Early Release version of AEO 2007 in some years could be as much as 2.5 to 3.0 Tcf/year.
By 2025 to 2030, even if there are no new environmental restrictions that favor gas-fired generation, under some plausible scenarios, power sector consumption of natural gas could increase by 5.0 Tcf/year or more, compared to EIA’s estimates in the Early Release version of AEO 2007.
The likelihood of higher power sector consumption of natural gas, even in a “no change” in environmental requirements scenario, is due to a number of different factors. The most important, however, are the following:
1. Inability of EIA’s model to accurately model power sector consumption of natural gas on peak summer days, when consumption of natural gas is at its height. While EIA does not provide sufficient information to fully evaluate the modeling on which it bases its estimates, from the data EIA reports, it appears that the production cost model EIA uses to simulate system dispatch understates the extent of power sector consumption of natural gas on peak summer days (possibly by treating peak demand as a large block spread out evenly over a relatively long period, rather than compressing it into a relatively small number of days). This is critical, since it’s the “peakiness” of demand on hot summer days that creates the greatest need to use gas-fired generation – which is generally the last to be dispatched.
As a result of this flaw, EIA starts its analysis with an estimate of current year consumption of natural gas which is too low – understating power sector consumption of natural gas by approximately 330 Bcf in 2006, compared to the separate estimate of 2006 demand published in EIA’s Short-term Energy Outlook, after adjusting for differences in categorization between the two reports. This has the effect of skewing all of its estimates for subsequent years.
By 2020 or 2025, the underestimate could easily be twice as large as the underestimate for 2006, given the critical role peak summer demand plays in driving total power sector consumption of natural gas for the year.
2. Inappropriate assumptions regarding summer weather conditions. EIA further distorts its analysis by basing its modeling on 30-year climatological norms. While this might initially appear to be a neutral basis for preparing EIA’s forecasts, as noted earlier, only 1 recent summer has been normal or milder than normal compared to this standard (i.e., the summer of 2004). Further, at least currently, the 30-year climatological norm is not nearly as “neutral” or “objective” a standard as might initially appear to be the case. This is because the first 15-years used in calculating the current 30-year norm (i.e., the period from 1971 to 1985) included the longest stretch of generally mild summer weather of the past 75 years:
Just replacing the current 30-year norm starting in 1970, therefore, with a sequence starting in 1980 (as will occur anyway in another few years) would lead to different results.
This is not a minor issue. Basing EIA’s forecasts on a 10-year rolling average, for example (an approach used by many analysts) would increase the assumed number of Cooling Degree Days by approximately 15 %, for the summer months alone:
This would be likely to increase EIA’s estimate of power sector consumption of natural gas by 200 Bcf or more for 2007 – and more in subsequent years.
3. Inappropriate assumptions regarding hydro availability in the Pacific Northwest. Similarly, EIA uses 30-year norms in calculating hydro availability in the Pacific Northwest – which also tends to create a downward bias in its estimate of natural gas consumption. While hydro availability will sometimes match the 30-year average (as it has in 2006), the trend for some time now has been that these years are the exception rather than the norm, with availability using falling well below this level. Once again, therefore, the effect of the assumption used in EIA’s Reference Case is to understate natural gas consumption in a typical case; use of a 10-year rolling average would produce a more defensible baseline estimate.
4. Consistent use of assumptions that are far too optimistic regarding the potential for a sudden jump in the output of existing coal-fired plants. Further, every year since at least 2000, in issuing its annual forecast, EIA has assumed that, in the near future, there will be a sudden huge jump in the output of existing coal-fired plants. The timing and magnitude of this expected sudden acceleration in the output of existing coal-fired plants has varied from forecast to forecast. In the last several Annual forecasts, however, it generally has been predicted to start the year after the forecast was issued (whenever that happened to be in a particular year):
The problem, unfortunately, is that EIA’s assumption that this sudden jump in output will occur has proven to be incorrect every time. Some growth in output has occurred, but seldom at even close to the levels EIA has predicted. The fact that EIA has been wrong in this prediction, time after time, has had little impact, however, on the Agency’s apparent optimism that a major change in performance is “just around the corner” – which EIA has persisting in building into its Base Case estimates for the better part of a decade.
In the Early Release version of AEO 2007, EIA finally appears to be taking at least a slightly different approach. The period in which rapid growth in output is assumed to occur has been pushed back slightly, to begin in 2009 rather than 2008, and then continue through 2012. Further, the assumed increases are not quite as aggressive as in previous years.
During this period between 2009 and 2012, however, the output of existing coal-fired plants still is assumed to increase by almost 100,000 GWhrs, over and above the increased output that might reasonably be expected from new coal-fired plants expected to be added during this period.
The effect of this paper assumption is to eliminate, with the wave of a hand, up to 800 to 850 Bcf per year in potential power sector consumption of natural gas that is likely to occur if the output of existing plants fails to suddenly jump as EIA’s projections assume.
Is it theoretically possible that the increases EIA assumes can be achieved? Yes, the possibility that EIA’s assumed output levels could be achieved, at least in some years, cannot be entirely ruled out.
The question, however, is whether EIA’s assumption is a reasonable starting point for developing a Base Case forecast of demand for natural gas. Here, the answer is almost certainly no.
If there ever was the potential to achieve the sudden increases EIA assumes, the best prospects probably were over the past few years. Natural gas prices certainly have been high enough to create strong incentives to operate existing coal-fired units at maximum capacity. Further, in some Regions of the country, there still were extended periods in which available coal-fired units were not fully utilized for extended periods of the year.
With each passing year, however, the nation’s existing fleet of coal-fired units continues to age; a surprising large percentage of these units a very old (> 50 years) and quite small (< 200 MW).
Especially as environmental requirements tighten, it is less clear that it will make sense to continue to pour large amounts of capital into the maintenance of these units. As a result, maintaining current availability levels at many of these units may become increasingly difficult and retirements could significantly exceed the levels EIA assumes (especially if natural gas prices were anywhere close to the levels that EIA forecasts).
Further, every year, there are fewer and fewer hours in which most existing coal-fired units are not already being dispatched at or near maximum capacity.
If anything, therefore, in future years, the output from existing units may be increasingly likely to plateau – or potentially even decline (depending in part on the number of retirements).
In preparing a Base Case forecast, therefore, it would have been more reasonable for EIA to take a more cautious approach – e.g., cutting the assumed rate of growth in output from existing units at least in half, if not further.
If it did so, by the end of the next decade, this would add another 400 to 500 Bcf to EIA’s estimate of projected power industry consumption of natural gas.
5. Assumptions regarding the addition of new coal-fired capacity could also prove to be wildly over-optimistic. Finally, particularly during the period beginning in approximately 2014, EIA’s estimates of natural gas consumption are highly dependent upon its assumptions regarding the addition of new coal-fired plants. EIA assumes, for example, that a total of 32,400 MW of as yet-uncommitted plants will be added by 2020, and an additional 46,400 MW in the next 5 years, bringing total additions by 2025 to 78,800 MW.
One can fervently hope that these plants will be built – since the need is clearly present, from the standpoint of national energy policy. Whether it is appropriate, however, to treat these plants as if they already were “steel in the ground” is an entirely different matter – particularly given legislative initiatives already being considered at both the federal and state level that might strongly discourage (or at least delay for several years) construction of new coal-fired plants.
Absent the construction of these plants, however, during the decade beginning in 2020, power sector consumption of natural gas could nearly double compared to the levels estimated in the Early Release version of AEO 2007 – i.e., quite literally, the amount of natural gas needed to “keep the lights on” in the U.S. could increase by as much as 5.0 to 7.5 Tcf/year compared to the level EIA estimates in the Early Release version of AEO 2007. Further, this increase could occur merely as a result of a reluctance to build new coal-fired plants, even if there are no new environmental restrictions that specifically preclude the construction of these plants.
At a minimum, therefore, given the critical importance of this issue, in preparing its annual forecast, EIA should explicitly model a range of different scenarios regarding both the amount of new coal-fired capacity that might be built over the next two decades and the likely output of existing coal-fired plants, and present the results of this analysis, rather than presenting as its Reference Case a scenario that assumes that coal-fired units will be available and dispatched whenever, on paper it would be consistent with principles of economic dispatch to build and utilize coal-fired generation.
In all five instances discussed above, therefore, EIA has used a methodology that has the potential to greatly minimize projected future power sector consumption of natural gas, compared to what is likely to occur in “real world” conditions.
Its Reference Case forecast, therefore, is hardly a “most likely” scenario. Instead, even in a “no new environmental restriction” scenario, it is much closer to an extreme “best case” scenario, in which on a whole series of critical issues, EIA has made assumptions that it should have recognized are likely to greatly understate the likely use of gas-fired capacity in the conditions that are likely to actually occur over the next 15 to 20 years.
The extent to which EIA has underestimated future consumption levels will depend on many factors, and varies for different time periods.
At a minimum, even in a “no environmental change” scenario, by the mid to later part of the next decade, in many years power sector consumption of natural gas will be 1.5 to 2.0 Tcf per year higher than EIA’s estimate.
This is a stunning figure. By 2015, for example, EIA estimates that power sector consumption of natural gas will increase by 1.30 Tcf/year over the level assumed for 2006 in its forecast – an increase of 3.6 Bcf/day. If the actual increase in power industry consumption, however, even in the “no change” scenario, is 1.5 to 2.0 Tcf/year greater (i.e., more than twice as great) this potentially would equate to an increase in power sector consumption of natural gas of as much as 9 Bcf/day – even in the unlikely event that there is no further tightening of environmental requirements that increase demand for gas-fired generation.
Further, during this same period, use of natural gas in other sectors is projected to increase by 1.76 Tcf/day. In the aggregate, therefore, by 2015, the amount of natural gas required to meet the needs of the U.S. economy, even under a “no change” scenario, could easily increase by at least 12.5 to 13.9 Bcf/day:
Increased Demand by 2015 – “No Change” in Environmental Requirements
Increased demand of this magnitude – viz., 12.5 to 13.9 Bcf/day – would be sufficient to precipitate a major crisis, even if there were no other flaws in EIA’s forecast, since there is no apparent source of supply that is likely to be adequate to meet an increase in demand of this magnitude at prices even remotely in the range EIA is forecasting.
As we move closer to 2020, however, unless large amounts of new coal-fired capacity begin to be added, the increase in the amount of natural gas required to meet the needs of the U.S. economy could begin to escalate at a much more rapid rate – potentially escalating by 400 to 500 Bcf/year on a weather-adjusted basis every year.
By the 2025 to 2030 time frame, therefore, even in the “no change” scenario, the total increase in the amount of natural gas needed to generate electricity could exceed EIA’s estimate by as much as 5.0 to 7.5 Tcf/year.
This huge potential increase in demand, of course, is no where apparent in EIA’s forecast, since EIA simply assumes that whatever coal-fired capacity is needed to prevent this demand from arising will be built – and brought on line exactly when it is needed, apparently irrespective of whether there is a regulatory mechanism in place that ensures full cost recovery for the entities expected to build these plants.
Unfortunately, however, this underestimate – as severe as it is – is just the “tip of the iceberg” in terms of the flaws in EIA’s estimate of the amount natural gas that potentially could be required to meet the future needs of the U.S. economy or any of the other factors that potentially might deter generators from building these plants.
EIA Does Not Attempt to Identify or Systematically Assess Potential Sources of Increased Demand for Natural Gas
Perhaps the most serious flaw in this regard is that EIA never attempts to systematically identify or assess the potential impact of possible sources of increased demand for natural gas, on either a short-term or long-term basis.
This is a fundamental shortcoming in EIA’s analysis, particularly because: (i) short-term elasticity of supply for natural gas is the lowest of any major fuel; (ii) even over a several year period, elasticity of supply is only modestly greater (as the experience of the past several years amply demonstrates); and (iii) natural gas is the most expensive fuel to store.
As a result, even relatively modest increases in demand (e.g., perhaps 200 to 400 Bcf in a particular 12-month period) potentially can lead to major price spikes (e.g., the quadrupling of prices that occurred in 2000).
Any omission in EIA’s estimate of future demand, therefore, potentially can “blow out of the water” EIA’s price forecast – leading to steep price increases, potentially comparable in magnitude to the total wellhead prices predicted in AEO 2006 (i.e., up to $ 4.00 to 5.00/mmbtu). These price increases, in turn, potentially could equate to $ 100 billion or more in increased customer costs in a single year.
To develop an accurate estimate of potential future costs for natural gas, therefore, it’s essential to identify and properly quantify every major factor that could lead to such increases – on either a sustained or short-term basis.
The Need to Develop Alternative Fuels is Already Beginning to Significantly Impact Demand for Natural Gas
In the past 12 to 18 months, for example, it’s already become quite clear that the need to develop alternative fuels to supplemental conventional oil supplies and replace the use of MTBE as an additive is likely to have a dramatic impact on demand for natural gas over time.
The decline in oil supplies effects demand for natural gas in at least 3 different ways:
- Huge amounts of natural gas are needed in Canada to mine and process oil sands (reducing supplies of natural gas available for export to the U.S.);
- Large increases in the use of natural gas in the U.S. already have begun to occur in connection with the production of ethanol and bio-diesel, both to grow crops and to produce bio-fuel;
- The amount of natural gas needed in the refining process has begun to increase significantly because of the need for additional hydrogen, to process heavier grades of crude.
These developing trends still are at an early stage and are likely to intensify over time. Within the next 5 years, for example, consumption of natural gas for production of bio-fuels could easily increase by an additional 500 Bcf/year, over and above the increases in the past year – making the bio-fuels industry one of the largest natural gas-using industries in the U.S. economy.
At best, only a portion of this potential growth in demand is factored into EIA’s most recent estimates of future U.S. consumption. At a time when both U.S. production and imports from Canada are likely to be declining, this is more than enough to lead to a significant increase in prices in the U.S. market.
If oil shale is developed on a major scale, the resulting increase in demand for natural gas as a fuel to generate electricity could be several times this size.
In the Current Environment, it is Irresponsible to Issue Natural Gas Price Forecasts that Fail to Explicitly Assess the Potential Impact of New Environmental Restrictions That Could Result in Increased Use of Natural Gas
Further, in estimating the potential demand for natural gas in the U.S. market, EIA makes four other limiting assumptions that would cause its forecast to be highly misleading even if there were no other fundamental flaws in its analysis. Specifically:
1. Perhaps most significantly – for purposes of its Annual Forecast, EIA does not make any effort to assess in a systematic, rigorous manner the potential impact of possible new environmental restrictions, and instead explicitly assumes that there will be no tightening in applicable emissions requirements at any time in the 25-year period covered by its forecast, even though legislation already pending before Congress could lead to dramatic increases in demand for natural gas;
2. EIA attempts to evaluate potential demand for natural gas and likely price levels only under assumed “normal” weather conditions;
3. For purposes of its analysis of likely power sector consumption of natural gas, EIA assumes “normal” availability of all other forms of generation (i.e., nuclear, hydro and coal), with only minimal future retirements; and
4. EIA examines only a narrow range of scenarios in terms of potential future oil prices, and does not explicitly link the price or availability of LNG to the price of oil.
These are huge flaws, which go to the heart of the mission we should be asking EIA to undertake.
At first blush, to an outsider to the industry, EIA’s use of these simplifying assumptions might seem understandable. Attempting to systematically assess uncertainties, and developing a reasonable set of scenarios to evaluate is not an easy task – and could easily create controversy. Indulging in the assumption that there will be “no change” in the future, and that weather conditions will match historical norms, simplifies EIA’s work, and – at least in a limited sense – eliminates a potential source of controversy.
Fundamentally, however, the issue is whether EIA’s mission is to inform policy-makers, energy users and others of the range of potential outcomes which reasonably could occur in future years or instead to assess future supply and price of natural gas only in purely academic terms, in an assumed “future world” that in all likelihood will never exist.
If EIA’s goal is to provide useful information to decision-markers, it has no choice other than to try to face squarely the major uncertainties that affect the future price of natural gas and to lay out clearly and candidly the range of possible price outcomes that could occur depending upon how these uncertainties are resolved.
At least thus far, however, EIA has steadfastly refused to undertake this task – at least for purposes of its Annual Energy Outlook, the official Agency forecast relied upon by many policy-makers at the federal and state level, and the only forecast ever reviewed by many in the private sector.
The result has been a disaster – i.e., an annual price forecast that ultimately misleads more than informs. The examples listed above helps to illustrate this point:
1. Assumption of “no change” in environmental requirements. Perhaps the single most important flaw in EIA’s annual forecasts is that it does not explicitly present or evaluate the potential impact on demand for natural gas or on natural gases of potential changes in environmental requirements. (At the request of certain members of Congress, EIA has periodically performed analyzes of the potential impact of certain legislative proposals pending before Congress. While these analyzes are available on EIA’s web site, many members of the public are not aware they exist. EIA does not discuss these studies in its annual forecasts or attempt to incorporate the results of these studies into the analysis its presents in its Annual Energy Outlook.)
The end result is that EIA’s forecast is highly vulnerable to being misinterpreted by policy-makers, and runs the risk of giving highly misleading signals to both energy producers and end use customers regarding potential future price levels.
EIA’s long-term forecast covers a 25-year period. Implicitly, therefore, EIA’s forecast is based on the assumption that there will be no changes in relevant environmental requirements that will significantly affect the use of natural gas or other fuels over the next 25 years.
It is possible this assumption will prove to be valid, but it obviously is not the only plausible outcome. Instead, changes already are started to be adopted at the state level that could result in huge increases in demand for natural gas, and it clearly possible that other such changes will be considered at the state level or the federal level or both over the next several years.
Some of these proposals could increase the amount of natural gas needed by the U.S. economy by the 2020 to 2025 time frame by as much as 2 to 4 Tcf per year, with far-reaching potential repercussions for the price of natural gas and electricity in the U.S. market.
There is a critical need, therefore, for EIA to explicitly address the potential impact of these requirements on future demand for natural gas and future price levels as part of its annual forecast – both to better inform decision-makers of the potential consequences of their decisions, and to inform end users of the potential price risks down the road.
Major decisions are being made every month regarding long-term investments in energy supply infrastructure and in buildings and equipment across the entire economy tied to the future cost of energy.
There is no hope that these decisions can be made intelligently if they are made in a complete vacuum regarding the range of possible future energy price scenarios. Right now, however, that is exactly what is occurring – with potentially disastrous consequences.
2. “Normal” weather assumption. Historically, for purposes of its natural gas price forecasts, EIA – like many other forecasters -- has prepared its forecasts based upon an assumption of “typical” weather. To take this approach, however, reflects a profound misunderstanding of how the natural gas market functions in the U.S. and the risks facing end use customers – virtually guaranteeing that EIA’s price forecasts, on average, will significantly understate the price end use customers pay for natural gas.
Why? Because natural gas prices are extremely sensitive to weather, but the impact of weather on natural gas prices is not symmetrical. When extreme mild weather conditions occur – as occurred last winter – the market finds ways of reducing supply (e.g., shutting in production, burning natural gas in lieu of oil-fired or coal-fired generation). As a result, price declines tend to be temporary and modest. By contrast, when weather-related demand is high – as may well occur this winter – the market’s ability to adjust is more limited. Prices could spike to very high levels – and potentially stay at high levels for several annual cycles, unless there is intervening mild weather (as we also may learn soon).
If we have learned anything in the past few years, therefore, it should be that events that put pressure on the system can lead to steep and last price spikes that can lead to huge cost increases.
By its choice of a simplifying methodology, however, EIA has simply factored these events out of the equation. In effect, it’s put itself in the same position as FEMA – planning for the possibility of Category 3 Hurricane striking New Orleans, but willfully ignoring the potential consequences of a Category 4 or Category 5 events, even though it’s these events that will cause the greatest harm.
This doesn’t mean, of course, that natural gas price forecasts should be developed on the assumption that weather conditions always will be extreme – or that the extremes will necessarily be unfavorable. (To the contrary, it is only a successive of three very mild winters, back-to-back-to-back that has kept natural gas prices from exploding to even higher levels over the past 3 years.)
It does mean, however, that: (i) potential natural gas demand and potential future prices cannot be properly assessed without looking at a wide range of scenarios (both favorable and unfavorable): and (ii) any such examination is likely to show the potential risk of much higher prices, potentially over sustained, several year periods than EIA projects using its current methodology.
3. Assumption of “normal” availability of non-gas fired generating capacity. EIA then compounds its error with respect to the weather by assuming the same level of hydro availability that will occur in climatologically normal years, and assuming high availability for nuclear generating capacity. In the past decade, however, “normal” hydro availability has been rare (this year is the exception); using this assumption, therefore, significantly biases EIA’s forecast. Further, while we certainly can hope for continued strong (i.e., in recent years, near perfect) performance from this nation’s nuclear fleet, as existing plants continue to age, this assumption becomes increasingly problematic, and the potential for at least some early retirements (not factored into EIA’s analysis) becomes an increasingly significant risk.
These are huge factors, which during the next decade could easily add 500 Bcf/year to power sector demand for natural gas in some years.
4. Restriction to examining only a limited range of oil price scenarios/failure to consider link between oil prices and availability of LNG. EIA further compounds the errors in its analysis by examining only a narrow range of potential future oil price scenarios. There is a direct link, however, between higher oil prices and increased demand for natural gas, and potentially between high oil prices and reduced availability of LNG. Failing to consider these factors, therefore, seriously distorts EIA’s analysis.
Cumulatively, the affect of these factors is to vastly understate the potential future natural gas requirements of the U.S. economy. Even in a normal weather scenario, by 2025, the total amount of natural gas required to meet the needs of the U.S. economy would be likely to increase by an additional 2 to 5 Tcf/year:
Potential EIA Underestimate of Future Natural Gas Needed by U.S. Economy
In years in which weather conditions are unfavorable, or in scenarios in which nuclear generation declines, or any of a number of EIA’s other assumptions prove to be invalid, this shortfall could easily increase by still another 2 to 4 Tcf/year.
EIA’s estimate of the amount of natural gas that is likely to be needed by the U.S. economy in future years, therefore, is profoundly flawed. Unfortunately, these errors – as serious as they may be – are not necessarily the most serious deficiency in EIA’s assessment of future supply and demand. Instead, as we’ll discuss in Part III, if anything, the errors in EIA’s estimate of the supplies that are likely to be available to the U.S. market in future years are even more severe.
Get your event listing in front of more than 100,000 industry professionals by posting on EnergyCentral's Event Center.