US Heading For Merchant Plant Overdevelopment
by Christopher Seiple and Dr. Arnold Leitner, with RDI Consulting in Boulder, Colorado
Power shortages and price spikes in many areas of the United States this summer have put the electric power industry in the public spotlight and sparked demands for reregulation of the industry. However, our research indicates that these price spikes and shortages will be temporary and that as early as next year some regions could be facing a glut of oversupply in the wholesale electricity market.
This glut is likely to begin just after developers and financiers ante up more than $30 billion in investment in new generating capacity.
Critical to success in a commodity market, such as electricity generation, is knowing when to buy, sell, or build generating assets. Academic research indicates that firms that are best able to drive their strategy by understanding cyclical trends are able to increase their return on investment by 3 to 4%.
No one can precisely predict boom and bust cycles in the future. However, a carefully structured analysis of the supply and demand balance that identifies the key sources of uncertainties and uses quantitative tools to assess the impact of these uncertainties provides a strong framework within which to develop a corporate strategy. To provide such an analysis, RDI employed a probabilistic model based on decision tree theory. This model incorporates uncertainty by applying probabilities to possible events and analyzing these events in hundreds of possible scenarios.
Our research indicates that four primary factors contribute to cyclical pricing trends in commodity industries.
Lumpy Capacity Additions
One of these factors is the lumpiness of capacity additions in relation to a commodity’s demand growth. This typically occurs in new or small industries where there are large economies of scale associated with capacity additions in relation to overall demand growth. For instance, demand for a new product may be increasing at 30% per year, but the construction of one new manufacturing facility may double manufacturing supply. Such conditions would likely create oversupply conditions for this product.
RDI believes that this factor could influence generation markets that are small in size due to a lack of transmission interconnections to neighboring regions. For instance, in eastern New York – an area where capacity is currently scarce – the peak demand is approximately 10,000 megawatts. Due to weak transmission interconnections, eastern New York is relatively isolated from the rest of the New York electricity grid. PG&E Generating is currently pursuing development of a 1,000 megawatt power plant in the region. Such a plant would increase overall supply by more than 10% in a market that is growing at a rate of less than 2% a year. Such a large capacity addition could meet future demand growth for as much as the next five years, causing a prolonged bust period in electricity prices.
Another factor contributing to cyclical pricing trends is supply-side uncertainty. In some markets, the industry as a whole may be unaware of how much new supply is in the pipeline. Thus, companies acting independently pursue new capacity development that results in oversupply for the market as a whole.
Many conditions point to supply-side uncertainty in electricity markets:
- The long lead time of new power plant development – up to three years – and the uncertainty associated with the likelihood of individual projects going forward.
- Increases in capacity at existing units are occurring without public announcements.
- The development of unanticipated distributed generation could add to existing supply.
- Improvements in plant performance, combined with additional interruptible demand, could reduce the amount of reserve capacity required to provide the same level of reliability.
Our research indicates that development of new power plants is currently the key driver of potential market downturns in electricity. The table on the previous page provides RDI’s most recent projections of new capacity additions. This table includes only plants that have begun operating, are under construction, or in the advanced stages of development. Developers have proposed a total of more than 290,000 megawatts of new capacity.
Availability of Capital
A third factor contributing to cyclical pricing trends is the availability of capital. In general, companies tend to invest only when returns are high and funds are available either internally or from capital markets. As a result, too much capacity is typically added at the top of a cycle and too little capacity is added at the bottom of the cycle. In most commodity industries, this is the primary driver of boom-and-bust cycles.
In electricity markets it is clear that substantial amounts of capital are currently available for investment. Electricity marketers, such as PECO Energy, Williams and Coral, have played a large role in supporting the availability of capital due to their willingness to sign 20- to 30-year power purchase agreements that limit risk for the developer and for banks financing the project. The substantial cash flow of utilities – especially those securitizing stranded costs – has also contributed to capital availability. Finally, the general fondness the stock market has shown for companies like Calpine and AES is a sign that capital markets are willing to make substantial amounts of capital available for merchant developers.
Incorrect Demand Forecasts
The final factor driving cyclical pricing trends is that producers planning new capacity forecast demand incorrectly. Incorrect demand forecasts have played a substantial role in contributing to the current price spikes of the market. In some regions of the country, electricity demand recently increased by more than 4% annually – substantially higher than was anticipated by most forecasters.
There is ample room for error when trying to predict future demand because of the many variables that must be taken into account. For instance, incorrect forecasts of gross domestic product can contribute to incorrect demand forecasts. Other variables that we considered in our analysis include price elasticities, the feasibility of developing dispatchable demand, the impact of computers on electricity demand, and the weather.
The following table shows our predictions of which regions of the country will be in boom portions of the cycle and which regions will be in bust portions of the cycle in the years 2000, 2001, and 2002. This is based on the results from our probabilistic boom-bust model using decision tree theory and taking into account all of the factors discussed earlier. Bust regions are assumed to have at least 5% more capacity than needed, and boom regions are assumed to have at least 1% less capacity than needed.
Our analysis has led us to a number of other conclusions.
This year, most of the country will either be in a boom portion of the cycle or at least close to market equilibrium levels in which prices are high enough to support new capacity development. With almost 30,000 megawatts of new capacity coming on line, we expect 2000 to be the year in which supply catches up with demand.
Due primarily to new capacity development, it is extremely likely that many regions of the country will enter bust portions of the cycle next summer. In the space of just two years – 2000 and 2001 – a minimum of 60,000 megawatts of new capacity will come on line. It is likely that total capacity additions by the end of next year will reach 75,000 megawatts. Total capacity additions during all of the 1990s were only slightly higher than 75,000 megawatts. In Texas and the northeast, we expect the market will have at least 20% more capacity than is required. Almost all of this capacity is already under construction. Only retirement of substantial amounts of capacity in these regions could provide price recovery.
By 2002, nearly all large markets in the US will be in the bust portion of the cycle. SERC is the only large market we predict may be at equilibrium levels. However, we believe this finding must be heeded with a bit of caution in that surplus capacity in ECAR/MAIN and SPP could potentially depress pricing in SERC as well. In smaller regions such as MAPP – where we predict equilibrium conditions – it would only take one or two large projects to move the market into oversupply conditions.
The most attractive regions for new development efforts include the southeast and mid-Atlantic. Florida is another attractive region, but the political climate is currently stymieing the efforts of developers to build new plants.
Doom and Gloom Scenario
Based on the insights gained from the model, RDI identified a longer-term scenario that could potentially create a prolonged period of low prices and low returns for generators.
The first requirement for this scenario actually to occur is that electricity markets must be deregulated. That is, generators must be subjected to the disciplining force of market prices and consumers – or at a minimum marketers serving consumers – must be exposed to the volatility of these same prices. Price caps, standard offer rates, and partial deregulation in only a few states would impede the development of this scenario.
Because this scenario is driven by the imposition of supply and demand economics on the electric business, we refer to it as the economic rationalization scenario.
In this doom and gloom scenario, the imposition of supply and demand economics creates the following impacts. First, prior to 2003, generators build new capacity to meet expected demand as is occurring now. Next, persistent price spikes cause some level of dynamic demand to develop so that peak firm demand is reduced by 5% from expected levels between 2003 and 2008. Next, producers, trying to improve profitability, increase availability factors from an average of 82% to 88% between 2003 and 2008. Finally, generators are able to increase the capacity of their existing facilities by 1% per year between 2003 and 2008.
To consider the implications of this scenario, RDI used its electric simulation model to forecast future electricity prices in the midwestern US. In our base case, prices are at relatively high levels today due to shortages of capacity and high turbine prices. By 2002, prices reach long-run equilibrium levels and stay at that level over the forecast horizon. However, in the economic rationalization scenario, the combination of factors described above leads to substantial oversupply conditions for the duration of the forecast horizon. Prices are approximately 20% lower than in the base-case forecast.
This oversupply occurs for several reasons. First, the development of dynamic demand causes modest reductions in firm demand. Second, generators are able to produce more capacity from the existing system and improvements in availability factors also create more capacity. Third, the development of dynamic demand in combination with more reliable generators results in the market being able to provide the same level of reliability to customers with less capacity. Thus, overall target reserve margins are reduced.
It is difficult to assess the likelihood of this scenario actually occurring, but we believe it is an important scenario to watch for. Early signs of development would include the development of the infrastructure to facilitate dispatchable demand, continued price spikes, and improvements in plant performance.
Developers, power marketers and capital markets have responded to power shortage conditions and are rapidly building new plants that will provide customers with a reliable supply of electricity. New power plants are getting built in markets with regulated reserve requirements — like NEPOOL — and in markets with no reserve requirements — like Texas and western states. They are getting built in regions with independent system operators, or “ISOs,” and in regions without ISOs. New power plants are even getting built in markets with significant regulatory risk — like California — or significant permitting and environmental hurdles — like the northeast. Our analysis indicates that developers should worry more about their investment returns than regulators should worry that plants will not get built in a deregulated market. Policymakers just need to ensure that the power plant development process is as easy, quick and fair as possible.
Finally, even though we expect most regions of the country will soon head into a period of low electricity prices, someday in the not too distant future, boom conditions will again return to the marketplace. Our analysis indicates that slight changes in the supply-demand balance can cause large changes in electricity prices. Markets with a 2% capacity shortfall have experienced significant price spikes, but regions with a 2% surplus have experienced very low electricity prices. There is one unknown that could reduce the threat of extreme price spikes — if customers begin to develop demand that can be curtailed during peak hours, price spikes could be diminished. Development of such demand should therefore be an important policy imperative.