To the rescue, hundreds of new generation facilities were announced across the
country. Plans to build our way to energy security were well under way. No one
questioned the cost of this generation, since these power plants were backed by
high-flying energy companies (with their high-flying stock valuations). Financial
firms lined up to invest and cash in on the largest technological investment in
decades.

June 2001 came, and the people of California (through conservation, good luck,
and a pittance of new generation) managed to survive the summer with no blackouts
– period. The rush to build was quickly reevaluated. Overnight, the anticipated
construction boom caused electricity prices to fall through the floor. This
Boom-Bust cycle was not limited to California alone, as Texas and certain parts
of the Midwest have experiences similar fluctuations as early as 1999.

Riding The Roller Coaster

As this roller coaster continues, is there a way to cut through the financial
and political hyperbole and answer the question: “Just how much generation will
be built?”

There is, and the method used is most commonly associated with the task of
placing a value on a company’s receivables. When a company issues credit to
a customer, it writes a receivable and has its accounting department age it
as it ferments in a file. Receivables can be grouped into different categories
as they age: 30-day, 60-day, 90-day and so on. Unlike a fine wine, receivables
do not get more valuable with age. This is because the probability lessens that
a receivable will actually be paid as it grows older. The trick in actually
finding the value of a receivable is to identify the probability it will progress
through the different stages of a receivable’s life, up to its ultimate state
– paid or written off. Once these probabilities are determined, a Markov Chain
Analysis can be used to determine the ultimate likelihood the receivable will
be paid or written off.

Proposed power plants face similar situations to receivables. A power plant
project lifecycle begins with a public announcement, and it is immediately placed
into an early development state. While in this state, developers, financiers
and engineers work to determine the viability of the site, fuel availability,
and a thousand other details. If the proposed site is successful, it will receive
permits, financing, local support, and so on. To the extent that the proposing
company will actually begin final design work and purchasing of equipment. At
this point the project has entered the advanced development state. However at
this stage, the firm could lose financing, the site deemed too environmentally
fragile for construction, or a thousand other details could present a high enough
hurdle to cause the project to be cancelled.

If the project is fortunate enough to survive the advanced development phase,
the time may come where a shovel will actually break ground. At this time, the
plant is now considered under construction. Once the plant is being built, it
is still not completely safe to assume it will be completed. On rare occasions,
a major party may lose interest or an organized effort can halt construction.
But if the project progresses, the plant could still be derated slightly, causing
the number of announced megawatts to be reduced.

As the chart above shows, the project lifecycle will have one of two ultimate
outcomes: commercial operation or cancellation. The path a project takes can
vary depending on many factors. A project may be announced, only to be delayed
for many months while permits are obtained, allowing it to proceed to the advanced
development state. But for the analysis to proceed, a consistent method to determine
the probability of plant progression is needed.

Data Handling

The method used to capture data on project progression for the Markovian Chain
Analysis is to review all announced plants across the country on a monthly basis.
Every plant’s current development status is compared to its previous status.
The process is simplified when we realize that there are a limited number of
state changes available to a plant. For example, if a plant was in the early
development state in the previous month, there are only three possible states
that plant could be in the current month. The plant could have proceeded to
advanced development, been cancelled or remain in early development.

Totaling the amount of proposed generation that has transitioned among the
various states of development, we can arrive at a frequency of generation advancement
by dividing the transition amounts by the total amount of generation in each
of the original states. To illustrate, suppose there were 10,000 megawatts in
early development last month and 1,000 of those megawatts transitioned to advanced
development. This indicates a 10 percent frequency, or probability, of monthly
transition from early to advanced development.

Once several months’ data has been analyzed, a series of monthly transitional
probabilities are developed. Since the transition of a single large plant could
cause an individual month to have a greatly distorted set of transitional probabilities,
several months’ probabilities should be averaged. It is recommended that at
least one year’s worth of transitions be used in this average, as projects tend
to be scheduled to achieve commercial operation just prior to the summer peak,
thereby introducing a seasonality component to the data.

The Markov Analysis

Once the averaged monthly transitional probabilities have been calculated,
a Markovian matrix can be built.

As this example shows, a project initially in the early development state stands
a three-percent chance of cancellation and a two-percent chance of advancing
by the end of the month. For this analysis it is further assumed that once a
plant has gone commercial or been cancelled, it will remain in that state forever.
Therefore, the transition probabilities for the commercial operation and cancelled
states are 100 percent.

Using matrix manipulation , a solution matrix is developed which indicates
the probability that a project currently in early development, advanced development
or under construction will ultimately be placed into operation or cancelled.

For the example given above, the solution given would say that 81 percent of
early development plants ultimately would be cancelled.

National Survey Results

After studying all United States plant construction projects since May 2000,
an analysis for the nation’s ability to convert an announcement into a megawatt
was performed. The total amount of generation in each of the three preliminary
states is multiplied by the success rates for plants in that phase of development.
The study results show that 75 percent of all megawatts in development today
will likely be cancelled at some point.

This study leads to the conclusion that the glut of power prophesied may not
materialize. This prediction is being born out by a raft of recent project cancellations,
as financiers look elsewhere to invest.

Conclusion

The large amount of generation announced in response to the California energy
situation could not be supported financially once credit to the electricity
sector began to tighten in the Third Quarter of 2001. As proposed power projects
quietly fall by the wayside, the potential for spot market price volatility
increases. As volatility increases and energy marketers garner profits from
power price swings, investors will once again be drawn to the market increasing
the amount of new generation.

In short, the boom and bust cycle in power plant construction appears to have
undergone a period of high amplitude and short cycle time. Whether this cycle
is attenuating or becoming more pronounced, only time will tell. This financial
cycle appears to be faster than the actual time required to develop and construct
a new power plant. The speed of the financial boom-bust cycle should keep the
physical boom-bust cycle from occurring.