EMPIRICAL TESTS OF MAY SPOT
PRICES
© Copyright 2004 by Michael A. S. Guth. All Rights Reserved. No portion of this site, including the contents of this web page may be copied, retransmitted, reposted, duplicated, or otherwise used without the express written permission of Dr. Michael Guth.
|
|
MICHAEL A. S. GUTH, Ph.D., J.D.
|
Financial Economics Homepage || Attorney at Law Homepage
Empirical
Tests of May Power Spot Prices
Despite NYMEX’s recent decision to delist its power futures contracts, power trading continues in and around these former futures hubs. The delisting of the futures contracts has made price discovery more challenging. But brokers and commodity data services are trying to fill the void by showing forward trading bid-ask spreads good for 50 MW - 100 MW trades. By the end of March, power traders will be focusing their attention on the forward prices for power in May. Traders, and trading floor managers, will ask whether the May 2002 forward bid-ask spreads are above or below the expected spot prices for various regions.
Ideally, the expected spot market price forecast should be determined by various pieces of forward-looking information. For example, power marketing firms might have regional price forecasting models that they calibrated to particular regions where they trade. Analysts would provide inputs to these models on forward fuel prices and other measurable forward variables, and then the model would yield an estimate for the equilibrium price of electricity in the May 2002 spot market.
In practice, firms do not have access to all the forward variables they desire. Consequently, power marketing firms often use historical averages of past daily spot market prices as a first approximation to the expected spot price. This technique works reasonably well as long as you believe electricity consumption and supply in May 2002 will not be radically different from May consumption and supply trends in prior years. That assumption seems relatively innocuous. The following tables show the minimum, maximum, and average power prices in May for select eastern hubs over the period 1997 – 2001. The source for all the data shown in the tables and graphs in this article is McGraw-Hill’s MegaWatt Daily database.
|
Cinergy Prices $/MWh |
|||
|
|
Min of
Low |
Max of
High |
Avg of
Wtd Avg |
|
1997 |
14 |
20.25 |
16.77333333 |
|
1998 |
16.88 |
325 |
42.1045 |
|
1999 |
15 |
32.5 |
24.0045 |
|
2000 |
12 |
175 |
40.13956522 |
|
2001 |
5 |
45.25 |
33.21869565 |
|
5 Year |
5 |
325 |
31.41738318 |
|
Variance |
|
|
477.7050922 |
For all four regions, the five-year historical average price is below the average spot prices that prevailed in 2000 and 2001. Absent some prediction for unusually hot temperatures to the contrary, spot market prices should be lower in 2002 than they were in 2001, due to the onslaught of combined-cycle generating capacity coming on-line. Using historical averages as an indicator, the Cinergy May 2002 daily power spot prices should average around $31/MWh.
|
Entergy Prices $/MWh |
|||
|
|
Min of
Low |
Max of
High |
Avg of
Wtd Avg |
|
1997 |
12.13 |
22.5 |
15.83818182 |
|
1998 |
15.88 |
200 |
42.39 |
|
1999 |
20 |
32.5 |
24.533 |
|
2000 |
29 |
175 |
54.95 |
|
2001 |
17.5 |
46 |
39.015 |
|
5 Year |
12.13 |
200 |
35.41632075 |
|
Variance |
|
|
507.7673006 |
Entergy’s prices will likely be about $4/MWh higher than Cinergy’s prices in May 2002. Cinergy’s power generating capacity relies heavily on coal, whereas Entergy’s fuel mix has a much larger share for natural gas and comparatively little coal. Entergy’s five-year variance of power spot market prices is higher than Cinergy’s variance, which reflects the fact that the underlying volatility of natural gas prices has exceeded the volatility of coal prices.
|
PJM Prices $/MWh |
|||
|
|
Min of
Low |
Max of
High |
Avg of
Wtd Avg |
|
1997 |
15.38 |
22.35 |
18.57238095 |
|
1998 |
20.25 |
68 |
30.402 |
|
1999 |
21.25 |
34.25 |
28.15809524 |
|
2000 |
22.25 |
225 |
49.68318182 |
|
2001 |
11.5 |
45.38 |
40.17956522 |
|
5 Year |
11.5 |
225 |
33.70598131 |
|
Variance |
|
|
414.6173156 |
PJM’s five-year average price rests, not surprisingly, between the five-year average prices for Cinergy and Entergy. If historical averages are a reasonable approximation for the future, then PJM’s average price in May 2002 should be in the $33/MWh - $34/MWh range.
|
TVA Prices $/MWh |
|||
|
|
Min of
Low |
Max of
High |
Avg of
Wtd Avg |
|
1997 |
13 |
20 |
15.27230769 |
|
1998 |
16.63 |
185 |
37.356 |
|
1999 |
15 |
34 |
23.882 |
|
2000 |
21 |
125 |
40.9426087 |
|
2001 |
8 |
32.5 |
34.35136364 |
|
5 year |
8 |
185 |
31.84397959 |
|
Variance |
|
|
314.8210627 |
TVA was never a NYMEX futures hub. The TVA futures contract was sponsored by the Chicago Mercantile Exchange, which is a part of the Chicago Board of Trade. On first impression from inspecting the above table, TVA prices for May 2002 would seem to average about $32/MWh. However, the price relationship between Cinergy and TVA continues to evolve. In much of 2001, the price for power at TVA was $1 or $2 above the Cinergy price. Now TVA power is at or below the Cinergy price. Therefore, for May 2002, I would supplement the historical average with additional insights about market dynamics, and conclude that TVA’s May 2002 spot market price should average at or below Cinergy’s $31/MWh.
Readers of The Risk Desk may recall the January 2002 issue contained an article in which Gloria Zhang and I presented some statistical tests of the correct distributions governing electricity spot prices. The Inverse Gauss, Loglogistic, and Pearson 5 distributions were the best fitting distributions as ranked by the Anderson-Darling and Komolgorov-Smirnoff test statistics. The data in the January article comprised July-August and January-February power prices at the eastern hubs.
It comes as a pleasant surprise to find that for a non-peak consumption month like May, these same three distributions rank at the top of a list of the best-fitted distributions. This somewhat surprising result means that electricity prices - finally - seem to be settling down into a stable set of distributions, which analysts can use for developing strategies and determining appropriate risk control measures.
CINERGY MAY DAILY SPOT
PRICES 1997 - 2001
Cinergy May power spot prices best fit the LogLogistic distribution, which is skewed to the right and has longer tails than the normal or lognormal distributions. We ranked the possible distributions using the Komolgorov-Smirnoff test statistic, because it does a better job of ranking distributions clustered about a mean than the Anderson-Darling statistic, which is better-suited to characterizing the tails of alternative distributions. Power spot prices in PJM tell much the same story.
PJM MAY DAILY SPOT PRICES 1997 - 2001
Although they are not displayed here in the interests of brevity, the May power spot prices for Entergy best fit the Pearson 5 distribution. The May power spot prices for TVA best fit the Inverse Gauss distribution. The next step is to take this information about the best fitted empirical distributions and form 90% and 95% confidence intervals around the mean, and also to form one-sided 90% and 95% confidence intervals about the price level that the average will be above or below with either 90% or 95% confidence. These tasks are left to the proverbial interested reader, or a firm can contact me about performing this analysis for trading strategies on a single term or ongoing basis.
© Copyright 2004 by Michael A. S. Guth. All Rights Reserved. No portion of this site, including the contents of this web page may be copied, retransmitted, reposted, duplicated, or otherwise used without the express written permission of Dr. Michael Guth. Reprinted from TheDesk (March 15, 2002) with permission of the publisher, Scudder Publishing Group, LLC. www.scudderpublishing.com.