The Relationship Between
Natural Gas and Electricity Prices
By John Herbert
John H. Herbert is adjunct professor of statistics at Virginia Polytechnic Institute (“Virginia Tech”) & State University Graduate
Center in Falls Church, Virginia. He also teaches statistics in Johns Hopkins University MBA program. He has published one book
and more than 50 refereed journal articles. After 18 years service as senior economist at the Energy Information Administration,
John took an early retirement in late 1997. John authored or co-authored more than 50 articles and reports on every aspect of the
natural gas industry while at EIA. In January 1998, John set up a private practice to include market analysis, price risk management
and arbitrage analyses of both financial and physical assets, statistical and econometric analyses, investment analyses, expert
testimony and workshops.
Natural gas prices do not drive electricity prices in the West. This study of 52 weeks of pricing data reveals that
the wholesale price of electricity and the wholesale price of natural gas appear to fluctuate completely
independently of one another. This conclusion holds true whether the daily, weekly, or monthly markets are
examined, and whether one examines the direction of the price change or the magnitude of the price change. Nor,
over the last 12 months, do electricity prices lead natural gas prices or vice versa. Finally, regardless of the specific
power or natural gas Hub studied in California, it continues to hold true that little correlation exists between natural
gas and power prices.
This study examines natural gas pricing data in West Texas, the major natural gas market serving California, and
power pricing data for Palo Verde, Arizona, a major wholesale power market serving California. The pricing data
was examined closely and analyzed statistically from April 2000 to April 2001 to determine whether natural gas
and power prices:
? Increase at the same time
? Move together in the same direction by day, week, or month
? Fluctuate to the same degree of magnitude
? Lead or lag each other.
Like all North American natural gas markets, the wholesale or spot price of natural gas in West Texas rose in the
Year 2000, as did power prices in the West. Observing these phenomena, some have concluded that high natural
gas prices cause high power prices. This study examines whether pricing data can support such a statement.
Something Happened—April 26, 2000
Increases in the price for natural gas and power began a year ago on April 26, 2000. On that day, a Wednesday,
wholesale power prices in the Palo Verde market area nearly doubled from their level at the beginning of the week.
Yet, wholesale natural gas prices in West Texas remained practically unchanged. Refer to Table 1.
Power Prices Increase While Gas Prices Unchanged
April 24, 2000
April 26, 2000
Natural Gas Price --West
Power Price -- Palo Verde,
Source: Gas Daily and Megawatt Daily
On Monday, April 24, 2000, the cost of natural gas in West Texas was $3.00/MMBtu. Wholesale power prices at
Palo Verde were $34.83/Mwh. By April 26, 2000, wholesale power prices in the Palo Verde market were
$67.79/Mwh, a 95 percent increase from April 24, 2000. Still the natural gas price in West Texas was a bit less at
$2.97/MMBtu. Based on daily data, electricity prices rose substantially with no increase in natural gas prices.
A correlation in price also cannot be found when examining weekly and monthly data. For the week ending April
21, 2000, the weekly wholesale power price at Palo Verde was $34.36/Mwh, while wholesale natural gas prices in
West Texas were $3.01/MMBtu. Two weeks later for the week ending May 5, 2000, average weekly wholesale
power was $80.15/Mwh. On that same date, weekly wholesale natural gas prices were $2.99/MMBtu, slightly
lower than their level two weeks earlier. In fact, in this example there is an inverse relationship between gas and
Using monthly averages, wholesale power prices increased from $34.74/Mwh to $73.62/Mwh from the 30-day
period before April 25, 2000, to the 30-day period after that date. During the same time, wholesale natural gas
prices in West Texas only increased from $2.88/MMBtu to $3.27/MMBtu. In other words, while both prices
increased when examining monthly averages, the increase for natural gas was a small percentage of the power price
Thus, examination of daily, weekly, and monthly prices indicates that power prices started to increase prior to
natural gas prices at the beginning of the structural change in both markets in late April 2000. Based on our
analysis, it is evident that natural gas price increases did not cause power prices to increase. On the contrary, to the
extent there is any causation at all, it appears that the increase in power prices may have kick-started the increase in
natural gas prices.
No Relationship between Gas and Power Prices:
April 2000 to April 2001
Figure 1 shows the price of natural gas versus the price of electricity for the 12-month period starting April 2000.
This figure reveals the stark independence between natural gas and electric prices as the markets for these two
commodities diverge. Whatever relationship might have existed prior to April 2000 evaporated after that time.
Figure 1. After April 25, 2000, No Relationship Between
Power and Natural Gas Prices
January 4, 1999 to April 25, 2000
April 26, 2000 to April 20, 2001
5 6 0 . 0 0
April 25 2000
4 8 0 . 0 0
4 0 0 . 0 0
3 2 0 . 0 0
2 4 0 . 0 0
Natural Gas Prices
1 6 0 . 0 0
8 0 . 0 0
0 . 0 0
1 / 4 /
/ 4 /
/ 9 /
/ 9 /
6 / / 9
7 / / 9
8 / / 9
9 / / 9
1 / /
2 / /
5/ / 0
3 / / 0
4 / / 0
5 / / 0
6 / / 0
8 / / 0
/ 5 /
1 0/ 5
1 / / 0
2 / / 0
3 / / 0
4 / / 0
1 8 / 0
Source: Gas Daily, Megawatt Daily
Although the analysis in Figure 1 seems clear, applying statistical techniques yields further confirmation that there
is no support for assuming a relationship between changes in prices in the two markets in the last year. Our ability
to predict power price changes from natural gas price changes is no better than our ability to predict the outcome of
a flip of a fair coin on the next toss given the current toss. This lack of a relationship holds true whether examining
monthly, weekly, or daily data.
When we examine whether monthly increases or decreases in wholesale power prices are coupled with monthly
increases or decreases, respectively, in natural gas prices for the 12 months starting in April 2000, we find the
following: in 5 of the 12 months an increase in power prices was coupled with an increase in natural gas prices.
But in 7 of the months either an increase in power prices was coupled with a decline in gas prices or a decline in
power prices was coupled with an increase in gas prices. In other words, about half the time an increase in natural
gas prices occurred with an increase in power prices, and about half the time the direction of the price change was
opposite for power and natural gas.
Weekly data suggest much the same result except they are even more striking. When we look at 52 weeks of
trading, starting on April 25, 2000, we find that power prices and natural gas prices move in the same direction,
either up or down, about half the trading weeks. We also find that an increase or decrease in natural gas prices is
followed by an increase or decrease, respectively, in power prices in a subsequent week about half the weekly time
When we examine two measures of the correlation coefficient1 -- which is the most common measure of
association used in research -- for daily price changes in power and natural gas for the selected series, we find
corroborating evidence that there is not a relationship between the two energy price series. The correlation
numbers are very low. Indeed, they are very near zero, which is as low as they can be expected to be.
For the correlation between the magnitude of the change in prices between natural gas and electricity, we obtain a
value of 0.19. For the rank correlation coefficient, which measures whether largest (and smallest) price changes are
related, we obtain an even smaller value 0.09. Thus, we conclude from these results that there is little association,
yet alone a causal relationship, between changes in price on wholesale natural gas and power
Overall changes in power markets appear independent of price changes in natural gas markets during the last year.
This appears true except at the very beginning of the price rise when power prices appear to have preceded the
increase in natural gas prices. This study does not attempt to outline the factors that contributed to these changes.
All that can be said is whatever that “something” is, we see little evidence it was an increase in natural gas prices.
At the California Border
The analysis undertaken for this paper focused on the West Texas natural gas producing region. However, even
when we compare natural gas prices at the California border, there is little relationship between those prices and
electricity prices much of the time. Indeed, a straightforward examination of data performed elsewhere shows that
when wholesale power prices began to rise in May 2000, natural gas prices rose, but not nearly to the extent power
prices rose (see Figure 2).3
1 The correlation coefficient as it is used here measures whether two prices move together. If they move together perfectly, the
coefficient will be 1. If the prices move completely independent of each other, the coefficient is zero. If the prices move in precisely
the opposite direction (one increases while the other decreases), the coefficient is -1. All correlation coefficients fall between -1 and 1.
2 For there to be any justification in saying that there was a strong relationship between these two price series, the correlation
coefficient would have to be much closer to 1.00, say 0.9 or higher. That is why the correlation coefficients we find of 0.09 and 0.19
are strong indicators of no relationship at all.
3 For a discussion of natural gas price in California markets and Figure 2, see John H. Herbert, "The Gas-Fired Future: Boom or
Bust?" Public Utilities Fortnightly, April 1 2001, pg. 26-27.
Figure 2. Major Wholesale Power Markets Serving California and a
Wholesale California Natural Gas Market are not Well Connected Much
of the Time
SoCal Gas large package
Source: Gas Daily, Megawatt Daily
A Closer Look at October to December 2000
When we look selectively and more closely at the prices in Figure 1, we can see one period where changes in
natural gas prices appear to be related to changes in power prices. For seven of the eight weeks between October
20, 2000, and December 15, 2000, we find that changes in power prices were coupled with like changes in natural
gas prices. But even here the relationship is very weak in terms of the magnitude of change for both sources of
energy (see Figure 3).
Figure 3. Magnitude of Electricity Price Changes
Unrelated to Natural Gas Price Changes
After the week ending December 15 2000 the average weekly
power price declined to $286.00/Mwh
Changes (Palo Verde)
10-Nov 17-Nov 24-Nov
Figure 3 shows that while the two prices series tend to move in the same direction, the magnitude of each change is
unrelated. For the most part, the changes in electricity prices are much larger than changes in natural gas prices.
We can conclude from this analysis that while natural gas certainly is a factor in the cost of electricity, it is often a
relatively small factor in the overall market price of electricity.
Nor does it matter whether we consider other producing basins. In terms of the magnitude of the change, results
are roughly similar when we examine price changes for the other producing markets shipping natural gas to
California, including the Western Canadian Sedimentary Basin, the San Juan Basin and the several basins in the
Rockies near southwestern Wyoming.4 Power price changes are much larger than natural gas price changes.
Results are similar even when we examine the price of natural gas within California such as prices for the
“Southern California Gas large gas packages” market5 and a related wholesale power price series.6 Equally
important, the finding of a weak correlation holds true even when looking at other natural gas hubs.7 The pattern
of prices over time is similar to the pattern of prices in West Texas and the other producing markets. Yet the
natural gas prices are higher in California than in West Texas especially in the November and December time
frame. These high gas prices are likely a consequence of transportation and other marketing issues, given that this
statistical study does not support the contention that the high electricity prices are due to an increase in the cost of
the natural gas commodity in major producing markets.
Note that the lack of a relationship between natural gas prices and power prices holds true regardless of the
efficiency of the natural gas turbines. Natural gas is the most likely fuel to have been used as the incremental or
marginal fuel for producing power for sale onto the California Power Exchange market. Often the marginal units
are inefficient units that are brought on line last because they are the most expensive to run. Thus, the amount of
Btus required to generate a megawatt-hour of power is much greater than the level required for more efficient units.
For instance, it would not be surprising if inefficient natural gas units had a heat rate of 16,000 Btu/Mwh instead of
the 8,000 Btu/Mwh or less generally used as an indication of the average heat rate of new combustion turbines.8
The effect of a natural gas price increase on the cost of power generation would be twice as much for a gas turbine
with a 16,000 Btu/Mwh heat rate than for one with an 8,000 Btu/Mwh heat rate. Using this example, when natural
gas reached $10/MMBtu in late December 2000, the cost of power from these inefficient natural gas generators
would have been at least $160/Mwh. Thus, the cost of gas in these units contributed to a higher wholesale cost of
power. But the commodity cost of natural gas, while high, was only a contributing, not a causal, factor.
No matter how one looks at the price data available for April 2000 to April 2001, no consistent relationship can be
found between changes in the price of natural gas--either in the producing region or at the California border--and
changes in the price of electricity to the California market. In general, the prices do not move together at all.
When, for short periods of time, there does seem to be a relationship in the movement of the prices, there is no
relationship between the magnitudes of the changes. The electricity price swings are much greater than the natural
gas swings. The inescapable conclusion is that while natural gas costs are certainly a factor in the price of
electricity, they have not been a major contributing factor impacting the price of electric generation in California.
4 For a brief discussion of power and natural gas prices in these other markets and graphical displays of related price information see
John H. Herbert “High Palo prices stretch spark spread in San Juan Basin,” Megawatt Daily, September 19, 2000 and “Mid-C’s high
power prices give big boost to gas plants,” Megawatt Daily, October 10, 2000.
5 The Southern California Gas large gas packages market is the largest volume wholesale natural gas market near the border in
6 Herbert, Public Utilities Fortnightly, April 1 2001, op. cit.
7 Specifically, the market areas of El Paso-non bonded, Kern River/Opal Plant, and Nova AECO-NIT. Herbert, Gas Daily, California
Gas and Power Prices: Joined at the Hip?, May 2, 2001, p. 1.
8 The heat rate refers to the efficiency of converting from natural gas to electricity. The lower the heat rate, the higher the efficiency.
A unit with a heat rate of 8000/Mwh is twice as efficient as a unit with a heat rate of 16,000/Mwh.