Ethanol and gasoline prices, 2009 article

advertisement
Does Ethanol Matter to Gasoline Prices?
Dr. Thomas E. Elam, FarmEcon LLC, 9-22-08
A link between ethanol production and gasoline prices has been cited by the
government and ethanol support groups in defense of federal ethanol support.
Iowa State University researchers (Du and Hayes) have published a paper in
which they claimed to show increasing ethanol production has depressed U.S.
gasoline prices by $0.29 to $0.40 per gallon. The model relies on a complex set
of assumptions to show reduced refiner margins are causing from lower gasoline
prices.
If increasing ethanol production is reducing gasoline refining margins we would
expect to see gasoline prices decline relative to the price of crude oil. The chart
below shows clearly that this is not the case. In fact, the gasoline/crude oil price
spread has actually INCREASED as ethanol production has increased.
Monthly Gasoline/Crude Oil Price Spread and Ethanol Production
January 2000 – April 2008
20,000
Refinery Outages
18,000
70¢
MTBE
60¢
16,000
14,000
Katrina
50¢
12,000
40¢
10,000
30¢
8,000
6,000
20¢
4,000
10¢
2,000
-
Ja
n00
Ju
l-0
Ja 0
n01
Ju
l-0
Ja 1
n02
Ju
l-0
Ja 2
n03
Ju
l-0
Ja 3
n04
Ju
l-0
Ja 4
n05
Ju
l-0
Ja 5
n06
Ju
l-0
Ja 6
n07
Ju
l-0
Ja 7
n08
0¢
Gasoline/Crude Oil Price Spread
Monthly Ethanol Production
Gasoline/Crude Spread Trend
The trend in the gasoline/crude price spread is positive while ethanol production
is increasing.
Statistical Models:
Is there a relationship between gasoline prices or margins and ethanol
production? To test the effects of ethanol production two similar models were
constructed. The first is:
Ethanol Production, 1,000 Barrels
Cents/Gallon Difference to Crude Oil
80¢
PG = PC, EP, GI, Season, Supply Interruption
The second model looks at the 3:2:1 refinery crack spread. The spread is the
value of 2 gallons of gasoline plus 1 gallon of heating oil minus the value of 3
gallons of crude oil, all divided by 3 to get a value per gallon.
The second model is:
3:2:1 CS = PC, EP, RU, CI, GI, Season, Supply Interruption
Where:
PG = New York Harbor conventional regular gasoline, spot price (cents/gallon)
3:2:1 CS = value of 2 gallons of gasoline plus 1 gallon of heating oil minus the
value of 3 gallons of crude oil
PC = Cushing, OK West Texas Intermediate crude oil spot price (cents/gallon)
EP = Monthly fuel ethanol production
GI = U.S. motor gasoline ending inventory (million barrels)
Season = monthly seasonal price difference relative to December
Supply Interruption = statistical corrections for Katrina (Aug-Sep 2005), MTBE
(Apr-Aug 2006) and Refinery Outages (Mar-Jul 2007)
RU = Refinery utilization rate (%)
CI = Crude ending inventory, net of Strategic Petroleum Reserve (million barrels)
Monthly data from 1/1/2000 through 4/30/2008 were used. Data came from the
Department of Energy and are available at www.farmecon.com (Gasoline Price
Study link).
Results, First Model:
Results show that ethanol production is not a statistically significant factor in
gasoline prices. Although statistically not meaningful, results show a very weak
POSITIVE, not NEGATIVE relationship between gasoline prices ethanol
production.
Crude oil price is the major factor in gasoline prices. A 1 cent per gallon increase
in crude oil price means very close to a 1 cent per gallon increase in gasoline
price.
A one million barrel increase in gasoline inventories depresses gasoline prices by
about 0.568 cents per gallon.
Relative to December, prices April-June are higher - the summer season price
peak.
The three 2005-2007 supply interruptions caused price spikes that averaged 21.5
to 29.2 cents per gallon.
Table 1:
Regression of Ethanol Production, Oil Price, Gasoline Inventory, Season and
Supply Interruptions on NY Wholesale Regular Gasoline Price
Variable
Constant
Ethanol Production, 1,000 Barrels
Cushing, OK WTI Spot Price FOB (Cents per
Gallon)
U.S. Motor Gasoline Ending Inventory (Million
Barrels)
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Katrina Effect Sept-Oct 2005
MTBE Effect Apr-Aug 2006
2007 Refinery Outages Mar-Jul 2008
Coefficient
124.810*
0.00102
Standard
Error
22.945
0.0073
t Score
5.440
1.392
0.978*
0.051
19.176
-0.568*
5.659
4.974
2.901
7.363*
12.440*
7.863*
4.495
1.944
2.615
-6.589
-0.616
24.957*
21.547*
29.176*
0.112
3.487
3.487
3.412
3.465
3.596
3.643
3.590
3.693
3.604
3.697
3.494
5.450
3.683
4.091
-5.074
1.623
1.427
0.850
2.125
3.459
2.158
1.252
0.526
0.726
-1.782
-0.176
4.580
5.850
7.133
R-Squared = 98.79% (% of gasoline price variation explained)
*Statistically different from zero, 95% confidence
Results, Second Model:
Results show that ethanol production has no statistically significant effect on the
3:2:1 price spread, and thus no effect on gasoline prices. The second model is
very similar to the Iowa State study.
Table 2:
Regression of Ethanol Production, Oil Price, Refinery Utilization, Crude Oil
Inventory, Gasoline Inventory, Season and Supply Interruptions on 3:2:1 Spread
Variable
Constant
Ethanol Production, 1,000 Barrels
Cushing, OK WTI Spot Price FOB
(Cents/Gallon)
Refinery Utilization Rate (%)
U.S. Crude Oil Ending Inventory Excluding SPR
(Million Barrels)
U.S. Motor Gasoline Ending Inventory (Million
Barrels)
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Katrina Effect Sept-Oct 2005
MTBE Effect Apr-Aug 2006
2007 Refinery Outages Mar-Jul 2008
Coefficient
Standard
Error
t Score
19.97285
0.00031
30.81738
0.00048
0.64810
0.63536
0.08372*
0.72736*
0.03326
0.29642
2.51725
2.45383
0.06529*
0.03057
2.13551
-0.48959*
6.39954*
5.36580*
1.47060
1.93555
3.25994
-0.40094
-3.12204
-4.68482
-1.37292
-6.48020*
-2.21959
25.54203
12.48663
22.66495
0.07973
2.55920
2.62476
2.39818
2.36410
2.52700
2.58155
2.50774
2.61179
2.35307
2.47380
2.29635
4.17069
2.48704
2.85293
-6.14069
2.50060
2.04430
0.61321
0.81872
1.29004
-0.15531
-1.24496
-1.79372
-0.58346
-2.61953
-0.96657
6.12418
5.02067
7.94443
R-Squared = 85.95% (% of 3:2:1 price spread variation explained)
*Statistically different from zero, 95% confidence
As crude oil prices increase the spread increases. Some crude oil is used in
refining, so cost increases as crude oil price increases. As refinery utilization
rate increases so do refining margins. As crude oil inventories increase they
pressure crude oil prices downward, increasing the margin. As gasoline
inventories increase gasoline prices are pressured downward, causing the
refining margin to decline. All these variables are statistically significant.
Seasonal patterns show a strong margin in the January-February heating oil
peak season and a weaker margin late in the year as heating oil stocks are
increasing. The three supply disruptions had strong, significant, positive effects
on the margin. This second model reinforces the conclusion of no ethanol
production effect on gasoline prices.
Implication:
Increasing U.S. ethanol production is having no effect on U.S. gasoline prices or
refining margins. Crude oil prices, gasoline inventories, crude oil inventories,
refinery utilization rate, seasonal demand patterns and temporary gasoline
supply issues of 2005-2007 have been the drivers of gasoline prices and
margins.
Reference:
Xiaodong Du and Dermot J. Hayes. The Impact of Ethanol Production on U.S.
and Regional Gasoline Prices and on the Profitability of the U.S. Oil Refinery
Industry. Working Paper 08-WP 467. Center for Agricultural and Rural
Development. Iowa State University. April 2008
Download