The blending of oxygenates, such as fuel ethanol and methyl tertiary butyl ether (MTBE), into motor gasoline has increased dramatically in the last few years because of the oxygenated and reformulated gasoline programs.(1) Because of the significant role oxygenates now have in petroleum product markets, the Short-Term Integrated Forecasting System (STIFS) was revised to include supply and demand balances for fuel ethanol and MTBE. The STIFS model is used for producing forecasts in the Short-Term Energy Outlook. A review of the historical data sources and forecasting methodology for oxygenate production, imports, inventories, and demand is presented in this report.
Fuel ethanol and MTBE usage has grown steadily since the early 1980's in response to octane demand resulting initially from the phaseout of lead from gasoline and later from rising demand for premium gasoline. Federal and local tax incentives for blending renewable fuels into motor gasoline have contributed to the growth in demand for fuel ethanol.(2) The oxygenated and reformulated gasoline programs stimulated a dramatic increase in oxygenate demand and production capacity between 1991 and 1995 (Table 1). Oxygenates now account for over 4 percent of the finished motor gasoline pool.
[Click here for monthly ethanol and MTBE production data in Excel .XLS file, 268 KB]
Table 1. U.S. Oxygenate Capacity and Production (Thousand barrels per calendar day) | ||||
Oxygenate Production Capacity | ||||
Ethanol | MTBE | TAME | ETBE | |
January 1, 1991 | 82.6 | 122.5 | 0.5 | 0.0 |
January 1, 1992 | 93.5 | 135.1 | 3.7 | 0.0 |
January 1, 1993 | 90.1 | 170.2 | 5.0 | 10.3 |
January 1, 1994 | 90.7 | 223.2 | 14.5 | 0.8 |
January 1, 1995 | 103.6 | 250.9 | 18.1 | 4.0 |
Annual Average Production | ||||
Ethanol | MTBE | |||
1990 | 49 | 84 | ||
1991 | 56 | 101 | ||
1992 | 70 | 101 | ||
1993 | 75 | 136 | ||
1994 | 83 | 144 | ||
1995 | 88 | 163 | ||
1996 | 63 | 185 | ||
1997 | 83 | 198 | ||
Notes:
| ||||
Sources:
|
The recent growth in oxygenate blending into motor gasoline has been demand-driven because of the minimum oxygen content mandates in the oxygenated and reformulated gasoline programs. Oxygenated gasoline must contain a minimum 2.7 percent oxygen by weight while reformulated gasoline requires a minimum 2.0 percent oxygen by weight. Supply and demand projections for fuel ethanol and MTBE begin with estimates of total oxygenate demand.
Total oxygenate demand forecasts are based on estimated market shares for the following types of motor gasoline:
Estimates of market shares for the regulated gasolines (oxygenated motor gasoline, RFG, and OPRG) generally begin with estimates of the fraction of the U.S. population that reside in each control area that require one of the regulated gasolines. Calculated population shares must then be corrected to arrive at motor gasoline demand shares because per capita gasoline demand varies throughout the country. For example, the District of Columbia contains 0.20 percent of the U.S. population but represents only 0.15 percent of the U.S. retail motor gasoline market. Wyoming, on the other hand, has a retail gasoline market share that is over 1.5 times its population share. Population share correction factors were estimated for each State using motor gasoline demand shares calculated from Federal Highway Administration (FHWA) 1995 motor gasoline sales data.(3) Finally, estimated control area motor gasoline demand shares are then adjusted for spill over (delivery of regulated motor gasoline to areas that do not require it).(4) Table 2 provides a sample of this estimation method for RFG in mandated areas.
[Click here for population correction factors in Excel .XLS file, 268 KB]
Table 2. Estimated RFG Demand Share - Mandated RFG Areas, January 1998 | |||||
Control Area | Estimated Population 7/1/96 (thousands) |
U.S. Population Share | Population Correction Factor | Estimated Gasoline Demand Share | |
Baltimore, MD | 2,436 | 0.0092 | 0.942 | 0.0087 | |
Chicago-Gary-Lake Co., IL-IN-WI | 8,275 | 0.0312 | 0.889 | 0.0277 | |
Hartford, CT | 1,862 | 0.0070 | 0.865 | 0.0061 | |
Houston-Galveston-Brazoria, TX | 4,253 | 0.0160 | 1.037 | 0.0166 | |
Los Angeles-Anaheim-Riverside, CA | 15,495 | 0.0584 | 0.912 | 0.0533 | |
Milwaukee-Racine, WI | 1,784 | 0.0067 | 1.004 | 0.0068 | |
New York City, NY-NJ-CT | 18,373 | 0.0693 | 0.779 | 0.0540 | |
Philadelphia, PA-NJ-DE-MD | 6,092 | 0.0230 | 0.903 | 0.0207 | |
Sacramento, CA | 2,073 | 0.0078 | 0.912 | 0.0071 | |
San Diego, CA | 2,655 | 0.0100 | 0.912 | 0.0091 | |
Total unadjusted share of total U.S. motor gasoline retail market | 0.210 | ||||
Correction for spill over | x 1.02 | ||||
Total adjusted share of total U.S. motor gasoline retail market | 0.214 | ||||
Notes:
|
|||||
Sources:
|
Regulated gasoline demand shares are converted to monthly volumes using the forecast of refinery output of motor gasoline generated by the STIFS model.(5) Because there is about a one month lag between production of finished motor gasoline at refineries and retail sale, demand shares are lagged one month to convert them to production shares. In other words, while 11.8 percent of the motor gasoline sold at retail outlets in January 1998 should be Oxygenated or OPRG, refineries are expected to produce these grades of regulated gasolines in December 1997.
Because fuel ethanol and MTBE have different oxygen contents, volumetric oxygenate demands are usually presented on an MTBE-equivalent basis. About 2 gallons of MTBE have the same oxygen content as 1 gallon of ethanol.(6) Oxygenated motor gasoline and OPRG are assumed to contain 15.2 percent MTBE by volume, and RFG is assumed to require 11.7 percent MTBE by volume.(7) Given estimates of total refinery production of motor gasoline, regulated gasoline production shares, and required oxygenate content, oxygenate demand for blending into regulated motor gasoline can be derived.
Continued demand for ethanol in gasohol blending, and demand for MTBE as an octane blendstock in conventional gasoline, is added to the demand for oxygenates in regulated motor gasolines to arrive at total oxygenate demand. A simple forecast of continued demand for oxygenates in gasohol and octane blending of an average 70 thousand barrels per day MTBE-equivalent volume during the summer months and an average 40 thousand barrels per day during the winter is assumed.
The sum of oxygenate demand for regulated motor gasolines, gasohol, and octane blending equals total oxygenate demand:
OZTCPUS = [0.152 * (OXFRAC + OPFRAC) + 0.117 * RFFRAC] * MGROPUS + OZTCPAD
Total oxygenate demand is then disaggregated into ethanol and MTBE (and other ethers) demands based on the assumption that ethanol demand is supply-driven and that MTBE satisfies the remaining demand.
The STIFS fuel ethanol balance involves the following 5 variables:
EOTCPUS = EOPRPUS + EONIPUS - d(EOPSPUS)
Most fuel ethanol blending into motor gasoline takes place at terminals and racks (often referred to as "splash" blending) that are not included in EIA surveys. Fuel ethanol splash blending is classified as field production in the EIA Petroleum Supply Monthly and is obtained from the following identity:
EOFPPUS = EOTCPUS - Refinery Inputs of Fuel Ethanol
Refinery inputs of fuel ethanol are not explicitly identified in EIA publications. However, total field production of fuel ethanol can be calculated from the Petroleum Supply Monthly using the identity:
EOFPPUS = MBFPPUS + MGFPPUS
[Click here for monthly ethanol data in Excel .XLS file, 268 KB]
Table 3. Fuel Ethanol Annual Statistics All units in thousand barrels per day, except stocks are million barrels | |||||||||
Production | Exports | Imports | End-of-Year Stocks |
Demand | Imputed Demand | ||||
Source: | ATF | EIA | ATF | ATF | EIA | EIA | FHWA | ATF | EIA |
1997 | - | 83 | - | - | 0 | 2,889 | - | - | 81 |
1996 | - | 63 | - | - | 1 | 2,065 | 70 | - | 64 |
1995 | - | 88 | - | - | 1 | 2,186 | 79 | - | 90 |
1994 | - | 83 | - | - | 1 | 2,393 | 68 | - | 83 |
1993 | - | 75 | - | - | 1 | 2,114 | 64 | - | 75 |
1992 | 62 | 70 | 0.4 | 2.4 | - | 1,791 | 58 | 64 | 68 |
1991 | 55 | - | 2.8 | 0.8 | - | - | 56 | 53 | - |
1990 | 70 | - | 8.6 | 1.0 | - | - | 49 | 62 | - |
1989 | 46 | - | 0.7 | 1.5 | - | - | 45 | 47 | - |
1988 | 51 | - | 0.8 | 0.6 | - | - | 53 | 51 | - |
1987 | 49 | - | 1.2 | 1.3 | - | - | 52 | 49 | - |
1986 | 60 | - | 0.2 | 6.5 | - | - | 51 | 66 | - |
1985 | 48 | - | 0.2 | 5.2 | - | - | 51 | 53 | - |
Notes: | FHWA fuel ethanol demand based on estimate of ethanol contained in gasohol sales reported by States
| ||||||||
Sources: | ATF | Bureau of Alcohol, Tobacco, and Firearms, Monthly Distilled Spirits Report, Report Symbol 76 and Alcohol Fuels Report, internal quarterly report. | |||||||
EIA | Production and 1992 stocks: Energy Information Administration, EIA-819M Monthly Oxygenate Telephone Report, Table B2
Imports and stocks: Energy Information Administration, Petroleum Supply Annual, Tables 20 and 30, respectively. | ||||||||
FHWA | Federal Highway Administration, Highway Statistics Summary to 1995, "Gasohol Sales by State, 1980-1992" (Table MF-233GLA), and Estimated Use of Gasohol, 1993-1995 (Table MF-233E); Highway Statistics 1996, "Estimated Use of Gasohol" (Table MF-33E). |
However, because of the lack of monthly data for some of the series, some simplifying assumptions are made for the STIFS historical database:
The Short-Term Energy Outlook's forecast for ethanol production is assumed to remain flat at 90 thousand barrels per day over the forecast period. Net imports are assumed to be 0, and inventories are assumed to remain constant at the most recent level reported in the Petroleum Supply Monthly. Thus, fuel ethanol demand for the forecast period equals fuel ethanol production. Refinery inputs are assumed to average 30 thousand barrels per day, and field production 60 thousand barrels per day.
The STIFS balance for MTBE and other ethers involves the following 4 variables:
EIA began collecting MTBE data at the same time as the fuel ethanol data (Table 4). The MTBE data do not include a difference between plant production and field production as in the fuel ethanol balance.
[Click here for monthly MTBE data in Excel .XLS file, 268 KB]
Table 4. EIA MTBE Annual Statistics All units in thousand barrels per day, except stocks are million barrels | ||||||
1997 | 1996 | 1995 | 1994 | 1993 | 1992 | |
Plant Production | 198 | 185 | 163 | 142 | 136 | 101 |
Gross Imports | 59 | 48 | 45 | 39 | 20 | - |
Gross Exports | 12 | 12 | 7 | 11 | - | - |
Net Imports | 47 | 36 | 39 | 28 | - | - |
End-of-Year Stocks | 8,228 | 10,008 | 8,702 | 13,769 | 10,035 | 13,818 |
Imputed Demand | 249 | 218 | 215 | 160 | 161 | 106 |
Sources:
|
MTBE supply history before January 1992 is more scant than that for fuel ethanol. Monthly gross imports for 1992 and gross exports for 1992 and 1993 were obtained from the Oil Market Listener.(9) Estimates of annual average MTBE production, gross imports and gross exports for 1985 through 1991 were provided by DeWitt and Company, Inc. The annual average production data were disaggregated into monthly volumes by fitting a cubic spline curve. MTBE net imports are assumed to average 3 thousand barrels per day in 1991, and 0 for all months before January 1991. MTBE stocks are assumed to remain constant at 1 million barrels for all months before December 1990, and then increase steadily at about 1 million barrels per month through January 1992.
Historical MTBE demand is calculated from a material balance around production, net imports (imports - exports), and stock change:
MTTCPUS = MTPRPUS + MTNIPUS - d(MTPSPUS)
The MTBE demand forecast is derived from the difference between estimated total oxygenate demand and assumed ethanol demand (converted to MTBE-equivalent volume):
MTTCPUS = OZTCPUS - 2.0 * EOTCPUS
MTBE net imports are assumed to remain constant at 40 thousand barrels per day over the forecast period. An MTBE production forecast is assumed, and adjusted to produce a reasonable stock path. These exogenously specified production and stock numbers may be converted to estimated regression equations once a consistent production/stock history is accumulated.
The MTBE and fuel ethanol balances are converted into aggregates that correspond to volumes reported in the Petroleum Supply Monthly and the Short-Term Energy Outlook.
Field Production: Field production of other hydrocarbons/oxygenates is one of 6 categories of field production reported in the Petroleum Supply Monthly and modeled in STIFS (other categories include crude oil, pentanes plus, liquefied petroleum gases, motor gasoline blend components, and finished motor gasoline). All MTBE production and a small volume of ethanol blended into motor gasoline at refineries are included in the category field production of other hydrocarbons/oxygenates.
Field production of other hydrocarbons/oxygenates (OHRIPUS) is estimated in the STIFS model as a linear function of MTBE production (MTPRPUS), a dummy variable representing January 1993 (when new MTBE inventory survey data was incorporated into the Petroleum Supply Monthly), and monthly dummies:
|
[Click here for Regression Results, Table A1] |
[Click here for Graph of OHRIPUS] | |
[Monthly regression data in Excel .XLS file, 268 KB] |
Refinery Inputs: Six categories of refinery inputs are modeled in STIFS: crude oil, pentanes plus, liquefied petroleum gases, unfinished oils, aviation gas blending components, and "other" petroleum products. The "other" petroleum products category includes other hydrocarbons/oxygenates and motor gasoline blending components.
Refinery inputs of other petroleum products (PSRIPUS) is estimated as a function of MTBE demand (MTTCPUS), field production of motor gasoline blend components (MBFPPUS), a dummy variable to capture the months Oct. 1996 and Dec. 1996 through May 1997 when imports and refinery inputs were unusually high, and monthly dummies:
|
[Click here for Regression Results, Table A2] |
[Click here for Graph of PSRIPUS] | |
[Monthly regression data in Excel .XLS file, 268 KB] |
The STIFS model also includes an estimating equation for the subcategory refinery inputs of other hydrocarbons/oxygenates (OXRIPUS), which is estimated as a function of MTBE demand (MTTCPUS) and monthly dummies:
|
[Click here for Regression Results, Table A3] |
[Click here for Graph of OXRIPUS] | |
[Monthly regression data in Excel .XLS file, 268 KB] |
Inventories: Inventories of MTBE (MTPSPUS) and fuel ethanol (EOPSPUS) are aggregated into total oxygenate stocks (OXPSPUS). Also included in this category are stocks of other oxygenates such as methanol, tertiary butyl alcohol (TBA), and other ethers, which are assumed to remain constant at 700 thousand barrels through the forecast period:
OXPSPUS = EOPSPUS + MTPSPUS + 0.700
Total oxygenate stocks are then added to stocks of other hydrocarbons/hydrogen to yield total stocks of other hydrocarbons/hydrogen/oxygenates (OHPSPUS). Stocks of other hydrocarbons/hydrogen are assumed to remain constant at 50 thousand barrels through the forecast period:
OHPSPUS = OXPSPUS + 0.050
1. For reviews of the oxygenated and reformulated motor gasoline program requirements and oxygenate supply and demand issues refer to these other analyses published by the Energy Information Administration:
For State tax exemptions refer to Federal Highway Administration, Highway Statistics, Section I, Table MF-121T.
3. An equivalent description of this method is that the fraction of a State's population that resides in a control area is multiplied by the State's share of the U.S. retail gasoline market.
4. Earlier estimates of the correction factor included the expected effects of reduced automobile fuel efficiency and price elasticity of demand. These corrections are not required when gasoline demand shares are estimated using 1995 or later gasoline demand data, which should include these effects.
5. For analysis of forecasting model and regression results for refinery output of motor gasoline, refer to: Energy Information Administration, Short-Term Energy Outlook Model Documentation Report.
6. The volumetric ratio between MTBE and ethanol may vary by plusmn;0.05 depending on the assumed ethanol and MTBE product purities.
7. These percentages may change by as much as plusmn;0.5 percent absolute (i.e., MTBE in oxygenated gasoline may range from 14.7 to 15.7 volume percent) depending on the density of the motor gasoline, the purity of the oxygenate, and the assumed average oxygen content.
8. EIA reports only gross imports of fuel ethanol in the Petroleum Supply Monthly (Table 33). Net imports are assumed to equal gross imports in the STIFS historical database (i.e., gross exports are assumed to be zero).
9. Energy Information Ltd., "US MTBE Imports Remain Strong While Stocks Rebuild With End of Oxy Season," Oil Market Listener (San Francisco, CA, April 6, 1994) p. 2.
Table A1. Field Production Other Hydrocarbons/Oxygenates (million barrels per day)
Equation | DF Model | DF Error | SSE | MSE | Root MSE | R-Square | Adj R-Sq | Durbin-Watson |
OHRIPUS | 14 | 56 | 0.06820 | 0.0012178 | 0.03490 | 0.828 | 0.788 | 1.93 |
Parameter | Estimate | Approx. Std. Err. | 'T' Ratio | Independent Variable |
OHRI_B0 | -0.047963 | 0.02469 | -1.94 | Constant Coefficient |
OHRI_MT | 1.726869 | 0.12456 | 13.9 | MTPRPUS, MTBE production |
OHRI_D9301 | 0.320144 | 0.03833 | 8.35 | D9301 = 1 if January 1993, 0 otherwise |
OHRI_E1 | 0.024326 | 0.02216 | 1.10 | JAN = 1 if January, 0 otherwise |
OHRI_E2 | -0.002370 | 0.02119 | -0.11 | FEB = 1 if February, 0 otherwise |
OHRI_E3 | 0.000512 | 0.02124 | 0.02 | MAR = 1 if March, 0 otherwise |
OHRI_E4 | -0.007310 | 0.02114 | -0.35 | APR = 1 if April, 0 otherwise |
OHRI_E5 | -0.015904 | 0.02113 | -0.75 | MAY = 1 if May, 0 otherwise |
OHRI_E6 | -0.019539 | 0.02113 | -0.92 | JUN = 1 if June, 0 otherwise |
OHRI_E7 | 0.001694 | 0.02118 | 0.08 | JUL = 1 if July, 0 otherwise |
OHRI_E8 | -0.008982 | 0.02115 | -0.42 | AUG = 1 if August, 0 otherwise |
OHRI_E9 | -0.015690 | 0.02116 | -0.74 | SEP = 1 if September, 0 otherwise |
OHRI_E10 | -0.009161 | 0.02118 | -0.43 | OCT = 1 if October, 0 otherwise |
OHRI_E11 | -0.003124 | 0.02207 | -0.14 | NOV = 1 if November, 0 otherwise |
Method of estimation: Ordinary least squares Estimation period: January 1992 through October 1997 |
Table A2. Refinery Inputs "Other" Petroleum Liquids (million barrels per day)
Equation | DF Model | DF Error | SSE | MSE | Root MSE | R-Square | Adj R-Sq | Durbin-Watson |
PSRIPUS | 15 | 55 | 0.43809 | 0.0079653 | 0.08925 | 0.738 | 0.671 | 1.26 |
Parameter | Estimate | Approx. Std. Err. | 'T' Ratio | Independent Variable |
PSRI_BO | -0.116219 | 0.04749 | -2.45 | Constant Coefficient |
PSRI_MB | 0.417911 | 0.26202 | 1.59 | MBFPPUS, field production motor gasoline blend components |
PSRI_MT | 1.626040 | 0.18066 | 9.00 | MTTCPUS, MTBE demand |
PSRI_DY | 0.174293 | 0.03900 | 4.47 | DY9697 = 1 if Oct 1997 or Dec 1997 - Mar 1998 |
PSRI_E2 | 0.103094 | 0.05164 | 2.00 | FEB = 1 if February, 0 otherwise |
PSRI_E3 | 0.147015 | 0.05203 | 2.83 | MAR = 1 if March, 0 otherwise |
PSRI_E4 | 0.148479 | 0.05240 | 2.83 | APR = 1 if April, 0 otherwise |
PSRI_E5 | 0.115947 | 0.05318 | 2.18 | MAY = 1 if May, 0 otherwise |
PSRI_E6 | 0.131098 | 0.05390 | 2.43 | JUN = 1 if June, 0 otherwise |
PSRI_E7 | 0.162030 | 0.05457 | 2.97 | JUL = 1 if July, 0 otherwise |
PSRI_E8 | 0.177580 | 0.05590 | 3.18 | AUG = 1 if August, 0 otherwise |
PSRI_E9 | -0.009130 | 0.05426 | -0.17 | SEP = 1 if September, 0 otherwise |
PSRI_E10 | 0.048537 | 0.05275 | 0.92 | OCT = 1 if October, 0 otherwise |
PSRI_E11 | 0.053504 | 0.05634 | 0.95 | NOV = 1 if November, 0 otherwise |
PSRI_E12 | 0.025127 | 0.05656 | 0.44 | DEC = 1 if December, 0 otherwise |
Method of estimation: Ordinary least squares Estimation period: January 1992 through October 1997 |
Table A3. Refinery Inputs Oxygenates (million barrels per day)
Equation | DF Model | DF Error | SSE | MSE | Root MSE | R-Square | Adj R-Sq | Durbin-Watson |
OXRIPUS | 13 | 57 | 0.07464 | 0.0013095 | 0.03619 | 0.876 | 0.850 | 1.62 |
Parameter | Estimate | Approx. Std. Err. | 'T' Ratio | Independent Variable |
OXRI_B0 | 0.025888 | 0.01909 | 1.36 | Constant Coefficient |
OXRI_MT | 1.250655 | 0.06835 | 18.3 | MTTCPUS, MTBE demand |
OXRI_E2 | -0.011611 | 0.02090 | -0.56 | FEB = 1 if February, 0 otherwise |
OXRI_E3 | -0.008702 | 0.02098 | -0.41 | MAR = 1 if March, 0 otherwise |
OXRI_E4 | -0.011842 | 0.02091 | -0.57 | APR = 1 if April, 0 otherwise |
OXRI_E5 | -0.020854 | 0.02090 | -1.00 | MAY = 1 if May, 0 otherwise |
OXRI_E6 | -0.010917 | 0.02092 | -0.52 | JUN = 1 if June, 0 otherwise |
OXRI_E7 | 0.007517 | 0.02104 | 0.36 | JUL = 1 if July, 0 otherwise |
OXRI_E8 | -0.002976 | 0.02097 | -0.14 | AUG = 1 if August, 0 otherwise |
OXRI_E9 | -0.034592 | 0.02106 | -1.64 | SEP = 1 if September, 0 otherwise |
OXRI_E10 | -0.021436 | 0.02125 | -1.01 | OCT = 1 if October, 0 otherwise |
OXRI_E11 | -0.025004 | 0.02241 | -1.12 | NOV = 1 if November, 0 otherwise |
OXRI_E12 | -0.023332 | 0.02232 | -1.05 | DEC = 1 if December, 0 otherwise |
Method of estimation: Ordinary least squares Estimation period: January 1992 through October 1997 |
File last modified: March 6, 1998