In the STIFS model, nonutility petroleum products consist of the following: motor gasoline, jet fuel, nonutility distillate fuel oil, nonutility residual fuel oil, liquefied petroleum gases (LPG's), and other (minor) petroleum products. The major determinants of demand for these products are: transportation activity, economic activity (i.e. gross domestic product, transportation activity, manufacturing output, etc.), prices and weather. Most of the estimating relationships incorporate monthly seasonal dummies and dummy variables (Dxxxx) to capture one-time events or conditions.
Utility demand for distillate and residual fuel oil is derived separately through simulation of the electricity model (see Electricity Supply and Demand section).
Two estimating relationships define the motor gasoline demand model. They are: motor gasoline demand (MGTCPUS) and highway travel activity
(MVVMPUS). The first equation requires projected highway travel data from the second one, which, in turn, is determined by variables exogenous to
the model (retail gasoline prices and disposable income*time). These estimating relationships are presented below:
|
[Click here for Regression Results] |
|
[Click here for Regression Results] |
The two main determinants of the motor gasoline demand estimating relationship are highway travel and inflation-adjusted average retail motor gasoline prices (MGEIAUS/CICPIUS). The dummy variables, D8302, D8412, and D9401, refer to months of severe weather that affected motor gasoline deliveries. The dummy variables, DRVP90 and D91ON, refer to modifications of RVP standards previously implemented. D95ON reflects the implementation of reformulated gasoline regulations starting in 1995.
The second equation, highway travel activity, is also one of the components in the first equation. The main arguments are real disposable income and inflation-adjusted cost per mile (MOGP) with a lag of 12 months and a polynomial degree of 2. The dummy variables, D8501, D9401 and D9412 pertain to weather-related disruptions in travel, and the variable D89ON relates to a change in the reporting methodology on vehicle miles traveled.
The jet fuel demand model is characterized by four linked behavioral relationships. They are: jet fuel demand (JFTCPUS), industry-wide capacity utilization (RMZZPUS), total available capacity (RMZTPUS), and consumer ticket prices (ACTKFUS). For modeling purposes, kerosene jet fuel and naphtha jet fuel are regarded as one product as a result of the phase-out of naphtha jet fuel use by the military. The estimating relationships are presented below:
|
[Click here for Regression Results] |
|
[Click here for Regression Results] |
|
[Click here for Regression Results] |
|
[Click here for Regression Results] |
The first relationship shows that changes in jet fuel demand are related to changes in air travel capacity and winter weather in the Northeast, and a dummy variable for December, 1989, to capture the additional demand brought about by the beginning of the Gulf War conflict. Total air travel (utilization) reflects the joint influences of real disposable income and industrial production. Air travel capacity is shown to respond to shifts in utilization over a period of 12 months. The time variable is a proxy for long-term trend in load factors. Consumer ticket prices are shown to respond to shifts in jet fuel prices and a time trend to reflect the impact of increased concentration in air travel markets.
This product is modeled as four separate sectoral linear equations. The sectors are: (1) transportation, (2) residential, (3) commercial, and (4) industrial and other nonutility consumers. Each equation comprises a lagged variable of sectoral demand, monthly seasonal dummy variables, and an adjustment factor that shifts projected values according to regression errors during the estimation interval.
Transportation demand (DFACPUS), depicted below, is a function of total industrial production (ZOTOIUS) and the inflation-adjusted producer price of diesel fuel (DSRTUUS/WPCPIUS):
|
[Click here for Regression Results] |
Residential demand (DFRCPUS) and commercial demand (DFCCPUS) have the same model structure. They are each determined by daily average population-weighted heating-degree-days in the Northeast (ZWHDDNO/ZSAJQUS) from October through April:
|
[Click here for Regression Results] |
|
[Click here for Regression Results] |
The industrial and other sector (DFICPUSX) is modeled as a function of an adjustment factor (NGINPUSX - NGINPUS) that defines curtailments of natural gas deliveries to the industrial sector during peak periods (see Natural Gas Supply and Demand section). The factor (0.362*1.030/5.825) is the share (0.362) of total estimated switchable gas capacity in the industrial manufacturing sector (converted to million barrels per day) thought to be dedicated primarily to distillate fuel as an alternate fuel.(1)
Thus, any reduction in industrial gas demand due to gas supply constraints are assumed to be made up by other fuels, with 36.2 percent (by Btu content) coming from distillate fuel oil.
|
[Click here for Regression Results] |
Total nonutility demand (DFNUPUS) for distillate fuel is the sum of the four sectoral demands:
Sales of no. 2 heating oil residential volume (D2RCPUS) and no. 2 diesel fuel through company-owned outlets (DSTCPUS) are also estimated for reporting purposes only. Residential volume is related to the combined total of residential and commercial deliveries (DFHCPUS); diesel volumes are a function of deliveries to the transportation sector and a time trend.
|
[Click here for Regression Results] |
|
[Click here for Regression Results] |
As with distillate fuel oil, nonutility heavy fuel oil product supplied (RFNUPUS) is modeled sectorally with linear functions. The following equations define the transportation, commercial and industrial sectors, respectively:
|
[Click here for Regression Results] |
|
[Click here for Regression Results] |
|
[Click here for Regression Results] |
Transportation demand responds primarily to growth in gross domestic product (GDPQXUS) and commercial demand is related to fluctuations in the weather during the heating season (ZWHDDUS/ ZSAJQUS) * (OCT...APR). Demand in the industrial sector is a function of average daily heating degree-day deviations from the norm during the winter months, the relative price of retail residual fuel oil (RFTCUUS*RFTCZPUS) to industrial natural gas (NGICUUS/NGNUKUS), and the proxy for curtailments of natural gas deliveries (NGINPUSX - NGINPUS). The factor (0.321*1.030/6.287) is the share (0.321) of total estimated switchable gas capacity in the industrial manufacturing sector (converted to million barrels per day) thought to be dedicated primarily to residual fuel as an alternate fuel. Thus, any reduction in industrial gas demand due to gas supply constraints are assumed to be made up by other fuels, with 32.1 percent (by Btu content) coming from residual fuel oil.
Total nonutility residual fuel oil demand (RFNUPUS) is the sum of the three sectoral demands:
Liquefied petroleum gases (LGTCPUS) product supplied is disaggregated into ethane (ETTCPUS) and LPG's heavier than ethane (LXTCPUS):
LGTCPUS = ETTCPUS + LXTCPUS
The demand for deseasonalized ethane (ETTCPUSA) is estimated using the following linear equation:
|
[Click here for Regression Results] |
The relative price of wholesale distillate to the price for natural gas to electric utilities (D2WHUUSA/ NGEUDUSA) and petrochemical output (ZO28IUS) are the principal drivers in the equation. The ethane equation is reseasonalized after the initial demand estimation:
Demand for LPG's excluding ethane (LXTCPUS) is a linear function of the previous month's demand, day-weighted heating degree-days' deviations from normal (ZWHDPUS-ZWHNPUS)/ZSAJQUS, the index of chemical production (ZO28IUS), and an adjustment factor (NGINPUSX-NGINPUS) that defines interruptions of natural gas deliveries to industrial customers. The factor (0.273*1.030/3.836) is the share (0.273) of total estimated switchable gas capacity in the industrial manufacturing sector (converted to million barrels per day) thought to be dedicated primarily to LPG's as an alternate fuel. Thus, any reduction in industrial gas demand due to gas supply constraints is assumed to be made up by other fuels, with 27.3 percent (by Btu content) coming from liquefied petroleum gas.
|
[Click here for Regression Results] |
Subsequent reseasonalization yields the non-seasonally adjusted demand for this product group, as shown below:
Although propane (PRTCPUS) is part of the LPG product group (LXTCPUS), the STIFS model includes a separate estimating relationship for propane. This variable is used in the estimation of propane retail sales volume in the petroleum supply portion of the STIFS model. Because propane constitutes the bulk of the LXTCPUS group, LXTCPUS is the principal explanatory variable in the equation:
|
[Click here for Regression Results] |
File last modified: October 28, 1998
This page's URL: http://www.eia.doe/emeu/steo/pub/document/textpt.html