EIA Logo

Short-Term Energy Outlook

Coal Model Description


Contents

Related Coal Model Documentation


Introduction

The Energy Information Administration (EIA) of the U.S. Energy Department (DOE) developed the Short-Term Integrated Forecasting (STIFS) model to generate short-term (up to 8 quarters), monthly forecasts of U.S. supplies, demands, imports, stocks, and prices of various forms of energy. The purpose of this report is to define the coal model in STIFS and describe its basic properties. This report documents the October 1998 version of the coal equations in STIFS.

This report is written for persons who want to know how short-term energy markets forecasts are produced by EIA. The report is intended as a reference document for model analysts, users, and the public.


Overview

This section discusses the methodology for forecasting coal consumption published in the Outlook. The STIFS model determines total coal consumption (CLTCPUS) as the total demand for four major sectors: electric utilities (CLEUPUS); coke plants (CLKCPUS); retail and general industry (CLZCPUS); and coal consumed by independent power producers (CLIPPUS).

CLTCPUS = CLEUPUS + CLKCPUS + CLZCPUS + CLIPPUS


Coal Equations


Coal Production

The forecasts of coal production (CLPRPUS), along with the other supply elements that appear in STIFS: imports (CLIMPUS), exports (CLEXPUS), waste coal supplied (CLWCPUS), and producer and distributor stocks CLSDPUS), are provided by the Energy Information Administration's Office of Coal, Nuclear, Electric and Alternative Fuels, Coal Division, Data Analysis and Forecasting Branch. The coal production equation (CLPRPUSY) is provided to give the analyst an idea of supply requirements needed by the demand forecast, and to provide an interim (before the final exogenous forecast is provided) balanced coal forecast. Coal production is modeled as a function of a time trend, and a dummy variabe for the strike that occurred from April 1981 - May 1981 and monthly dummy variables:

CLPRPUSY = CLPR_01
+ CLPR_02 * TIME
+ CLPR_18 * STRIKE2
+ monthly dummy variables
[Click here for Regression Results]

The demand-implied coal production (CLPRPUSX) is calculated as total coal consumption (CLTCPUS) less the amount of waste coal supplied (CLWCPUS) plus the changes in producer stocks (CLSDPUS) and consumer stocks (CLSTPUS) plus exports (CLEXPUS) less imports (CLIMPUS):

CLPRPUSX = CLTCPUS - CLWCPUS + (CLSDPUS - LAG1(CLSDPUS))/ZSAJQUS + (CLSTPUS - LAG1(CLSTPUS))/ZSAJQUS + CLEXPUS - CLIMPUS

Then the difference (CLAJPUSX) between the forecasted production (CLPRPUSY) and the demand-implied production (CLPRPUSX) is calculated:

CLAJPUSX = CLPRPUSY - CLPRPUSX

The difference (AJDIFF) between this number and the forecasted coal balance discrepancy (CLAJTAR) is calculated. Please note that in earlier versions of STIFS a coal balance discrepancy was provided as an exogenous forecast, but the model currently assumes no inbalance between supply and demand (i.e. CLAJTAR = 0) in the forecast period:

AJDIFF = CLAJPUSX - CLAJTAR

Final coal production (CLPRPUS) is calucated by subtracting this difference from the forecasted coal production:

CLPRPUS = CLPRPUSY - AJDIFF


Coal Inventories

Electric Utility Coal Stocks

Electric utility stocks of coal (CLSEPUSX) is a function of elctric utility coal demand (CLEUPUS), a time trend and monthly dummy variables:

CLSEPUSX = CLSE_01
+ CLSE_02 * CLEUPUS
+ CLSE_18 * TIME
+ monthly dummy variables
[Click here for Regression Results]

A target stock level (CLSEPUSY) is derived from the electric utility coaL demand forecast (CLEUPUS) and exogenously specified target days-of-supply (CLDESTAR):

CLSEPUSY = CLEUPUS * CLDESTAR

Final electric utility stocks (CLSEPUS) is calculated as the minimum of the two:

CLSEPUS = MIN(CLSEPUSX,CLSEPUSY)


Retail and General Industry Coal Stocks

Retail and general industry stocks of coal (CLSOPUSX) is a function of retail and general industry coal demand (CLZCPUS), a time trend and monthly dummy variables:

CLSOPUSX = CLSO_01
+ CLSO_02 * CLZCPUS
+ CLSO_18 * TIME
+ monthly dummy variables
[Click here for Regression Results]

A target stock level (CLSOPUSY) is derived from the demand forecast (CLZCPUS) and exogenously specified target days- of-supply (CLDOSTAR):

CLSOPUSY = CLZCPUS * CLDOSTAR

Final retail and general industry stocks (CLSOPUS) is calculated as the minimum of the two:

CLSOPUS = MIN(CLSOPUSX,CLSOPUSY)


Coking Coal Stocks

Coking coal stocks (CLSKPUSX) is a function of coking coal demand (CLKCPUS), a time trend and monthly dummy variables:

CLSKPUSX = CLSO_01
+ CLSK_02 * CLKCPUS
+ CLSK_18 * TIME
+ monthly dummy variables
[Click here for Regression Results]

A target stock level (CLSKPUSY) is derived from the demand forecast (CLKCPUS) and exogenously specified target days- of-supply (CLDKSTAR):

CLSKPUSY = CLKCPUS * CLDKSTAR

Final retail and general industry stocks (CLSKPUS) is calculated as the minimum of the two:

CLSKPUS = MIN(CLSKPUSX,CLSKPUSY)

Tota secondary coal stocks (CLSTPUS) then represents the sum of stocks in each sector:

CLSTPUS = CLSEPUS + CLSOPUS + CLSKPUS


Retail and General Industry Coal Demand

Three equations are used to forecast coal consumption in the retail and general industry sector. Two equations forecast consumption in the residential and commercial sectors, and the third forecasts the consumption of coal in the industrial sector (excluding use at coke plants and synfuel plants). Coal used in the manufacture of synfuels (CLFCPUS) is assumed to remain constant at the level of 1.7 million short tons per quarter.

Residential Coal Demand

Residential coal consumption (CLRCPUS) is a small and relatively stable portion of total retail and general industry coal consumption. It is modeled as a function of U.S. population-weighted heating degree days (ZWHDPUS), time dummy variables and monthly dummy variables:

CLRCPUS = CLRC_01
+ CLRC_02 * D8401ON
+ CLRC_03 *TREND84
+ CLRC_04 * ZWHDPUS
+ monthly dummy variables

[Click here for Regression Results]


Commercial Coal Demand

Commercial coal consumption (CLCCPUS) is a small and relatively stable portion of total retail and general industry coal consumption. It is modeled as a function of U.S. population-weighted heating degree days (ZWHDPUS), time dummy variables and monthly dummy variables:

CLCCPUS = CLCC_01
+ CLCC_02 * D8401ON
+ CLCC_03 *TREND84
+ CLCC_04 * ZWHDPUS
+ monthly dummy variables
[Click here for Regression Results]

CLHCPUS = CLRCPUS + CLCCPUS


Other Industrial Coal Demand

Other industrial sector coal consumption net of synfuel consumption (CLXCPUS) is modeled as a function of the coal-weighted industrial production index (ZOSIIUS), time dummy variables and monthly dummy variables:

CLXCPUS = CLXC_01
+ CLXC_02 * D8401ON
+ CLXC_03 *TREND84
+ CLXC_04 * ZOSIIUS
+ monthly dummy variables
[Click here for Regression Results]

Total industrial sector coal demand (CLYCPUS) is then determined by the following identity:

CLYCPUS = CLFCPUS + CLXCPUS

Total retail and general industry coal consumption (CLZCPUS) is given by the identity:

CLZCPUS = CLYCPUS + CLHCPUS


Coal Consumed to Generate Electricity

Two of the four major sectors of total coal consumption (CLTCPUS) utilize the coal to generate electricity. Coal consumed to generate electricity (CLEUPUS and CLIPPUS) accounted for 89.7 percent of all coal used in the United States in 1997.

Electric Utility Coal Demand

The model for electric utility coal-fired generation (CLEOPUS) is discussed in the section on electric supply. To translate coal generation requrements into coal consumption, a simple equation is used. The equation does not assume strict proportionality between coal input an electricity output, but attempts to measure changes in coal consumption due to changes in electricity requirements. This equation allows for a trend and seaonality in the average net converson rate of coal to electric power. Seasonaliy arises from normal geographic shifts of electricity requirements form month to month form areas with newer, more efficient plants, or acess to higher quality coal, to areas with older, less efficient plants to access to lower quality fuel. Ambient atmospheric conditions (temperature extremes) may affect conversion loss rates, and these conditions can change significantly over the course of a year. Consumption of coal at electric utilities (CLEUPUS) is expressed as a function of coal-fired generation (CLEOPUS), a time trend, and monthly dummy variables:

CLEUPUS = CLEU_01
+ CLEU_02 * CLEOPUS
+ CLEU_18 * TIME
+ monthly dummy variables
[Click here for Regression Results]


Independent Power Producer Coal Demand

The forecast for coal consumed by independent power producers (CLIPPUS) is provided by the Energy Information Administration's Office of Coal, Nuclear, Electric and Alternative Fuels, Coal Division, Data Analysis and Forecasting Branch.


Coking Coal Demand

Coking coal is used in the manufacture of coke, which fuels blast furnaces that produce molten iron for the production of steel. Thus, coking coal demand is derived from the demand for steel. Coke is only used in steel plants that employ basic oxygen (open-hearth) furnaces. Determining the domestic demand for coking coal requires estimates of how much of coke demand will be satisfied from coke production, coke stocks, and net coke imports.

Coke demand (CCTCPUS) is derived from a forecast of total raw steel production (RSPRPUS) and the ratio of coke consumption to coke-based raw steel production (K1):

CCTCPUSX = K1 * (RSPRPUS - RSELPUS);

where,

K1 = K1_01
+ K1_02 * TIME
+ monthly dummy variables
[Click here for Regression Results]

Net imports of coal coke (CCNIPUS) are derived as a fraction of total raw steel production (RSPRPUS), the ratio of coke bet imports to coke consumption (K2), the ratio of coke consumption to coke-based raw steel production (K1), and the ratio of coke-based raw steel production to total raw steel production (K5):

CCNIPUS = K1 * K2 * K5 * RSPRPUS

where,

K2 = K2_01
+ K2_02 * TIME
+ monthly dummy variables
[Click here for Regression Results]

and,

K5 = K3_01
+ K5_02 * TIME
+ monthly dummy variables
[Click here for Regression Results]

Coke stocks (end of period) are derived as a fixed fraction of coke consumption:

CCSDPUS = (CCTCPUSX * K3 * 3.0) * ZSAJQUS

where,

K3 = K3_01
+ K3_02 * TIME
+ monthly dummy variables
[Click here for Regression Results]

Total coke production is detrmined by the identity:

CCPRPUS = CLKCPUS / K4

where,

K4 = K4_01
+ K4_02 * TIME
+ monthly dummy variables
[Click here for Regression Results]

The above relationships are combined into one reduced-form equation for coking coal demand:

CLKCPUSX = K1 * (1-K2+K3) * K4 * K5 * RSPRPUS - RSPRPUS(t-1) * K1 * K3 * K4 * K5


The only remaining task for forecasting coking coal consumption is to provide an estimate of total raw steel production (RSPRPUS). Seasonally adjusted raw steel production is estimated to be a function of the change in manufacturing inventories (KRDRXUS), real fixed investment (I87RXUS), and a time trend:

RSPRPUSA = RSPR_01
+ RSPR_02 * KRDRXUS
+ RSPR_03 * I87RXUS
+ RSPR_04 * TIME
[Click here for Regression Results]

And reasonalizing:

RSPRPUS = RSPRPUSA * RSPRPUSS

Non coked-based raw steel production is calculated by the following identity:

RSELPUS = (1 - K5) * RSPRPUS

Coke-based raw steel production is then calculated by:

RSBFPUS = RSPRPUS - RSELPUS

To ensure that the estimated value of coking coal consumption does not exceed a maximum monthly consumption capacity of 3.16667 million short tons, the following adjustment mechanism is used:

CLKCPUS = MIN[CLKCPUSX, (3.16667/ZSAJQUS)]

COKEBAL = CCTCPUSX - CCPRPUS - CCNIPUS + [CCSDPUS - CCSDPUS(t-1)] / ZSAJQUS

where,

COKEBAL = Difference between coke supply and demand.

Then calculate final coke consumption as:

CCTCPUS = CCTCPUSX - COKEBAL


EIA Home Page

File last modified: October 4, 2001

Contact:
Elias Johnson
Email: elias.johnson@eia.doe.gov
Phone: (202) 586-7277
Fax: (202) 586-9753

This page's URL: http://www.eia.doe/emeu/steo/pub/document/textcoal.html

If you are having technical problems with this site, please contact the EIA Webmaster at wmaster@eia.doe.gov.