Model structure for mid-term energy demand outlook
Primary energy demand consists of final energy demand and energy demand in the transformation sector. Forecasts are made by first breaking down final energy demand by different energy sources such as petroleum products, town gas, electricity, coal, and thermal and other energy.
Each energy source is broken down by purpose of use or demand sector, such as industry, transport, residential/commercial, and public/other. Consumption patterns and characteristics per source/sector are reflected when forecasting demand.
Model structure and methodology
1
[FigureⅡ-1] Outlook model structure
Use of econometric model for demand outlook by final energy source
A model by energy source/sector (use) is estimated by quarterly time-series data, after which input premise values (GDP, temperature variables, energy prices) are applied to forecast demand.
- The forecast results are tallied by energy source and sector to determine the total final energy forecast value.
Major explanatory variables that are used for mid-term metric model estimation and outlook are data on GDP, index of industrial product by industry type, energy prices per source/sector, and cooling degree days/heating degree days.
From among major explanatory variables, the premise values of the index of industrial product by industry type are determined within the model based on GDP.
ARDL (Autoregressive Distributed Lag) is used as a basic model for demand outlook by specific use.
The following method is adopted for forecasts in the transformation sector.
The fuel input amount is needed to produce secondary energy demand, such as electricity, town gas, and thermal energy that is forecasted in the area of final energy.
This fuel input amount is determined for each of the power generation, town gas production, and district heating thermal energy production sectors.
How forecasts are made on the amount of fuel input needed for electric power generation
- An outlook is made on the total electricity supply in consideration of total electricity demand, self-consumption, and power transmission/power distribution loss rate.
- The LP (Linear Programming) model is used to forecast the power generation amount by energy source required to satisfy total electricity supply.
- Generating efficiency forecasts are applied to the forecasted power generation amount by source to determine the fuel input amount.
- The “6th Electricity Supply and Demand Plan” is used as a major premise for forecasting energy demand in the power generation sector.
A similar method is employed to come up with forecasts on the fuel input amount in the
town gas and thermal energy production sector. The amount is determined based on reverse order of the ‘energy conversion process’.
Oil demand outlook method
Final energy consumption is categorized into three areas - transport, industry, and residential/commercial/public/other.
A forecast model is established by major product in each sector.
- Five products (gasoline, diesel, heavy oil, jet fuel, LPG) in the transport sector
- Six products (kerosene, diesel, heavy oil, LPG, naphtha, asphalt) in the industrial sector - Four products (kerosene, diesel, heavy oil, LPG) in the residential/commercial/public
and other sector
Major explanatory variables of each model include GDP (or index of industrial product), product prices, heating degree days, seasonal variables, and a lagged variable of actual consumption. Specific model settings are used for each product.
With regards to oil injected into the transformation sector (power generation, town gas production, thermal energy production), demand forecasts are determined for secondary energy sources (electricity, town gas, thermal energy), after which the input requirement is determined based on the transformation sector module.
- At this point, relations with other energy sources that can substitute for oil are simultaneously considered.
Electricity demand outlook method
Electricity demand is categorized into four sectors - for industrial use, for housing, for commercial/public use, and for transport.
Demand patterns and characteristics by sector are considered for individual model estimations, after which input premise values are used to forecast electricity demand for the forecast period.
In each model’s estimations, major explanatory variables that are used are quarterly GDP, index of industrial product, real power rates (unit cost of sales) by sector, and quarterly temperature information (cooling degree days and heating degree days).
- The index of industrial product is used as an explanatory variable instead of GDP to forecast electricity demand for industrial use.
LNG demand outlook method
For LNG demand forecasts, LNG demand is categorized into demand for town gas production and demand for power generation.
For forecasts on LNG demand for town gas production, town gas demand in the final sector is forecasted first.
- Town gas demand is categorized into different uses, such as for residential, general, and industrial use. An outlook is made for each use, using price, income, such temperature variables as CDD and HDD, and number of customers as variables from the supply perspective.
Next, an outlook is made on LNG demand for town gas production in consideration of the input ratio between LNG and LPG, which are raw materials for town gas production, as well as self-consumption and loss ratio.
LNG demand for power generation is determined through an LP model, which forecasts the amount of power generation by source in the power generation sector and the energy input amount by source.
Coal demand outlook method
For coal demand, a categorization is first made into anthracite and bituminous coal demand in the final consumption sector. Forecasts are made on demand by source and by use (industry, residential/commercial, and power generation), after which they are summed up. For coal demand for power generation, the coal input amount for power generation that is forecast in the transformation sector is used.
Anthracite demand is categorized into demand for residential/commercial use and for industrial use. The major explanatory variables that are used are GDP, a lagged variable, and seasonal variables.
Bituminous coal demand is categorized into demand for steel making, cement, and other industries to make forecasts. The major explanatory variables that are used for each
model include the pig iron production amount, cement production amount, and index of industrial product.
The major explanatory variables that are used for thermal energy and other energy demand outlook models include GDP, index of industrial product, temperature variables (CDD, HDD), a lagged variable, and seasonal variables.
Data on income, price, and temperature, which have the largest impact on energy demand, is used as major input premises for the mid-term energy demand outlook. The GDP growth rate was set for income forecasts, and international oil prices were adopted for price forecasts.
It was assumed that the GDP growth rate would go up from around 2.8% in 2013 to approximately 4.1% in 2014, which is about the potential growth rate. For economic growth rates in 2014 and onwards, the premise is that economic growth will gradually decelerate from the potential growth rate.
Notes: The economic growth rate of 2013 is a forecast by the Bank of Korea (Bank of Korea’s Economic Outlook, January 2013). The economic growth rates from 2014 through 2017 are the premise values in “2012 Long-term Korea Energy Demand Outlook” of KEEI.
For temperature variables that were used for the outlook, including CDD and HDD, the average temperature data for the last ten years was used.
It was assumed that average year temperatures would be maintained during the forecast period.
Outlook premise
2
Category 2012p 2013 2014 2015 2016 2017 average Annual
GDP growth rate (%) 2.0 2.8 4.1 3.9 3.7 3.6 3.6
<TableⅡ-1> Premise on economic growth rate for mid-term outlook
Notes: CDD (HDD) refers to the difference between the daily average temperature and baseline when the daily average temperature is higher (lower) than the baseline (18℃). Monthly CDD/HDD is the sum of the daily degree days of the corresponding month.
To consider the influence that energy prices have on energy demand, international crude oil prices were used as the outlook premise.
Base international oil prices are forecast at USD 106.23 per barrel for 2013, which is an approximately 2.6% drop from USD 109.06 per barrel in 2012.
For international oil prices for 2013, base oil prices that were forecast by KEEI (November 2012) were used. The forecast international oil prices were used to come up with forecasts for domestic petroleum product and town gas prices.
The premise for domestic petroleum product and town gas prices for 2014 and onwards is that the real price level of 2013 will be maintained.
Average
-2.5 1.0 5.5 12.2 18.3 22.6 24.6 25.9 21.5 15.3 8.0 -0.5 temperature
CDD 0 0 0 3 39 138 206 246 111 10 0 0
HDD 634 482 389 177 29 1 0 0 6 92 302 571
Category
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Year 2013 ~ Year 2017
<TableⅡ-2> Temperature variable premise