This dissertation studieswholesale and sector-wise electricity demand in SouthKorea. Electricity demand analysis provides useful insights formarket performance evaluation, load prediction,market restructuring, tariff schedule design, etc. In recent years, there has been a heated debate in Korea on howto restructure the electricity market, since low reserve margins that have been in operation (6.7% on average in 2010 for instance) have been threatening the stability and integrityof the electricity system. This dissertation thus attempts to address three importantquestions about Korean electricity demand-side market restructuring: (1) What are the estimates of the price elasticity of electricity demand in the wholesale and retail markets, including the residential, industrial, and commercial sectors? (2) How do inter-temporal price changes affect electricity consumption, and what are the estimates of the inter-temporal electricity cross-price elasticities in the wholesale market? (3) Except for the electricity price, what other factors affect electricity consumption in the wholesale and retail markets, including the residential, industrial, and commercial sectors?In Chapter 2, I review current studies on electricity demand estimations, with the emphasis on price elasticity after the year 2000. Twenty papers (selected on the basis of the author'sjudgment) are summarized and evaluated, along with six papers that are discussed in relatively more detail. I also present evaluations and critiques of these works.In Chapter 3, I briefly introduce the Korean electricity market and how it functions. In Chapter 4, I investigate the underlying features of the data in each market and sector and present these features both graphically and statistically.In Chapter 5, I study the wholesale electricity market. Under the Real Time Pricing (RTP) structure, I discuss the model specification with respect to hourly consumption data with a consideration of aggregate utilization behaviors to control the complicated cyclical consumption patterns. Identification is established when the exclusion condition is not satisfied in the demand and supply system. The estimated real-time aggregate price elasticity, based on the whole sample, is -0.0034, the corresponding long-run price elasticity is -0.0640, andthe estimated cross-price elasticitieswithin the previous twenty-two hours are all negative, suggesting complementarity price effects.Price elasticities are also affected by the size of responsive customers. The effects ofinterruptible service operated by Korea Electric Power Corporation (KEPCO) and large buyers in the wholesale market with on-site generators on the demand curve are not detectedbased on a smooth transition model. Price elasticities with regard to each hour within a day are also estimated. Temperature and different types of the day also affect aggregate electricity consumption.In Chapter 6, I study the retail electricity market, with a focus on the residential, industrial, and commercial sectors. Section 6.1 studies the residential sector. A basic regression model is built based on Ito (2012)'s finding that, contrary to theimplications of conventional economic theory, households respond to the average electricity price ratherthan the marginal price when the tariff structure is increasing stepwise. I show that, on average, households respond to the previous month's average electricityprice based on encompassing tests, which might be explained by the cognitive cost of a household obtaining the price information for thecurrent monthly bill, as Ito (2012) implied. A structural time series model (STSM) with four different specifications is also applied to take account of the Underlying Energy Demand Trend (UEDT). Theestimated aggregate price and income elasticities are around -0.2923 and 1.0388. Even though natural gas is a theoretical substitute for electricity, statistically, it does not affect electricity consumption. Other factors, such as temperature and holidays, have significant effects on electricity consumption. Moreover, the UEDT shows a steady decreasing usage trend, indicating, in the residential sector, that improved energy efficiency is the driving force of the UEDT.Section 6.2 studies the industrial and commercial sectors. A simple theoretical analysis is first provided to model electricity demand for each pricing interval under the Time of Use (TOU) tariff structure. An absence of daily/monthly sector consumption data in different pricing intervals prohibited me from applying the theoretical model in practice. Instead, I take advantage of monthly aggregate data and model demand as monthly aggregate consumption against the monthly average price. This modeling compromise would introduce some bias into the price coefficients, for instance, by masking own- and cross-price effects in different pricing intervals. Except for the basic log-log specification, a seemingly unrelated regressions (SUR) model and an STSM, used to take account of the UEDT, are also applied. I find that firms in the industrial sector are responsiveto electricity price variations, with the estimated price elasticity being around -0.19,but that firms in the commercial sector are not. Income elasticities in the commercial and industrial sectors are 1.7326 and 1.4585, respectively. Natural gas substitution elasticity is significant in the industrial sector with the basic and SUR models but this result is not robust to the STSM specification. Substitution effects are all insignificant in the commercial sector. Moreover, both sectors show an increasing UEDT trend. Further, once the UEDT is controlled, the estimated income elasticity becomes smaller (1.2483 in the commercial sector), indicating that part of the UEDT effects are confounded in the income coefficient when the UEDTis not specifically controlled. Other factors, such as temperature and holidays, have significant effects on electricity consumption.
PDF
download
878987797
028-85220240
