Residential End-use Electricity Demand: Development over Time.

AuthorDalen, Hanne Marit
  1. INTRODUCTION

    In recent years, there has been a great deal of interest in and debate about end-use energy consumption in Norway. One question is to what extent Norwegian residential electricity consumption for different end uses varies over time. Current and future policy measures used to increase energy efficiency and the consumption of renewable energy form the political background to the interest in end-use consumption. A range of instruments have been used: taxes on electricity and fuel oils; government subsidies for investing in more clean-burning wood stoves, pellets stoves, heat pumps and heating control systems; and a variety of softer policy measures, such as information campaigns and energy labeling. The EU has also adopted an integrated energy and climate policy aimed at cutting greenhouse gas emissions, increasing energy efficiency and increasing the renewable share, all by 20 per cent by the year 2020.

    To determine the effect and potential of policies aimed at changing the composition or intensity of different energy uses, it is important to know both the proportion and amount of electricity used for different purposes. Heating is often provided by different energy sources, and in order to predict the effect of changing the composition of these sources, it is important to have information about the amount of energy and, specifically, the amount of electricity used for heating. Furthermore, to determine the effect of future and past policies aimed at increasing energy efficiency, it is important to know the amount of electricity used for different purposes (including behavioral and technical aspects that influence actual consumption). To analyze the energy saving potential of efficiency measures, it is also important to know how electricity use for relevant purposes has developed. In this context, electricity consumption for space heating and other purposes is important, and it is also important whether changes have taken place over time and, if so, why.

    A large proportion of household energy consumption is used for stationary purposes such as heating, washing, lighting, cooking and other household uses. For heating households may use several different energy sources; they can use electricity, firewood, fuel oil, paraffin, gas, district heating, pellets etc., or any combination of these. A large share of Norwegian households (70 per cent) use more than one energy source to heat their home, and a combination of electricity and firewood is the most common. Thus, there are extensive substitution possibilities for space heating in Norwegian households, which in turn have a substantial effect on behavioral responses to environmental and energy policies targeting household stationary energy consumption. Residential electricity consumption is very heterogeneous, and several variables are important in relation to explaining both total electricity consumption and the different end uses. These explanatory variables vary both between households and over time. Explanatory variables for electricity consumption and electricity for different end uses include electricity prices, prices of other energy goods, outdoor temperature, technology, income, policy measures, efficiency, appliance ownership and other house-hold characteristics. In this paper, we study the importance of different explanatory variables to households' electricity consumption for different end uses over time, with particular focus on electrical appliances.

    The most common econometric approach to end-use estimation used in the literature is conditional demand analysis (CDA). CDA is a multivariate econometric technique combining information of households' total electricity consumption, household specific information regarding prices and weather and detailed household survey data on energy appliance stock and background data. The technique yields robust end-use estimates for energy consumption of different appliances. Early studies include Parti and Parti (1980), Aigner et al. (1984) and Lafrance and Perron (1994). Later studies have used data for directly metered electricity consumption for specific appliances in some households to improve the results from traditional CDA. Metering data are used in, for example, Bartels and Fiebig (1990), Aigner and Shonfeld (1990), Bauwens et al. (1994), Hsiao et al. (1995) and Bartels and Fiebig (2000). Several of these studies focus on end-use consumption over time in the form of load profiles. However, few studies of electricity consumption for different end uses focus on changes in end-use consumption over a period of years. Larsen and Nesbakken (2004) compare engineering and econometric models and study one year only (1990). We use a traditional econometric CDA as our initial approach, but we deduce the model for explicit use on annual cross-sectional samples of households. The data are taken from the Norwegian Survey of Consumer Expenditure expanded by questions on energy use. In these data, we also see variation in energy prices over households, which is not very common. In addition to data for 1990, we use more up-to-date data, i.e. data for 2001 and 2006. Based on repeated, independent cross-sectional data for 1990, 2001 and 2006, we have estimated end-use consumption for each of these years. Using the same econometric method together with annual cross-sectional data enables us to analyze end-use consumption consistently over a period of time. We use samples of households for three years, and our data for average electricity consumption for these years corresponds well with official statistics for total average electricity consumption in Norwegian households. In this way, our data is representative for the household sector, which is important as we seek to disaggregate the aggregate trend (the development of aggregate electricity consumption).

  2. THE ECONOMETRIC MODEL

    In the conditional demand (CDA) model for total electricity consumption, dummies for ownership of different appliances are included as explanatory variables, i.e. electricity consumption is conditional on having (or not having) an appliance. The coefficients of the appliance variables provide estimates of electricity consumption for the different appliances and form the basis for the end-use estimates. The main idea of the econometric model is that the estimated difference in electricity consumption between households with and without a particular appliance is interpreted as the mean electricity consumption for this appliance. Estimates of mean electricity consumption for each appliance, given possession of that specific appliance, are multiplied by the proportions of households possessing the appliances. This yields estimates of the mean electricity consumption for different appliances. Electricity consumption for each end use divided by total electricity consumption gives end-use shares. Details of the model are presented below.

    2.1 Econometric Model

    If we assume, as a base line, that annual electricity consumption for end usej for household i ([[epsion].sub.ij], i=1,..., N) is observed through direct metering, the following end-use equation can be estimated:

    [mathematical expression not reproducible] (1)

    where [C.sub.im] (m=1, 2,..., M) are economic and demographic variables; household and dwelling characteristics, electricity prices, heating degree days etc., and [C.sub.jm] is the mean value of these variables for households possessing appliance j. [e.sub.ij] is a stochastic error term. The parameter [[gamma].sub.j] represents the mean value of electricity for end usej given that household characteristics ([C.sub.im]) are equal to the respective variable means calculated over all households possessing end usej or given that no economic or demographic variables are relevant to the electricity consumption for end use j (i.e. [[rho].sub.jm] = 0 for all m). Household characteristics vary across households, however. Thus, the second term of equation (1) represents an adjustment of end-use consumption due to the impact of economic and demographic variables.

    We do not have data for electricity consumption for different end uses in each household, and equation (1) cannot therefore be estimated. Neither do we have national figures for end use electricity consumption on average for all households. However, the total electricity consumption of each household is observed. By summarizing electricity consumption over all end uses in equation (1), we arrive at the total electricity consumption of household i ([[epsion].sub.i]). In this connection, we have to take into account that not all households possess all types of appliances and that not all end uses can be specified. [D.sub.ij] is a dummy variable with value one if household i possesses appliance j and value zero if the household does not possess appliance j. Of a total of J possible end uses, we define S as electricity end uses that can be estimated separately, i.e. j = 1, 2,..., S,..., J and S

    [mathematical expression not reproducible] (2)

    where [u.sub.i] is a stochastic error term with the form:

    [mathematical expression not reproducible] (3)

    with expectation zero and non-constant variance. Heteroskedasticity follows from the model specification, but whether heteroskedasticity is a problem is an empirical question. Theoretically, the error term varies systematically with the explanatory variables, see equation (3). For example, the higher the dwelling size and number of electrical appliances, the higher the potential for higher variation in electricity consumption between households (due to variation in the use of the appliances). Consequently, the difference between estimated and observed electricity consumption may increase with the size of appliance holdings. Even though parameter values are not affected, we open for the possibility of non-constant variance in the estimations.

    The economic and demographic variables are included so that they adjust electricity consumption...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT