Modelling Energy Demand and Household Welfare Using Micro-Data

The paper is concerned with the empirical modelling of domestic demand for energy in the United Kingdom at the level of the individual household (most previous British work has used aggregate time-series data). The paper develops a two-stage budgeting model of the household's demand for energy conditional on its ownership of durables. Preferences at both stages of the budgeting process are determined implicitly by the cost functions we specify, the almost Ideal form at the first and the Gorman Polar form at the second stage. At the second stage the household is assumed to allocate its expenditure among gas, electricity and a composite good, "other fuels". This composite good is subject to such large measurement errors in the data set concerned that it is treated as an unobserved variable. Some simplifying assumptions lead to linear estimating equations for gas and electricity consumption and it is shown how the model's structural parameters, i.e. those of household preferences, can be identified for use in the calculation of welfare gains and losses from price changes. The model is estimated using a sample of over 50,000 households drawn from the annual Family Expenditure Survey (FES) for the years 1972-83. This data has not been fully exploited in the analysis of energy demand to date. Unrestricted reduced-form estimates are presented and the diversity of implied price and income elasticities is emphasized. Both gas and electricity are found to be inferior goods (conditional on durable ownership) for over half the sample.