Residential time-of-use pricing : an econometric assessment.
MetadataShow full item record
Constrained electrical power systems and the long lead times ne eded for new capacity necessitate interim demand side management measures such as t ime-of-use (TOU) pricing. This form of electricity pricing has the potential to reduce system peak demand and thus improve the efficiency of power systems. Such time differenti ated pricing mechanisms have been used successfully in the industrial and commercial secto rs to shift demand out of the peak periods but have yet to be implemented in the residential sector in South Africa (SA). TOU schemes are based on the cost of supply and reflect, in part , the changes in short-run marginal costs. In contrast the conventional residential tariffs in SA are based on flat rate structures and recover long-run costs only. The analysis of the impact of such schemes, for both the utility as well as the customers, is gaining importance once more, particularly when most utilities are contemplating the implementation of smart s ystems and advanced metering infrastructures and the costs associated with this. A recent TOU pilot project, HomeFlex, is analysed from an ec onometric point of view. Panel data sets for both treatment groups and the control group ar e obtained from the pilot project database for each customer in two separate experiments in two separate geographic areas. The Caves and Christensen approach is used and the const ant elasticity of substitution functional form is chosen. Conditioning variables such as daily consumpti on per customer as well as climate effects are included in the ordinary lea st squares regression in order to establish the relationship between peak and off-peak consumption and the extent of the substitutability of these two commodities. The elasticity of substitution estimates obtained for stage 1 of the analysis range from 0.339 to 0.384. The conditioning variables enter the analysis as modifie rs to the estimates but their effect is insignificant. The stage 2 estimates range from 0.457 to 0.518. The effect of the conditioning variables is also statistically insignificant at this stage. The effect of the daily and weekly price ratio is therefore the primary factor in det ermining the response of customers to TOU pricing in the HomeFlex project.