Estimating the Impact of Time-of-Use Pricing on Irish Electricity Demand.

• Author: Di Cosmo, Valeria

1. INTRODUCTION

   Electricity demand traditionally exhibits a substantial peak during a small number of hours each day. Policymakers are aware of the potential benefits that may be generated from a shift in energy consumption away from peak times. Smart meters, in conjunction with time-of-use (TOU) pricing, can facilitate an improvement in energy efficiency (1) by providing consumers with enhanced information about electricity consumption and costs, and thereby encourage a shift away from consumption during peak hours.

   In the EU, a number of recent pieces of legislation have promoted the use of smart metering, including the Electricity Directive 2009/72/EC, which requires Member States to ensure the implementation of intelligent metering systems and to carry out a cost-benefit analysis of the system by September 2012 (Commission for Energy Regulation, 2011b). In Ireland in May 2009 the first National Energy Efficiency Action Plan (NEEAP) was adopted in line with EU requirements, and included a commitment to encourage more energy efficient behaviour by households through the introduction of smart meters (Commission for Energy Regulation, 2011a).

   In 2007, the Irish Commission for Energy Regulation (CER) announced their intention to introduce a trial smart metering experiment in the Irish residential and small-to-medium enterprise (SME) electricity markets. (2) Smart meters, which replaced the existing mechanical meter readers, were introduced in approximately 5,000 households and 650 SMEs. While participating households self-selected into the trial, and therefore our results cannot be generalised to the overall population (see also Allcott, 2011b), participants were randomly assigned to control and treatment groups. Treatment groups were exposed to a variety of time-of-use (TOU) tariffs and information stimuli (in-home display (IHD) units, monthly billing, etc.). Data was collected over the period 14 July 2009 to 31 December 2010, and as the experiment began on 1 January 2010, six months of pretrial data are available for both the control and treatment groups.

   Numerous other countries have experimented with the use of smart meters (e.g U.S., Canada and Denmark), (3) and there is a growing international literature analysing the impact of TOU tariffs on residential and commercial electricity consumption. The availability of high-quality data on a large and representative sample allows us to estimate the impact of TOU pricing on electricity consumption in Ireland for the first time. (4) An earlier statistical analysis of the trial was conducted for the CER in 2011 (CER, 2011a); in this paper, we extent that analysis to analyse household responses to TOU pricing and various information stimuli using econometric methods. Ireland is an interesting case study as much of the international literature focuses on the U.S. where the use of air conditioning for residential use is common. As in Ireland there is no demand of air conditioning during the summer, the trial results show the impact of different TOU and information stimuli on residential electricity demand net of the air conditioning effects, which accounts for a large part of the household responses in the U.S. (Faruqui and Sergici, 2009). Finally, limited socioeconomic information on the participating households is also available. (5)

   The first aim of this paper is therefore to disentangle the effects of the different TOU tariffs (peak, day and night) on residential electricity (6) consumption during different times of the day. Our results show that different information stimuli lead to differences in household responses during different times of the day. In particular, the presence of an IHD that indicates the quantity and cost of electricity consumed on a real-time basis leads households to contract their consumption during the peak hours. This is consistent with international research that highlights the importance of instantaneous direct feedback in generating sustained demand responses. Furthermore, we find that the magnitude of the contraction increases as the ratio of peak to off-peak prices increases. However, the extent of the additional reduction in peak demand due to a steepening tariff schedule is very small in absolute terms. The other stimuli (i.e., bi-monthly and monthly paper billing) also give rise to reductions in peak demand when TOU tariffs are employed, but for them there is little evidence of further reductions as the ratio of peak to off-peak prices rises further. This suggests that the consumers in this experiment respond on the basis of a simple heuristic: they know peak prices are now higher than at other times of the day and they change their behaviour to reflect this, but further increases in the differential are either not fully perceived or evoke only a weak response for some other reason.

   Second, we investigate the determinants of electricity consumption during different times of the day. We find that controlling for day of the week, public holidays, climatic conditions and household appliance ownership, the presence of different TOU tariffs affects household electricity consumption during the peak hours, but does not lead to a significant change in electricity usage during the day and night periods.

   Finally, in an attempt to gain further insight into the demand response, we test for possible heterogeneity across different household types. We examine the variation in our results across different socio-economic groups, as proxied by the highest level of education completed by the chief income earner of the household. We find that households with higher education levels respond to TOU tariffs during the peak period (consistent with the overall results noted above), but that households with low education levels are less responsive to TOU tariffs.

   Section 2 discusses previous research in the area. Section 3 describes our data, while Section 4 outlines the methodology employed in this paper. Section 5 presents and discusses empirical results, while Section 6 summarises and concludes.

2. LITERATURE REVIEW

   Estimates of the price elasticity of electricity demand in the residential sector can be very different depending on the type of data used (time-series, cross-section, panel), context (national, regional or local economy), size of the variation in price and time periods covered (see also Alberini et al., 2011). Here we focus on studies that, similar to the approach used in this paper, use microdata on households and that examine the impact of price and information stimuli on electricity demand.

   The extent to which price elasticities differ across population groups is a common focus of research in this area. Baker et al. (1989) use data from the British Family Expenditure Survey over the period 1972-1983 to analyse household expenditure on electricity, gas and other fuels. Prices are national averages. They find a significant own-price elasticity of -0.758 for electricity demand, with considerable variation in the estimated own-price elasticity across different household types (e.g., by presence of children, type of heating, income, etc.). Alberini et al. (2011) estimate price elasticities of energy (electricity and gas) demand using data on over 74,000 households in the 50 largest metropolitan areas in the U.S. over the period 1997-2007. They report price elasticities of demand for electricity use that range from -0.67 to -0.86, with the elasticities slightly higher in poorer households.

   As TOU pricing is becoming more common, so too are studies evaluating households' responses to TOU pricing. Bartusch et al. (2011) examine the impact of the introduction of a demand-based TOU tariff on a pilot basis to a group of 500 households in Sweden. Using data before and after the introduction of the TOU tariff, they find that total electricity consumption declined by 11.1 per cent and 14.2 per cent in the first two years after the change to TOU pricing (with the size of the reductions higher in the winter months). They also find a shift in electricity demand from the peak to off-peak period of 0.8 and 1.2 percentage points in the first two years (with the shift greater during the summer months). Filippini (2011) analyse electricity data at the city level for 22 Swiss cities over the period 2000 to 2006. They find that the own-price elasticities vary between -0.80 and -0.89 during the peak period and between -0.90 and -0.95 during the off-peak period (positive cross-price elasticities imply that peak and off-peak electricity are substitutes). An earlier study, also using Swiss data, found similar results (Filippini, 1995). Matsukawa (2001) examine the impact of TOU pricing on residential electricity demand in Japan. The results show that (1) household response to the high price of the peak period is relatively modest, and (2) the relative magnitudes of the price and selection effects (i.e., participation in the trial) depend on the ownership of water heaters.

   A variant on TOU pricing is dynamic pricing, whereby rates respond to critical periods of electricity use. In the U.S., critical periods occur typically during the top one percent of the hours of the year where somewhere between 9-17 percent of the annual peak demand is concentrated. It is very expensive to serve power during these critical periods and even a modest reduction in demand can be very cost-effective (Faruqui and Sergici, 2009). A comprehensive review of 15 experiments (largely based in the U.S.) (7) with dynamic pricing of electricity was undertaken by Faruqui and Sergici (2009). They find conclusive evidence that households (residential customers) respond to higher prices by lowering use. The magnitude of the price response depends on several factors, such as the magnitude of the price increase, the presence of central air conditioning and the availability of enabling technologies such as two-way programmable communicating thermostats. Across the experiments studied...