Switching on Electricity Demand Response: Evidence for German Households.

• Author: Frondel, Manuel

1. INTRODUCTION

   Recent evidence from experimental economics indicates that the sluggishness of consumers' response to price changes may be due to insufficient price knowledge (Jessoe and Rapson, 2014). Ignoring this fact in estimating demand responses may result in elasticity estimates that incorrectly reflect the responses of consumers who are aware of prices. Further obstacles to consistently estimating demand elasticities emerge from the prevalence of tariffs that include a fixed fee, as this would lead to endogeneity bias (Taylor et al., 2004). This kind of tariff is standard in numerous retail markets, such as telecommunications, water, natural gas, as well as electricity markets. Endogeneity issues are all the more relevant if consumers are free to choose from a broad range of tariffs. Such is the circumstance in Germany since the liberalization of the electricity markets in the European Union (EU) in 1998.

   Employing an instrumental-variable (IV) approach to cope with the likely endogenous tariff choice, this article estimates the households' response to power price changes based on panel data originating from Germany's Residential Energy Consumption Survey (GRECS) for the years 2011 and 2012, thereby adding to the growing literature on the price responsiveness of residential electricity demand by investigating the role that information plays in influencing customer response. While being a standard method for dealing with the endogeneity of an explanatory variable, IV approaches are less frequently applied with respect to estimating electricity demand responses. A rare exception is the study by Alberini and Fillipini (2011), who use an instrument to remedy the measurement error in their price variable.

   As instrument for the endogenous price variable, we use the composite of grid and licence fees. Given that these fees are part of the end-use price and account for about 26% of average household prices, our instrument is clearly correlated with the endogenous price variable. Moreover, these fees are set by grid operators and are fixed at the regional level, so that it is highly warranted to assume that this price component is exogenous to consumers and, hence, does not affect consumers' tariff choice. Using components of a tariff as a source of exogenous variation, such as grid and licence fees, we follow a fairly common identification strategy in the empirical literature (e.g. Ito (2014). A distinguishing feature of our analysis is that we exploit sample information on the households' knowledge about power prices, which is only available for the GRECS waves for 2011 and 2012. On this basis, we combine the IV approach with an Endogenous Switching Regression Model to estimate price elasticities for two groups of households, finding that only those households that are informed about prices are sensitive to price changes, whereas the electricity demand of uninformed households is entirely price-inelastic.

   These results suggest increasing the transparency of tariffs and cast doubt on both the efficacy and the welfare effects of Germany's eco-tax, which was introduced in 1999 to curb the greenhouse gas emissions from electricity consumption. This tax contributed to the doubling of the electricity prices for households since the beginning of the new millennium (BDEW, 2017:29), when prices reached their minimum after the liberalization of the electricity market in 1998. A key consequence of this liberalization for households was the opportunity to freely choose their electricity provider. The improved competition was not sufficient, however, to moderate retail prices (Ros, 2017), which primarily increased due to the introduction of new taxes and levies (BNetzA, 2017). Currently, taxes and levies account for slightly more than half of the power prices for German households (BDEW, 2017:30).

   Another key reason for rising power prices was the introduction of a feed-in tariff scheme to support renewable energy technologies in 2000 (Andor et al., 2017a). While this support scheme was very effective in increasing the share of green electricity in gross production, from less than 7% in 2000 to around 36% in 2017 (BDEW, 2018), the resulting burden for consumers particularly mounted in recent years, above all due to the exploding expansion of photovoltaic capacities (Frondel et al., 2015). As a consequence, the levy with which electricity consumers have to finance the support for green electricity more than quadrupled between 2009 and 2018, increasing from 1.31 to 6.79 cents per kWh. Today, German households have to bear electricity prices that are--in terms of purchasing power standards--the highest in the EU (Eurostat, 2017). Yet, whether the doubling of average household electricity prices since 2000 has induced substantial reductions in electricity demand critically depends on households' responsiveness to price changes.

   Based on our key empirical result that only price-conscious consumers exhibit a price-elastic demand and given that Germany's electricity production still largely rests on coal and natural gas, an important policy implication suggests itself: to dampen the electricity consumption of the household sector and its environmental impact, low-cost information measures, such as increasing the transparency of tariffs, should be implemented on a large scale in order to increase the saliency of prices.

   The subsequent section provides a brief review of the literature on the residential demand for electricity. Section 3 concisely summarizes our database, followed by the presentation of the estimation results in Sections 4 and 5. Based on our empirical results, Section 6 presents some policy recommendations. The last section closes with a summary and conclusions.

2. FINDINGS FROM THE LITERATURE

   The received empirical literature suggests that price knowledge may substantially alter the demand for goods and services. For instance, analyzing the effect of detailed price information presented on water bills, Gaudin (2006) finds that households that are aware of price levels are considerably more sensitive to price changes than price-ignorant households. In a similar vein, examining the effect of price knowledge for various utility services, Carter and Milon (2005) also conclude that informed households are more responsive to price changes than those without any clue about prices.

   For electricity markets, though, empirical evidence on the impact of price information is scarce and primarily available for markets with real-time or time-of-use pricing schemes. Investigating the impact of price information on demand patterns under time-of-use pricing, recent studies, such as Harding and Lamarche (2016) and Pon (2017), detect a large price response due to intra-day demand shifting. With respect to real-time pricing, Martin and Rivers (2018) find evidence that households respond to this information in part by forming habits, rather than adjusting their load-shifting behavior, resulting in a reduction in average electricity consumption of about 3%, an effect that is roughly constant across hours of the day. This result is based on a large-scale field deployment in which close to 7,000 households were provided with real-time feedback on electricity consumption and prices. Exploring the implications of real-time pricing for Swedish households, Vesterberg and Krishnamurthy (2016) also estimate small cost savings from shifting load up to five hours ahead. These results indicate weak incentives for households and retailers to adopt dynamic pricing of electricity.

   There are also studies that investigate the effect of information on usage, rather than prices, on electricity demand. Based on experimental data from Japan, Matsukawa (2004), for instance, measures the effects of usage information received from a monitoring device on residential demand for electricity, the results indicating that this information substantially contributed to energy conservation. Another example is the randomized control trial used by Jessoe and Rapson (2014) to test the effect of high-frequency information about residential electricity usage on the price elasticity of demand. These authors provide experimental evidence that informed households are more responsive to temporary price increases, concluding that inefficiencies due to imperfect information about product attributes can be overcome by providing easy-to-grasp information on a low-cost basis.

   Our empirical analysis adds to this line of inquiry by providing evidence on the effect of price knowledge on electricity consumption levels in retail markets without real-time or time-of-use pricing schemes. Although the demand for electricity has been investigated by economists ever since its discovery, no broad consensus has been reached about the size of the response of residential electricity demand to changing power prices. In fact, price elasticity estimates cover a large range, stretching from 0, that is, an entirely price-inelastic demand, to a highly elastic response as indicated by an elasticity estimate of about -2.5 (Alberini and Fillipini, 2011; Fell et al., 2014; Labandeira et al., 2006; Reiss and White, 2005; Schulte and Heindl, 2017; Shin, 1985; Taylor, 1975).

   A key reason for these huge discrepancies is the specification of the price variable (Espey et al., 2004). While a central assumption in economic theory is that consumers optimize with respect to marginal prices (Ito, 2014:537), recent empirical findings suggest that consumers tend to react to alternative price measures. Ito (2014), for instance, finds strong evidence that households respond to average prices, rather than (expected) marginal prices. By analyzing the price measure issue as well, Borenstein (2009) comes to similar conclusions, claiming that, as a consequence of the non-linearity of tariffs, consumers do not respond to marginal prices due to the lack of precise price knowledge.

   While such results suggest the...