A System of Hourly Demand in the Italian Electricity Market.

AuthorBigerna, Simona
  1. INTRODUCTION

    The main purpose of this paper is to analyze the demand side in an organized electricity market and to test whether agents behave rationally. We postulate a general theoretical model assuming that agents react to price signals expressing some well-defined willingness to pay for each hourly quantity according to a complete, well-structured and simultaneous demand equation system. (1) This is the first attempt in the literature to estimate a simultaneous demand system for the 24 hours and to recover the entire structure of expenditure elasticities and cross price elasticities for all hours of the day in an organized electricity market, ascertaining whether hourly electricity demands can be considered normal or luxury goods and substitutes or complements.

    In this respect, we improve the simple single equation approach existing in the recent electricity market literature, such as those using ex post market equilibrium data (Bernstein and Griffin, 2006; Borenstein, 2004; Fezzi and Bunn, 2010; Horowitz, 2007; Huang and Huang, 2012; Labandeira et al., 2012) and those analyzing organized electricity market behavior (Bigerna and Bollino, 2014). We also proceed further in the direction of some recent studies analyzing individual household behavior facing dynamic pricing experiment (Faruqui and George, 2005; Joskow and Wolfram 2012; Reiss and White, 2005 and 2008 and Wolak, 2011). In particular, we consider those studies taking into account "information only" customers who are somehow informed about time-varying prices but are billed differently (Faruqui et al., 2012; Faruqui and Sergici, 2010).

    In this paper, we use data of the Italian day-ahead market (IPEX) in the period 2010-2011 and we assume that agents demand electricity every day considering the quantity purchased in each of the 24 hours of the day as a different commodity. This rather obvious assumption follows from the fact that agents express their bids simultaneously for the 24 hours in the day-ahead market and that there is a different equilibrium price in every hour. As it is well known (e.g. Newbery, 2005), such institutional arrangement has been mainly motivated by the need to determine an efficient merit order of generators on the supply side of the organized electricity market, but it certainly gives a price signal to the demand side, too. We note that in 84% of the hours the IPEX equilibrium price is determined by the intersection of the aggregate supply curve with the elastic section of the aggregate demand curve. This means that around the equilibrium price the aggregate demand behavior in the Italian market is surely price responsive.

    This fact may arise from the existence in the electricity market of several heterogeneous agents on the demand side, whose characteristics and behaviors have not been explicitly analyzed in the previous literature. There are two main groups of heterogeneous agents acting on the demand side in the organized electricity market: (i) final users who buy directly in the market (such as large industrial firms); (ii) traders and distributors, who buy electricity in each hour and resell it to small industrial and tertiary firms and households. We assume that all agents in IPEX are all cost minimizers and we define them simply as "consumers", providing a thorough discussion of their behavior in the next Section. (2)

    The aggregation of many different individual behaviors yields a market demand. We model aggregate consumer behavior for estimation, using ex ante demand price and quantity bids, i.e. bids submitted before engaging in market transaction, differently from the classic demand literature (based on ex post market data). Our fundamental testable hypothesis is to assess whether consumers in the market act rationally, according to standard microeconomic theory assumptions of adding up, symmetry, homogeneity and individual exact aggregation. A positive answer would allow us to recover significant structural parameters, namely own price and cross price elasticities and expenditure elasticities, attributable to aggregate consumer behavior in the market. The importance to estimate a complete demand system and to recover hour-by-hour elasticities, specifically own and cross price elasticities and expenditure elasticities cannot be overlooked.

    Electricity can be used for different fixed needs, which can be satisfied in different hours of the day, according to price convenience. This gives rise to substitutability among hours (e.g. usage of an equipment in an hour and not in the next). Alternatively, electricity can be required in complex industrial processes, which can use competing energy sources. This gives rise to complementarity among hours (e.g. usage for drying follows washing in the previous hour). In addition, electricity usage may increase more or less than total expenditure according to needs and this entails that electricity can be a necessity (e.g. running equipment for industry) or a luxury good (e.g. air conditioning for large shopping centers).

    We make direct use of collected and well-organized data in the IPEX. (3) According to IPEX rules, consumers have two alternatives to present their bids. They can either bid a price or not together with the desired quantity. If they bid a price, they are obviously elastic consumers, who are willing to purchase a certain electricity quantity only if they can pay their reservation price. If they do not bid a price, they have an ex-ante fully inelastic behavior. (4) Consequently, demand function aggregation yields a composite structure. Initially, demand schedule is vertical. Subsequently, after elastic consumers have entered in the computation, demand schedule shows a well-defined negative slope.

    The setup of this paper is as follows. Section 2 describes theoretical framework. Section 3 describes methodology, data sources and structure of our large database. Section 4 reports empirical results. Section 5 concludes and summarizes main findings.

  2. THEORETICAL FRAMEWORK

    We assume a theoretical framework to model electricity demand acknowledging that there are heterogeneous consumers operating in the Italian hourly electricity organized market. Some consumers directly buy electricity for productive activities, such as energy intensive industries, large public utilities, smaller industrial consumers; some others are traders and distributors who are intermediaries of both industrial and residential consumers (households).

    We assume that all consumers in the market are cost minimizers, demanding electricity to serve a variety of needs, as a production input into the production function for industrial firms and/or as input for traders and distributors who know their final customer needs. Obviously, industrial firms act in their own interest responding to hourly prices.

    However, because some of the consumers in the market are not final users, it is not immediately obvious that their bid behavior reflects some well-defined final user willingness to pay. (5) We assume that there is a principal-agent relationship between a final user and an electricity intermediary bidding in IPEX (similar to Bigerna and Bollino, 2014). The final user has some private information (e.g. his valuation) but the intermediary can develop a statistical forecasting model to predict the final user's behavior. Then, under standard regularity conditions on preferences, the intermediary and the final user will strike a Pareto optimal contract, which maximizes both of their respective objective functions. Moreover, assuming that it is the agent making offers to the principal, mechanism design theory tells us that the intermediary will achieve full surplus extraction from the final user (Maskin and Tirole, 1990). As such, absent strategic bidding on the electricity market, an intermediary has the incentive to submit bids that are reflective of the final user willingness to pay (i.e. those who ultimately maximize utility).

    Thus, assume that in IPEX perfectly competitive conditions prevail on the demand side, given that intermediaries have low market shares and so they do not possess any significant market power (differently from Wolak, 2001 conjecture). Further, assume that intermediaries must pay an unbalancing penalty to the regulator (as stated by the Energy Authority, AEEG, 2013) if their customer real time load deviates from their day-ahead commitment. Then, intermediaries will indeed abstain from strategic bidding, and instead submit bids of their (and their customer's) true willingness to pay.

    In this framework, we reckon that intermediaries react to hourly prices, because they know that a relevant proportion of their customer demand is price sensitive during the day. (6) This is so because the Energy Authority ruled in 2010 to compel distributors to offer to households a contract with two or three different prices during the day and to publicize it. Intermediaries are free to offer variable price contracts to other final users. Thus, Italian final users are very well informed that the electricity price varies on an hourly basis. (7)

    To provide a quantitative measure of the relevance of price sensitive final users in the period analyzed, notice that hourly price contract are undersigned by all large industrial users (directly connected to the HV network) which represent 13% of total electricity consumption and by more than half of MV users which represent 34%. Other LV users (tertiary) represent 27% of the total and households (residential) represent 26%, of which roughly two thirds have the compulsory two-price tariff and one third has the voluntary three-price tariff. Consequently, we know that hourly prices are faced by at least one-seventh (for sure, HV users) and possibly one-third of total final users (considering also MV users). Moreover, at least half of consumers face two or three prices during the day (residential and other LV users). Thus, we expect that there may exist...

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