Electricity Demand in Wholesale Italian Market.

AuthorBigerna, Simona
  1. INTRODUCTION

    Estimation of demand elasticity has been, ever since the foundation of modern economic theory, one of the most popular empirical exercises in applied economics. The relevance of knowing demand elasticity is quite obvious for advancement of research about consumer preferences and consumer willingness to pay in the market, as well as for guidance to the adoption of policy measure, ranging from taxation to welfare (as price response taxonomy is crucially based upon quantitative measure of demand elasticity).

    In general, typical data used for empirical estimation are market prices and quantities. This means that we observe one point along the demand curve in every state of the world and identification of a (stable) demand function can be obtained using appropriate covariates in econometric estimation procedure. This is true whether one uses aggregate market data, pretending to infer a representative consumer's behavior or one uses individual disaggregated data, coupling observed quantities purchased with market prices (Deaton and Muellbauer, 1980).

    Sometimes we find in the literature estimation based on data reflecting ex ante consumers' willingness to pay based on survey data, focusing on public goods and externalities, such as environmental quality, air pollution, health care, social services rendered by local municipalities, public transportation and so on. However, it is well known that consumer intention to purchase "goods" for which no market exists may be untruthful.

    In this paper, we would like to present a novel approach for estimation of demand elasticity, using collected and well organised data for ex ante consumer demand schedule, based on a large data set of consumer price and quantity bids for electricity in an organized market, notably, the day-ahead market in the Italian Power Exchange (IPEX). (1) In the day-ahead market, consumer bids price and quantity pairs for 24 hours, thus expressing a well-defined willingness to pay for each quantity, presumably according to a complete and well-structured behavioral strategy.

    Surprisingly, these data have never been used before to estimate demand elasticity, while many empirical analyses of electricity markets have used producer bids for price and quantity in the day-ahead market (Wolak 2003 and 2010), to estimate residual demand for each market participant in an oligopolistic market structure. In this case, "demand" behavior is estimated from "supply" data, while there is a more direct and obviously way to estimate demand behavior, i.e. using demand bid data. We criticize the use of "residual demand" facing an oligopolistic supplier to infer the notion of demand elasticity for two reasons. Firstly, each supplier bids in the ex-ante day ahead market without knowing total market demand quantity, thus making, at best, an efficient forecast about "expected" market demand. Secondly, computation of "residual demand" is conditional to a specific maintained hypothesis, namely that suppliers behave as Cournot oligopolists. It is therefore evident that individual and/or aggregate consumer behavior information is not used at all, resulting in a loss of efficiency, and a potential mis-specification in the case of other-than-Cournot suppliers behavior.

    Some studies have used electricity market equilibrium data to estimate demand elasticity (e.g. Bernstein and Griffin, 2006; Horowitz, 2007; Huang and Huang, 2012). A recent survey of empirical work can be found in Labandeira et al. (2012). Not surprisingly, using this data is not uncontroversial as in Fezzi and Bunn (2010): "Thus, there remains a prominent interest in modelling the interactions of supply and demand as a joint system, in particular to investigate the elusive demand elasticity issue from the perspective of publicly available, wholesale day-ahead market data." Some other studies have used individual household information, collected from billing data. Again, these data refer to ex-post household purchases and not to ex-ante demand function (e.g. Reiss and White, 2005 and 2008, using billing data from San Diego area; Wolak, 2010, using a dynamic price experiment in the District of Columbia).

    Another relevant problem highlighted in the previous literature is that "electricity demand elasticities are a subject of nearly endless contention. The relevant elasticity would be a short-run elasticity in the sense of the customer's ability to respond to potentially large hourly price volatility, but still recognizing that customers would know well in advance that prices could be quite volatile. The actual elasticity will depend in great part on technology, as automated response to price changes will surely become easier" (Borenstein, 2004), thus long-run elasticities might be larger.

    In order to identify well defined demand functions in the electricity market, we notice that demand data are expressed by individual consumers in the IPEX in two different ways. Some consumers express a quantity bid without corresponding indication of a price they would be willing to pay. These are consumers who show ex-ante a perfect inelastic behavior, as they are in principle willing to pay any price that would result from market clearing procedure. The IPEX market clearing procedure assigns a default price limit to these bids which has been varying in time, set equal to the maximum price cap to suppliers that is imposed by the Regulatory Authority within a broad range of market mitigation and competition policy measures. Default price assigned to these bids has increased in time from a level of 200 euro/MWh in 2004 to 3000 euro/MWh in 2009-2011.

    Some other consumers express a simultaneous quantity and price bid. These are obviously somehow elastic consumers, for they express a willingness to purchase a certain electricity quantity only if they can pay a certain well defined price. Obviously, when all individual behaviors are aggregated in a market demand function, demand schedule is vertical curve until the portion of "elastic consumers" is reached. From this point onward demand schedule shows a well defined negative slope.

    In principle, there are two possible market clearing outcomes which are crucially relevant here. One possible market outcome occurs when the upward supply curve intersects the vertical portion of the demand curve (see Figure 1.A); another possible market outcome is determined by intersection between the upward sloping supply curve and the downward sloping portion of the demand curve (Figure 1.B). It is clear from these considerations that the concept of market demand elasticity is well defined only in the second case. It is therefore an empirical issue to classify market outcomes according to observed demand elasticity in order to make realistic conjectures about consumer behavior.

    The setup of this paper is as follows. Section 2 describes the complexity of the IPEX due, primarily, to the geographical segmentation. Section 3 describes the theoretical framework. Section 4 describes methodology, data sources and data bank construction. Empirical results are shown in section 5, while Section 6 concludes and summarizes main findings.

  2. THE IPEX NORMATIVE FRAMEWORK

    There are some features of the Italian electricity market which are relevant to understand the operational way in which we have modeled consumer behavior for practical estimation.

    The Italian liberalization process began under the framework of the European Directive of energy sector liberalization, 96/92/CE with national legislation (Law 79/99) enacting a breakdown of the former vertically integrated monopolist (ENEL) to be realized through auction sales to the private market of three GENCO (new companies controlling about 50% of the existing generation capacity). In addition, other legislation (Law 07/02 and 240/04) has created an Independent System Operator (now TERNA), a Market Operator (GME), managing a non-compulsory pool market and a Single Buyer (AU) in charge of aggregating small and poor customer demand.

    Similar to other countries (Newbery, 2005), the Italian market has been organized as a sequence of three markets: day ahead, adjustment and dispatching resource market (ancillary services market). The equilibrium price in the day ahead market is set as the system marginal price (SMP) based on supply and demand bids. SMP is computed for each zone in which market is separated due to network congestion, as measured by excess of physical transmission capacity (Bollino and Polinori, 2005). These latter are the prices received by the supply side while on the demand side there is a unique national price, set as a weighted average of zonal supply prices.

    The adjustment market allows generators and loads to correct parts of schedules which cannot be implemented due to technical constraints. The ancillary services market is a single market allowing the System Operator to procure congestion relief and reserve margin resources, on a pay-as-bid basis. Real time deviations from ex-ante solutions create unbalancing in the system and are costly for bidders, who are accordingly charged balancing costs. (2)

    Other main characteristics of IPEX have been: (i) in the period April-December 2004 only suppliers were allowed to participate into the market while demand was inelastically represented by the best day-ahead forecast of the System Operator; (ii) in January 2005 active demand bids entered into the market while the System Operator maintained the privilege to bid in case of danger of system reliability (i.e. when total market demand is "too different" from day-ahead forecast used for network security management); (iii) in January 2005 AU was instructed to use contract for differences extensively and compulsorily buy into the market. As a result, market liquidity rose to above 60% due to AU dimension; (iv) in 2007 there has been a legislative attempt to reform the market, aimed at discouraging producer quantity withholding strategies. The...

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