Demand Response: Smart Market Designs for Smart Consumers.

AuthorAstier, Nicolas

    The U.S. is on its way to becoming a net energy exporter. Most current and planned energy production, whether conventional or renewable, will take place on farmland (Hitaj et al., 2018). However, little is known about how energy production influences farmland markets. Farmland makes up over 80 percent of farm sector assets and serves as a primary source of collateral for farm loans (Nickerson et al., 2012). U.S. energy exports have been fueled by a boom in natural gas production from new technology, such as hydraulic fracturing (DiChristopher, 2018). In 2017 shale gas production accounted for 60% of U.S. natural gas production (EIA, 2017) and 18% of domestic energy production (EIA, 2018). With this growth, there have been multiple debates on the impacts of shale gas development (SGD) on the economy, environment, and social welfare. As such, the impact of SGD on property values has become the focus of a growing body of literature (Muehlenbachs et al., 2015). Other research has considered how areas with energy production activities tend to have increased population, employment, business activities and government revenue in the short term (Weber, 2012; Fetzer, 2014). Hitaj et al. (2018) found that oil and gas payments represent over 6% of net cash farm income and almost 5% of off-farm household wages earned by all U.S. farm operators and landlords in 2014. Payments of this magnitude can influence farmland values as well as rural economies.

    Today, as more sophisticated metering technologies have been or are about to be rolled out in many countries, more complex tariff structures are becoming implementable. As a result, numerous studies have been conducted over the last couple of decades in order to investigate retail consumers' response to different tariffs (see Borenstein (2005) for an example of simulation, Wolak (2010) and Allcott (2011) for examples of field experiments, and Faruqui and Sergici (2010) and Newsham and Bowker (2010) for academic reviews). These tariffs usually fall into one of the following categories:

    * Real-Time Pricing (RTP): a direct passthrough of the spot market price.

    * Time-of-Use (TOU): a handful of time periods are defined and a different per-kWh price is set for each period. The span and price of each period is fixed ex ante. As a result, this family of tariffs only allows reflecting variations that are predictable well in advance.

    * Critical-Peak-Pricing (CPP), also called "passive demand response": a default constant price is set for all hours but a limited number of hours per year, chosen ex post, during which the price is set at a much higher level, most often defined ex ante. Alternatively, peak prices may be themselves state-dependent and computed after the signature of the contract. Such a design is called variable CPP (vCPP).

    * Peak-Time-Rebates (PTR), also called "active demand response": customers receive a financial reward if they decrease their consumption (relative to a counterfactual called baseline). In a sense, consumers thus resell electricity.

    Reviews of dynamic pricing field experiments have raised several issues regarding the efficiency of PTR tariffs (Faruqui and Sergici, 2010; Newsham and Bowker, 2010). First, they suggest that PTR may be less effective at reducing peak demand than CPP. (1) Consumers may indeed react more to CPP because they are loss averse, or react less to PTR because this design perfectly hedges them against a bill increase (Fenrick et al., 2014). Second, simplistic methods to establish baselines may decrease the magnitude of consumers' response due to asymmetric incentives: customers have an incentive to respond only if they actually hold a chance to beat their baseline. Finally, a PTR mechanism may reward random shocks in consumption (Ito, 2013), decreasing the cost-effectiveness of PTR programs. (2)

    Besides concerns regarding their efficiency, PTR programs have triggered two main debates in the literature. First, economists fiercely denounced an important flaw of initial designs, namely that consumers did not have an obligation to buy first the power they were then reselling (Chao, 2010; Hogan, 2010; Crampes and Leautier, 2012). This has led to vigorous debates and litigation, notably in the U.S. (Chen and Kleit, 2016). Second, since the counterfactual consumption ("what would have happened if the customer had consumed as usual?") is not observed, some methods have been and are being developed to estimate it (Grimm, 2008; Newsham et al., 2011).

    Baseline estimation raises an additional issue due to potential asymmetric information. Customers are indeed likely to be better informed than their retailer about their future consumption, at least on some dimensions. Since they know how their baseline is computed, they may try to influence its calculation. Chao and DePillis (2013) formalized this issue for a few methods typically used to compute baselines, explaining on these examples why consumers have both the ability and incentives to inflate their baseline. Their observations constitute the initial motivation of the present paper, which contributes to the literature by first generalizing their results, and then discussing the issue of consumers' incentive to opt-in an incentive-compatible PTR mechanism. Doing so, we highlight that policy-makers' willingness to maintain historical cross-subsidies among consumers or not has an influence on the optimal policy.

    Empirically, the magnitude of the information asymmetry may vary across categories of consumers. Typically, one may expect less gaming from small residential consumers (e.g. Wolak, 2007) than from bigger consumers (see for example the famous case of Baltimore's stadium (FERC, 2013). On the long run however, consumers are likely to learn how to game the mechanism over time, as observed in Chen and Kleit (2016). Note that such learning may also happen with small consumers as they may get the ability to contract with intermediaries.

    Two approaches can be contemplated to tackle asymmetric information. The first approach is to develop methods to decrease the magnitude of the information asymmetry and/or increase the cost of cheating (e.g. fraud detection algorithms). There is a risk that it may be an endless route. The second approach is to acknowledge some residual asymmetric information will always remain, at least for some categories of consumers, and to design contracts that explicitly take into account asymmetric information, as do Crampes and Leautier (2015) for balancing markets.

    This paper chooses the second approach. We start by investigating what a socially optimal Incentive Compatible (IC) PTR contract looks like. We show that classic PTR designs allow consumers to arbitrage between spot prices and the constant state-independent price at which they are allowed to buy baseline electricity, compromising incentive compatibility. Baseline electricity should instead be contracted forward at its (expected) spot price, removing the implicit subsidy awarded to consumers under PTR contracts. Under risk-neutrality, an IC PTR design then collapses to a variable CPP (vCPP) design and the relevant economic question becomes to design vCPP contracts optimally in order to achieve high enrollment rates under voluntary opt-in. The solution to this problem crucially depends on whether policy-makers decide to maintain the cross-subsidies embedded in the historical tariff or not. We suggest there may exist some complementarities between this political choice on the one hand (whether or not to maintain historical cross-subsidies between consumers), and the chosen structure for the electricity retail industry (liberalised or local monopolies) on the other hand.

    The rest of the paper is organized as follows. Section 2 presents the analytical framework and derives socially optimal IC PTR contracts. Section 3 then investigates, under different structures of the retail industry, consumers' opt-in choices depending on whether cross-subsidies to non-switchers are maintained or not. We discuss possible extensions in section 4. Finally section 5 concludes.


    2.1 Analytical framework

    We build on the partial equilibrium model by Spulber (1992), and focus on a single class of consumers, defined by common contractible and observable characteristics (e.g. residential consumers with a subscribed maximum power of 6 kVA). Risk-neutral customers are characterized by a one-dimensional type [theta] [member of] [[[theta].bar],[bar.[theta]]] (with pdf g and cdf G), which is their residual private information (e.g. their price elasticity). System conditions vary across numerous exogenous stochastic states of the world which are represented by t. Consume [theta]'s gross utility from consuming a quantity q of electricity in state t is U(q,[theta], t). We hence ignore intertemporal substitution, although it could be added to the model at the cost of much more complicated notations. Consumer [theta]'s marginal utility u(q,[theta],t) [equivalent to] [[partial derivative].sub.q]U(q,[theta],t) where u(q,[theta],t) > 0 and [[partial derivative].sub.q]u(q,[theta],t)

    As we focus on a single class of consumers and not the whole demand, the wholesale price p(t) is assumed to be exogenous and to represent the social cost of power in state t. The socially optimal level of consumption for consumer [theta] in state t is q * ([theta], t) [equivalent to] q(p(t), [theta], t). Finally, we will use the following notations throughout the paper:

    * For a given price p, V (p, [theta], t) [equivalent to] U (q(p, [theta], t), [theta], t) - pq(p, [theta], t) is consume [theta]'s net surplus in state t when she faces the price p.

    * For a given price p, W (p,[theta], t) [equivalent to] U (q(p,[theta], t),[theta], t) - p(t)q(p, [theta], t) is the net social surplus in state t from consumer [theta]'s consumption when she faces the price p.

    The model assumes that there is no uncertainty left...

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