THE DEMAN D SID E IN A ZERO - MARGINA L C OST W ORLD Jesse D. Jenkins PhD Candidate, Institute for Data, Systems and Society Researcher, MIT Energy Initiative, Electric Power Systems Center LET S THIN G ABOU T A N EXTREM E C A SE… AN E N TI R E LY (NEAR) ZERO MARGINAL C OST PO WER SYSTEM: Renew ables + Nuclear + S torage + Demand-side Flexibility 2 IN EQUILIBRIUM All resources still recover fixed costs via inframarginal rents • Increase in frequency of zero prices does not m e a n decrease in average price. • Requires very high prices during scarcity periods (well above most price caps) Prices are ~zero w h e n generation is adequate; rise to induce d e m a n d rationing and storage discharge w h e n generation is inadequate. • D e m a n d elasticity, storag e, and reserve scarcity pricin g all increase prevalance of “Scarcity periods” 1-‐7% of hours non-zero prices. when prices spike to drive demand • “Everyone is a pea ker” ra<oning or storage discharge - recover all fixed costs (at non-‐zero opportunity cost) during “scarcity periods.” 3 SHOR T AND LONG - RUN EFFICIENCY Long-run: efficiency determined by incentives for efficient entry and exit. • Long-term uncertainty is a significant factor: variable renew able energy supply, non-monotonic demand, policy intervention • C onsumers and producers are both risk averse, but with inverse preferences. Indicates potential for w elfare-enhancing exchang e of risk via long-term contracts or hedges (and coordination of entry to smooth investment cycles). • First-best outcomes depend on ability of d e m a n d to indicate (differentiated) willingness to p ay to be hedg ed against scarcity. Short-run: efficiency determined (almost) entirely by efficient rationing of d e m a n d (and operation of storage) during periods of generation scarcity. • Depends on active participation of elastic d e m a n d in real-time clearing of energy m a r ket and balancing supply/demand during “scarcity periods.” • Implication: D e m a n d can no longer be insulated from efficient price signals during scarcity periods. 4 THE ROL E O F DEMAND (At least) Three options: 1. Today’s para d i g m Only a sub-set of most elastic d e m a n d enters the “supply side” (of energy and/or capacity m a r kets) as “demand response” resources via ag gregators. Entails transaction costs that have limited participation to large loads to date. Most customers on fixed price (volumetric) tariffs. “Second-best” situation. 5 1. TO D A Y’S P ARADIG M (TRADITIONA L RESOURC E MIX) Almost en<rely inelas<c demand curve Almost en<rely inelas<c demand curve (non-‐scarcity period) ($/MWh) A handful of elas<c demands – bid into supply stack as P “demand response” resources Aggregate supply curve 6 Q – (MWh) 1. TO D A Y’S P ARADIG M (ZER O VARIABL E COS T RESOURC E MIX) Almost en<rely inelas<c demand curve Almost en<rely inelas<c demand curve (non-‐scarcity period) A few storage resources bidding non-‐zero opportunity costs A handful of elas<c demands ($/MWh) bid into supply stack as – P “demand response” resources Aggregate supply curve Zero variable cost resources 7 Q – (MWh) THE ROL E O F DEMAND (At least) Three options: 1. Today’s para d i g m Only a sub-set of most elastic d e m a n d enters the “supply side” (of energy and/or capacity m a r kets) as “demand response” resources via ag gregators. Entails transaction costs that have limited participation to large loads to date. Most customers on fixed price (volumetric) tariffs. “Second-best” situation. 2. The emerging frontier: dynamic retail pricing D ynami c retail pricing provides short-term marginal price signal to all demand. Prices signal scarcity value and induce elastic customers to reduce demand. Risk averse or inelastic customers can opt in to fixed price contracts with retailers w h o absorb volatility for a premium. “First-best” short-run outcome. 8 2. EMERGIN G P ARADIG M (TRADITIONA L RESOURC E MIX) Demand Inelas<c por<on of demand (likely (non-‐scarcity period) under fixed price contracts with retailers) (scarcity period) Elas<c demand animated by real-‐<me dynamic retail pricing (scarcity period) ($/MWh) – Aggregate supply curve P 9 Q – (MWh) 2. EMERGIN G P ARADIG M (ZER O VARIABL E COS T RESOURC E MIX) Demand Inelas<c por<on of demand (likely (non-‐scarcity period) under fixed price contracts with retailers) (scarcity period) Elas<c demand animated by real-‐<me dynamic retail pricing (scarcity period) ($/MWh) – P Aggregate supply curve A few storage resources bidding non-‐zero opportunity costs Zero variable cost resources 10 Q – (MWh) THE ROL E O F DEMAND (At least) Three options: 1. Today’s para d i g m Only a sub-set of most elastic d e m a n d enters the “supply side” (of energy and/or capacity m a r kets) as “demand response” resources via aggregators. Entails transaction costs that have limited participation to large loads to date. Most customers on fixed price (volumetric) tariffs. “Second-best” situation. 2. The emerging frontier: dynamic retail pricing D ynamic retail pricing provides short-term marginal price signal to all demand. Prices signal scarcity value and induce elastic customers to reduce demand. Risk averse or inelastic customers can opt in to fixed price contracts with retailers w h o absorb volatility for a premium. “First-best” short-run outcome. 3. Ideal future structure? As in #2 but customers also facilitate efficient long-term outcomes by entering into long-term hedge contracts (“reliability options”) with generators, either directly or via intermediaries (ag gregator/retailers). D e m a n d is hedged against scarcity prices (for N years) in exchange for fixed-price p aym e nt to generators, but still see efficient short-run signal to induce elastic d e m a n d to ration first. 11 3. IDEA L SHORT - RUN STRUCTUR E (SAME A S # 2 ) Demand Inelas<c por<on of demand (likely (non-‐scarcity period) under fixed price contracts with retailers) (scarcity period) Elas<c demand animated by real-‐<me dynamic retail pricing (scarcity period) ($/MWh) – P Aggregate supply curve A few storage resources bidding non-‐zero opportunity costs Zero variable cost resources 12 Q – (MWh) 3. A CENTRALIZE D LONG - TER M RELIABILIT Y OPTION S A UCTION? Conserva<ve es<mates for un-‐submiVed consumers added to upper leW of demand curve and considered inelas<c Aggregate demand curve for firm capacity reliability op<ons (call op<ons) submiVed by large customers - - and retail intermediaries with different willingness to pay Cleared price per MW-‐yr for fixed price payment in long-‐term reliability op<ons Aggregate supply curve for firm capacity supply (put op<ons) submiVed by generators and Cost or willingness to pay ($/MW pay willingness to or Cost storage – A diverse set of non-‐zero bids for zero variable cost P ‐yr) resources, as generator’s fixed costs and expecta<ons Cleared capacity on supply & demand side of energy market revenue (due to different expectated enter into reliability op<ons; uncleared output during scarcity periods) all differ. bids reflect unhedged generators/demand that can s<ll play in energy market Q – Firm capacity (MW) 13 A N “IDEAL” SCENARI O (BE C OMIN G POSSIBL E S O ON?) • Efficient short-run scarcity pricing is uncapped and passed to d e m a n d in dynami c prices. • C entralized and/or bilateral m a r kets for multi-year reliability options contracts betw een generators (and storage) and d e m a n d (or their intermediaries) exchange inverse risk preferences and help m a n a g e long-term uncertainty. • Retail competition facilitates consumers’ ability to indicate differentiated willingness to p ay for firm capacity and hedge against scarcity prices by allowi ng customers to sign up for different retailers or rates with a different portion of their peak d e m a n d hedged (at different monthly fee). Retailers then act as intermediaries in reliability options contracts. • No free-ridership: individual customers mu st be able to receive differentiated reliability, and in cases of administrative d e m a n d rationing, customers that have hedged least are first to have d e m a n d capped (at their contracted peak). • A utomation of d e m a n d management, contracts that allow direct control by ag gregators, and distributed energy resources can all increase demand-side elasticity and offer an alternative to p aying for reliability options, facilitating competition betw een elastic demand, self supply, and firm generation. 14 Jesse D. Jenkins PhD candidate, Institute for Data, Systems & Society Research assistant, MIT Energy Initiative, Electric Power Systems Center Massachusetts Institute of Technology [email protected] |Linkedn.com/in/jessedjenkins |Twitter @JesseJenkins.