Texas, Hedg’Em
The Texas crisis need not be the death of dynamic pricing.
The Texas energy disaster last month will yield many important lessons for assuring resource adequacy in a time of increasingly unpredictable weather. The supply shortfall in ERCOT, which Jim Bushnell wrote about two weeks ago, amounted to about 25% of what customers would have liked to consume and it continued for about four days. That’s roughly 10 times larger than the shortage California had last August, and lasting more than 20 times longer.
Still, one takeaway from Texas is the same as here: demand adjustment should be playing a much larger role during times of tight supply. Among nearly all energy economists, and an increasing number of grid operators and utility managers, there is a consensus that the current approach of paying for demand reduction is not working well, and that dynamic pricing is a more cost-effective approach to demand response during high stress on the grid.
(source: https://www.pinterest.ph/pin/561824122241064404/)
Yet, the Texas crisis made abundantly clear the problems of exposing customers to spot price volatility and risking mammoth bills. The “simple” solution that I have heard from many economists is hedging. For instance, a customer could purchase its expected consumption quantity in advance, before the weather is known, which on a forward market would have cost $20-$30 per MWh for Texas power in February.
Locking in a price in advance for a fixed quantity dampens the customer’s bill volatility. At the same time, the customer still has the full price incentive to reduce consumption during a price spike, because every MWh the customer saves is one less additional MWh it has to buy at the high spot price. (Or, if the hedge quantity is greater than its demand, it’s one more MWh it can sell back at the spot price).
(source:berico.com)
Customers hedging their expected consumption is a good start, but in Texas last month it would have been woefully inadequate.
To see just how inadequate, consider Max’s Strudel Hut, a Texas customer on a pricing plan that passes through the wholesale spot price to customers like the Strudel Hut. Let’s say that Max hedged this exposure to the spot market by purchasing the Hut’s expected February consumption through a forward market at $25/MWh. ERCOT estimates the full demand during the 95-hour crisis averaged about 65% above its typical daily February load. Say that the Strudel Hut’s demand rose during the crisis by the same proportion as system demand, so it ended up buying “only” about 40% (0.65/1.65) of its actual consumption at spot prices. That means that for the 95 hours of the extreme electricity crisis it bought 60% of its power at $25/MWh and 40% at ERCOT’s $9000/MWh price ceiling (where the spot price remained throughout the crisis). If you do the math, Max’s Strudel Hut would face an electricity bill for the month of February about 34 times higher than expected.[i]
The problem is that the Hut’s demand is positively correlated with the price, so hedging typical consumption turns out not to be very much protection. One way to address this correlation is to over-hedge, that is, to purchase on the forward market a quantity larger than the Hut’s expected consumption, as shown by McKinnon in 1967 (and rediscovered by me 40 years later).
But there is an even better solution: a hedge quantity that fluctuates proportionally with system demand. To be concrete, let’s say that Max’s Strudel Hut averages about 1 MWh per hour during normal times (yes, that’s a lot, but it’s really outstanding strudel), and its demand for electricity fluctuates in the same proportion as system demand. Max buys 1 unit of this “quantity-indexed” hedge at a given price for the month (we will get back to the price in a minute). Say the average system demand is 40 GW in February, so the hedge contract delivers 1 MWh of power at the contract price if system demand is 40GW in an hour. If the Strudel Hut consumes more in that hour, it has to buy the extra at the spot price, and if it consumes less, it automatically sells the residual quantity back to the market at the spot price. Either way, the cost (or opportunity cost) of consuming additional electricity is the spot price.
If system demand is 60 GW in a given hour, however, the hedge contract delivers 1.5 MW of electricity at the same contract price. And if the system demand is 30 GW, the contract delivers 0.75 MW. Regardless of the hedge quantity, any change in the Hut’s consumption still costs (or benefits) Max the spot price in that hour. Still, the fluctuating hedge quantity means that the hedge in a given hour is likely to match the Hut’s consumption much more closely than a constant-quantity contract. And if the Strudel Hut’s demand varies less (or more) than system demand, Max can go with a combination of the quantity-indexed hedge and a constant-quantity hedge to get as much or little variation as he needs to match the Hut’s correlation with system quantity. So, Max doesn’t have to worry about huge electricity bill fluctuations and can instead focus on developing his special recipe for lebkuchen before the holiday season.
(source: agilitypr.com)
Importantly, the hedge quantity would not change with the Strudel Hut’s own electricity usage; that would undermine the incentive to conserve when the price is high. Actually, that is what the most common pricing plans today do by giving a customer the option to buy all it wants at a preset price, which is called a “requirements contract”. Instead, the quantity-indexed hedge would change with the total system demand, a quantity the Hut’s usage doesn’t affect, so the hedge itself doesn’t change its incentive to consume.
In practice, well before each month begins, Max would choose how many units of the quantity-indexed hedge to buy, or his retail provider could do it for him based on the Hut’s past consumption and the goal of minimizing bill volatility. Either way, once he makes the purchase, he couldn’t change the hedge quantity in specific hours. So, sellers of the hedge (probably a utility or a financial market participant, possibly through a market mediated by an RTO/ISO) can price them by analyzing the relationship between market demand and the spot electricity price. For those who think about such things, the seller might not easily be able to balance such a contract completely with its own purchases. One natural counterparty, a generator, would face performance risk in its electricity production, as we saw in Texas. A retail provider selling such hedges, however, would still be far less exposed than it is now when it offers a requirements contract.
(source: amperesum.com)
Note the difference between such dynamic pricing with hedging and programs that pay customers for reduction from a baseline that is a function of the specific customer’s past demand. Programs that pay for reduction are subject to (a) baseline manipulation, (b) baseline selection – customers often get to choose among different baseline formulas, so understandably choose the one that delivers the highest payout with the least effort/quantity reduction, and (c) participation selection – customers who don’t find a program that is a winner for them simply choose to continue buying all they want during price spikes under a fixed-price requirements contract.
Dynamic pricing creates more powerful incentives for conservation during high-priced periods, and thoughtfully-constructed hedging greatly reduces bill volatility. At the same time, it reduces rewards for apparent reductions that are actually just due to clever baseline strategies or participation choices. Large and sophisticated customers could make their own hedging decisions, and would more effectively lower bill risk with a quantity-indexed edge as described here. Less sophisticated customers could have their retailer hedge for them, or they could sign up for a simpler dynamic pricing plan, such as optional critical peak pricing.
Dynamic pricing isn’t going to completely replace other demand response programs anytime soon. We will still want to have interruptible rates that a system operator can activate just before it has to resort to involuntary outages. But well-crafted dynamic pricing can greatly reduce the need for more drastic interventions without subjecting customers to unacceptable bill risk. And it would go a long ways towards keeping the Texas and California electricity markets out of the news.
I’m mostly tweeting energy news/research/blogs @BorensteinS .
Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.
Suggested citation: Borenstein, Severin. “Texas, Hedg’Em” Energy Institute Blog, UC Berkeley, March 15, 2021, https://energyathaas.wordpress.com/2021/03/15/texas-hedgem/
—-
[i] Showing my work: During an average February, Max’s Strudel Hut would purchase 1 MWh/hour, or 672 MWh, at $25/MWh for a total bill of $16,800. During the 95-hour crisis, it would have purchased an extra 0.65 MWh each hour, or 61.75 MWh, at $9000/MWh for an extra cost of $555,750, or a total bill of $572,550, 34 times higher than its expected bill.
Categories
Severin Borenstein View All
Severin Borenstein is Professor of the Graduate School in the Economic Analysis and Policy Group at the Haas School of Business and Faculty Director of the Energy Institute at Haas. He received his A.B. from U.C. Berkeley and Ph.D. in Economics from M.I.T. His research focuses on the economics of renewable energy, economic policies for reducing greenhouse gases, and alternative models of retail electricity pricing. Borenstein is also a research associate of the National Bureau of Economic Research in Cambridge, MA. He served on the Board of Governors of the California Power Exchange from 1997 to 2003. During 1999-2000, he was a member of the California Attorney General's Gasoline Price Task Force. In 2012-13, he served on the Emissions Market Assessment Committee, which advised the California Air Resources Board on the operation of California’s Cap and Trade market for greenhouse gases. In 2014, he was appointed to the California Energy Commission’s Petroleum Market Advisory Committee, which he chaired from 2015 until the Committee was dissolved in 2017. From 2015-2020, he served on the Advisory Council of the Bay Area Air Quality Management District. Since 2019, he has been a member of the Governing Board of the California Independent System Operator.
Energy Institute Blog 
I’m puzzled – this was a once-in-blue-moon weather event. Isn’t the solution for that sort of thing to have all the utilities have a finger on everyone’s heat pumps, like BGE does for many Maryland suburbanites, and just turn everyone’s thermostat down to something uncomfortable but not dangerous – 55 F or so? They could pay people to sign up, and make it a sliding scale with temperature – pay more for a lower set-point. If this were 4 days of everyone wearing sweaters and sweatpants to bed, instead of burst pipes, that would have been a really different (and much better) story, no?
In an ideal world, unlike the messy one we live in, we would all have smart EVs and smart buildings arbitraging distributed consumption and storage against wholesale energy prices that fully incorporate externalities such as the social cost of carbon, without consumers and EV owners facing frequent decisions about what is, in many cases, a small part of their monthly or annual expenditures. Given that we are probably a decade away from that world, the most difficult political and economic challenges are agreeing to make the transition, including the amelioration of economic inequality. Severin and his colleagues have recently addressed retail rate redesign separately, a subject that needs to be front and center during the transition.
I’d think that the best option for Max’s Strudel Hut would be to take his utility’s fixed-price option. If we are solving for Max’s utility, he (a) minimizes time spent thinking about electricity where he is undoubtedly at a competitive disadvantage (b) removes all price risk and (c) relies on the government / social trust to ensure fixed price contracts deliver as promised.
Moving to a different hedge (in today’s world where fixed price hedges are cheap) seems to be structurally inferior. Max spends more time thinking about his electricity consumption, increases his price risk, and still relies on governmental / social trust to ensure the lights stay on!!! Unless the utilities prioritized keeping the lights on for folks with dynamic pricing, Max still faced the same blackout risk last month!
What is the benefit to Max of moving away from a fixed price? What caused the problem in Texas? Could a power marketer in Texas offer such a plan? If so, is it a problem that people have a choice among hedges and choose the fixed price option? Should people be forced into alternate pricing arrangements or allowed to select their desired arrangement?
In my rudimentary mind, the breakdown in Texas was the disconnect between the sale of a fixed price contract (i.e., guaranteed electricity at a given price) and the reliable provision of electricity. In the old inefficient world, the utility provided the fixed price contract but also built/bought the electricity to ensure sufficient supply to satisfy the contract. In some areas, this reliability function has been delegated to the ISO/RTO but remains a binding reliability criteria, thus providing strong assurance that the supply will exist. No such centralized structures exist in Texas.
I suspect it was left to trust that power marketers / utilities will “firm up” their contracts somehow; maybe those parties assumed the traders they bought their hedges from were firming things up even higher up the chain. That didn’t work out as designed. I haven’t heard anybody state the grid couldn’t function in this weather (they do all the time), its that the parties involved in the provision of electricity did not believe the investments in supply reliability were cost effective.
Nate
Not so rudimentary. You really captured the key issues here. It’s not so simple to just implement dynamic pricing and then declare the problem solved. There’s other structural issues created by the network nature of the electricity grid that you’ve identified. We do have to rely on the collective response of consumers to mitigate these types of demand spikes, but we also need to answer the question of “what if that’s not enough?” Why should those customers who did respond suffer just the same as those who didn’t if the power is curtailed, which is the case now?
I’m also not convinced that there’s some sort of great inefficiency simply because we don’t price at the short run cost of electricity. No matter what we likely have to recover the long run costs of generation investment. Under the regulatory regime almost everywhere, utilities are guaranteed recovery of either prudent investments or PPA costs over a fixed period of time with a fixed rate of return. The amount of disallowances are miniscule to nil. What’s the point of pricing at short run costs if either those prices inevitably lead to generator bankruptcy or leave consumers exposed to very high prices needed to recover those investments through the lottery of price spikes? How do we know that any hedges will be priced to deliver the financial assurances that are required?
In California specifically, such a retail pricing vision can only be carried out if AB 57 which guarantees full cost recovery for investor owned utility generation assets, whether owned or contracted, is repealed. This is a necessary step before dynamic pricing can have any real impact other than superficially shuffling the chairs on the Titanic.
“Among nearly all energy economists, and an increasing number of grid operators and utility managers, there is a consensus that the current approach of paying for demand reduction is not working well, and that dynamic pricing is a more cost-effective approach to demand response during high stress on the grid.”
This statement is simply not true. There is much debate among energy economists and industry practitioners about the most effective strategy. Virtual power plants (VPPs) based on demand response are becoming widespread as an alternative to dynamic pricing. Intermediary-managed demand response appears to be a favored path over dynamic pricing. In PG&E’s GRC, the complexity of implementing real time pricing is becoming readily apparent.
In the Statewide Pilot Program, 80% of the response came from 20% of the participants. Why not target those customers instead? If response is dependent installing smart devices, we’ve created two income barriers, one for being a homeowner (only 55% of California households) and being well enough to spend on the devices.
We need to start looking at customers as prosumers, not consumers. We don’t threatened generators to force them to generate more–we offer them higher prices. We should be doing the same to encourage consistent demand response.
Severin,
You have laid out an interesting solution to a persistent problem. Thank you.
Two questions:
1. I don’t understand your assertion that “A retail provider selling such hedges, however, would still be far less exposed than it is now when it offers a requirements contract.” Let’s suppose that the hedge is for a portfolio of customers whose demand does follow the index (e.g., system load) quite closely. So under a requirements contract the supplier would be on the hook for the cost of the energy demanded by the customer group in each hour, and would charge a fixed price for that energy. The hedger would be on the hook for the cost of the same energy (assuming perfect correlation with the index) and would similarly receive a fixed rate per kWh. In the latter case, the customers may shift or curtail load in the high-priced hours, but they keep the savings, not the hedge provider. So how does this approach reduce the seller’s risk?
2. Any idea what that hedge might cost, above the weighted forward energy price?
Severin,
Thank you for another thoughtful look at the dynamics of the Texas market. I am left with several hesitations: 1. How many customers would have the where-with-all to play with hedging? 2. Hedging does not change the amount of economic loss resulting from a flawed market design. It just shifts it to someone else. 3. The Texas disaster provided a grand experiment in demand reduction, however involuntary. Yet, with a 25% reductions in load resulting from the rolling blackouts, there seemed to be no downward pressure on day-ahead prices. 4. Not everyone can hedge, because no one is going to sell hedges at the level, but if they did, and if everyone over-hedged as you suggested, dynamic pricing would have no short-term effect on load, since no end-user would be exposed to the dynamic price. 5. Since the initial premise was that unhedged dynamic prices can be punitive and economically damaging, and since short-term electric demand is relatively inelastic, what’s the point?
Best,
Steve
Steve, I’m sure that Severin can reply to your questions himself, but I’m taking a crack at it to demonstrate my reading comprehension skills.
1. How many customers would have the where-with-all to play with hedging?
If the utility or marketer offers the bundled RTP+hedge rate, everyone can participate.
2. Hedging does not change the amount of economic loss resulting from a flawed market design. It just shifts it to someone else.
I think you are wrong there. If a generator sells a hedge, it gets a guaranteed price from the hedge and its cost of serving the hedge is covered by the revenue it gets from its output. Of course, it needs to keep its plant on line during price spikes. And if the generators succeed in doing that, the spikes will be mitigated.
3. The Texas disaster provided a grand experiment in demand reduction, however involuntary. Yet, with a 25% reductions in load resulting from the rolling blackouts, there seemed to be no downward pressure on day-ahead prices.
The operation of the ERCOT reliability adders is complicated. It is not part of a normal supply curve
4. Not everyone can hedge, because no one is going to sell hedges at the level, but if they did, …
Outside of TX, almost all customers are hedged, by vertically-integrated utilities, fixed-price basic service from restructured utilities, or fixed-price contracts with marketers. Remember that most customers will not opt for RTP, anyway.
…and if everyone over-hedged as you suggested, dynamic pricing would have no short-term effect on load, since no end-user would be exposed to the dynamic price.
Reread Severin’s proposal. If a customer on his rate uses less than its hedge level, it sells the surplus at a high price.
5. Since the initial premise was that unhedged dynamic prices can be punitive and economically damaging, and since short-term electric demand is relatively inelastic, what’s the point?
Reread the proposal, You missed the punchline.
“…and if everyone over-hedged as you suggested, dynamic pricing would have no short-term effect on load, since no end-user would be exposed to the dynamic price.
Reread Severin’s proposal. If a customer on his rate uses less than its hedge level, it sells the surplus at a high price.”
I’m not sure I see how this is different from the problem of setting the baseline for demand response. An informed customer could play the hedge level like a financial market, choosing to overhedge in anticipation of large offsetting revenues when the market hits a high price level. Given that wealthy customers are more likely to be able to afford this financial play and be willing to take this calculated risk, this likely would lead to another regressive wealth transfer.
“How many customers would have the where-with-all to play with hedging?”
Severin, I’m with Steve. Max is busy making sure his apples are crisp and ripe, his almonds are freshly-slivered. As long as lights and coffee machine in the Hut are on and he can accept credit cards, the last thing he needs is to think about hedging his electricity consumption. Like most restaurateurs he’s already working 12-hour days – Max has better things to do.
I would love to understand the difference between “demand adjustment” and “service denial” – the contemporary trend of shifting service responsibility from vendors to consumers. After all, this extra time Max needs to save money on his electric bill has value, does it not? I would challenge anyone to show me evidence demand-adjustment programs have saved customers a dime, and not only allowed customers, at best, to maintain their current cost of electricity while imposing a significant cost in inconvenience – in re-scheduling their day to use electricity at the right time, in the right weather.
That sounds familiar…could it be another example of transferring the cost of solar and wind integration from generators to customers? For once we add the transmission unreliability of gas, to the variability of wind and solar, there could have been no other outcome in Texas – market design, notwithstanding.
“For once we add the transmission unreliability of gas, to the variability of wind and solar, there could have been no other outcome in Texas – market design, notwithstanding.”
Given that neighboring states, experiencing the same extreme weather, did not suffer the same scale of outages, it’s clear that other outcomes could and did occur with different market structures, regardless of resource mix.
This is an interesting idea. It would be useful to model how having a large number of electric cars plugged into the grid, both selling and buying power as the price fluctuates, would affect electric price and shortages. Eventually computer models may be able to use weather forecasts to estimate the best times to buy and sell stored electricity, minimizing the users cost and maximizing their return on investment in an electric car. If most Texas household had two electric cars, each with 100kwh batteries with 60 % of their storage available, would this have been able to get the household through the recent electricity shortage?