EMP S2E20: USC academics Matthew Kahn and Bhaskar Krishnamachari discuss the potential for smart grids and dynamic pricing to address climate emissions and reduce energy demand.

Researchers Matthew Kahn and Bhaskar Krishnamachari, respectively an economist and electrical engineer at the University of Southern California, discuss their recent commentary calling for greater dynamic pricing in the electric industry. By reducing peak electricity demand, more responsive demand can eliminate the need for new fossil-fuel power plants and help reduce climate-altering and other harmful emissions. They look to "experimentation" with opt-in, voluntary demand-response programs that expose electricity consumers to varying power prices to analyze and determine what best motivates them to conserve energy. They urge that Inflation Reduction Act clean-energy funding be directed to develop effective demand-response programs for all consumers, not just large industrial and commercial customers with the greatest economic incentive to reduce energy use. Artificial intelligence and machine learning tools can be developed that automatically respond to price signals on behalf of "Average Joe" electricity consumers, making decisions in response to price signals based on the wants and desires and needs of the individual customer, they say, and they urge that demand-response programs be designed with lower-income households in mind.

Transcript

EMP S2E20. Matthew Kahn and Bhaskar Krishnamachari, transcript
(edited for clarity)

EMP: Welcome to the Energy Markets Podcast. I’m your host, Bryan Lee, and our guests today are Matthew Kahn and Bhaskar Krishnamachari. Matthew Khan is the provost professor of economics and Spatial Sciences at the USC Dornsife College of Letters Arts and science. And Bhaskar Krishnamachari is professor - and helped me with the Chinese pronunciation here, Bhaskar, The Ming Hsieh Faculty Fellow in electrical engineering at the Viterbi School of Engineering, also at USC. And so welcome, gentlemen. And the reason I've invited you two on was you co-wrote a commentary in a forum called the Conversation. And you talked about the promise of smart grids and dynamic pricing as a way to help us use the grid to respond to the increasing threats of climate change, and its increasing intrusions into the reliability of the grid. So, um, why don't we talk about that? Let's start by, what is dynamic pricing?

BK: So, you know, just maybe to take a step back - in this article, we're talking about kind of this trilemma that we have with energy. We want our energy to be clean. Increasingly, we're moving towards more renewable sources. We want our energy to be reliable - as continuously available as possible so we don't have interruptions to our power supply. And we want it to be affordable. And these three goals are kind of, you know, at odds with each other. And often, we've seen incidents particularly recently and there's also all of this is happening with the backdrop of climate change. We're seeing increasingly in many parts of the country and around the world, days that are hotter than they used to be, colder than they used to be. And during those peak moments, which might increase the demand and the load on the system, the question is, how do you get the demand to be more flexible because your supply doesn't necessarily change when you have hotter days, right? 

EMP: So what you're talking about is introducing elasticity into the electricity market. 

BK: We want to be able to, you know, have an ability, a forcing function, if you will, to moderate the demand on those days. And today, the way a lot of that is being done is through, essentially, pleas, right? You get these power companies begging people to turn off their fans and their ACs. And you know, there are people who are goodhearted people, and really solid citizens and they do that and there are a lot of people that don't. And so the question is, are there incentive systems or mechanisms that can help do that? And dynamic pricing is an approach. And my colleague, Matthew Kahn will tell you more about why this is all Economics 101, but essentially giving a pricing signal is one way to adapt that that demand.

EMP: Yeah, so I guess I threw elasticity at the wrong person here. Matthew, you know, I was starting to say, this has been a perennial problem ever since we introduced competition into the electricity industry is, how do we get price-responsive demand? So why don't you talk about that a little bit.

MK: So, Bryan, in my Econ 101 classes I've hypnotize generations of students talking about demand curve sloped down - that when the price of electricity is lower, or any good, people want more of it. Bryan, what Bhaskar and I have been talking about is our diversity. Different people have different abilities to respond to price signals if we can confront them with these price signals. An example - Jimmy Carter back when he was president told us on a cold day to just put on a sweater, rather than cranking up the natural gas heat in our house. And Rosalind and Jimmy Carter, being good people or for whatever motivation, they were able and willing to conserve on energy on such a day. But other people, as Bhaskar was saying, might not be willing to engage in such voluntary restraint and we need to more forcefully nudge them with a price signal. But Bryan, we both know that when we raise prices for energy, poor people are less able to cope with this. And so Bhaskar and I are setting out on a research agenda of how do we have the win-win of the pricing incentive signal without making the poor poorer, and it's the sort of Goldilocks issue that we've been wrestling with.

EMP: So you're talking about manufacturing price signals, according to the wholesale price? I'm not sure what we're looking at here, in terms of, what is dynamic pricing.

BK: So, so the short version of dynamic pricing, first of all, it's telling us that the price is going to change, right? It's not the same price and all times. It's not the same price during the hottest days. It's not the same price at night as it is during the daytime. And so what would you want to do? So if you want to reduce the demand, we're going to have to increase the price. Why would we do that? We would do that because the supply - well there's multiple reasons - the supply itself might be time-varying. The more renewables we rely on there going to be times of the year when the sources of power we're relying on are not producing as much. So if it's wind power, there may just not be enough wind during certain times. So you have lower supply of wind power. Or solar power during cloudy, rainy days, right? And so, so the price signals have to adapt to that gap between supply and the demand that you're going to see. On hotter days, when naturally people are going to want to turn on their ACs, but if everybody did that, there'll be way too much load of the available supply. And so those are days we want to have higher prices. 

EMP: What I'm asking you is, how do we set those prices? Under what parameters? Do we have regulators trying to approximate a market or do we let a market work?

MK: So, Bryan, I love your question. Right now in economics, there is more and more humility that we know that we don't know how to bring this about. So I would want to see the local electric utility experiment with different dynamic pricing and to see if we confront Bryan Lee with a 40% increase on the hottest days, if you opt in to face this incentive, do you reduce your consumption by 4% by 9% by 15%? So, Bryan, the electric utility and their data scientists have access - have the ability to run these experiments to learn these unknown elasticities that you refer to. So these known unknowns can become known knowns. If economists and engineers can work with the data scientists to run these experiments to see what price increases do we need to confront manufacturing, commercial and residential consumers with to bring supply and demand in balance on the hottest days of the year to avoid the next Texas freeze blackout?

EMP: Yeah, well, we're doing that to a certain extent already. We've got wholesale markets being linked with the large industrial and commercial customers who you know, have the most financial incentive to curb their use during these high-price times. And we're struggling with how to do more of that. And you're talking about how to bring in the mom-and-pop electricity customers into that kind of a paradigm where in response to high prices in the wholesale market, you reduce consumption in the retail market.

BK: Yeah. And I think, you know, the key that Matthew just mentioned is the experimentation. I think we don't want to take the approach that this is a solved problem, it’s just a matter of implementing it. We just got to charge everyone high prices and you know, they'll just stop using electricity. It's not as simple as that, right? People, different people have different needs, different capabilities of scaling back on their demand. And so even in a given market, and some of those could change over time. And so there has to be a systematic approach that really starts from the principle that we don't know what is the right pricing structure. We don't know what the current demand elasticities are. But we can learn it and adapt it in a way that's better than the status quo, where you're just begging people to scale back and it's not happening.

MK: To piggyback on that, just for a moment - if my family expects that we're going to face dynamic pricing for a couple of years, this could affect the durables we buy. What air conditioner we buy. So, Bryan, what Bhaskar and I have been talking about is setting off a virtuous circle. If my family knows that if we can opt in for five years into dynamic pricing, then does a green air conditioner now pencil out because of the saved lower operating costs over the next five years? And so these synergies between facing higher prices and demanding greater durables is something that fascinates me and Bhaskar.

EMP: There’s been a lot of experimentation over the years. And you talked about pricing fairness, and not impacting someone who really can't afford to maybe pay, you know, $1,000 a megawatt hour for their electricity, which is something that never happens anyway, except in extreme instances like we saw in Texas last year. But there was an Edison Electric Institute Edison Foundation study, gosh, I don't know 12-15 years ago, and they looked at some of the experiments in dynamic pricing, and they found in that study - and hopefully I can find it now after all these years - but they found in that study, that those with lower incomes were more motivated to respond to the price signal than were those who were wealthier.

MK: I believe that. And so, if you have a diminishing - so I agree with that. Bryan, Bhaskar and I emphasize fairness in our work because we're worried about political backlash. I love the way you set up our discussion that we need price signals to help us both to reduce our carbon footprint and to adapt to extreme weather. But many politicians have been leery of adopting these incentives because they fear a political backlash. And so Bhaskar and I have emphasized an opt-in design, that the electric utilities would not order Bhaskar, this is the new rate you're going to face. They would offer Bhaskar’s family, or if he's an industrial consumer, his firm, an incentive to sign up and then see who self-selects on their own free will to participate. We think that that would defuse the political backlash issue.

EMP: I guess my point is that we've got a bifurcated electricity industry. FERC regulates the wholesale market and they've done a lot of work to promote wholesale power markets in regions across the country, and they're still trying to get them in certain regions. Then the other half of the equation is retail regulation by the states. And we've got maybe a dozen or so states that have ostensibly opened up their retail markets to competition - not very well, in my opinion, and many of the guests’ opinions here - the only state that has really gotten it right is Texas, because they've quarantined the utility from the retail market. But most of the states are still regulated. And so there's a regulated price that is divorced from the reality of the wholesale market, which is setting the price signal. So why would we have regulators or utility companies trying to approximate those price signals when they're already there? And we can make them open to the price signals and especially when we start pricing carbon, this is going to be very important. 

MK: So I certainly appreciate your points. And I agree with much of what you're saying. A fundamental idea to me is that all final consumers of electricity should face a scarcity price signal, and currently they do not. And so we - Bhaskar and I are beginning discussions with certain California electric utilities who are on the hook. They don't want power blackouts to occur. So Bryan, we haven't talked about power blackouts. Most of the time, the United States is able to keep the lights on but if we don't adopt dynamic pricing of how we reduce the risk of power blackouts. And so I greatly respect the point you made. I want America's consumers going forward to face the true marginal cost of their decisions when they leave a computer on - just for every activity involving electricity. Bhaskar, your thoughts?

BK: I absolutely agree. I think I think you know, making sure that those scarcity signals are reflected in the price is a big part of it. I think the other resistance that you sometimes get to dynamic pricing is people are worried about how do you actually absorb cognitively dynamic prices, right? What is the frequency with which these prices are going to change for the average consumer. If you tell me every hour the price of electricity is different and it's different on different days. I’m literally going to be you know looking at a live feed of this on my on my computer to decide whether to turn off something or not. That's not something the average consumer is going to be able to do. But I think there also - and this is where, you know, a lot of my research interests are really kind of connected to this question - has to do with advances in technology. So when we talk about the Internet of Things, we talk about connected devices, we talk about the use of AI and ML tools integrated into these devices. We talk about smart thermostats in our homes. We talk about kind of bringing in much more automated tools to operate our homes and the electricity that are then plugged into the dynamic price signals. So it's not the Average Joe that has to make these decisions but tools that can help them make these decisions and responses in response to those dynamic price signals. It isn't really there today but integrating dynamic prices with the with the new technology that's in our homes is important too.

EMP: Yeah. So what you're talking about is developing a program that, according to the customer’s wants and wishes, adapts for them to the pricing signals. And they just plug it in and walk away, right?

BK: Absolutely, right. In fact, a paper that we're working on together with a student at USC is looking exactly at what a financially autonomous thermostat might look like. So financially autonomous in the sense that, you give it a budget. You say this month, I'm able to spend about this much for my electricity needs. Let's just talk about the thermostat as an example because it's controlling the temperature in my home. It learns or I can tell it what my thermal comfort zone is. I don't want it to be too cold, too hot. What does that mean for me? It will take into account the daily temperature readings it might even look at predictions of the temperature readings over the rest of the month. It will look at the dynamic price signals. It'll take into account my thermal comfort and my budget and make decisions that try to keep me as happy as possible within the budget that I've granted. And perhaps if it's a particularly hot month, it might come to me and ask for a slight raise to be able to keep me comfortable and I can turn it up or down. But the point is that these sort of fine-grained decisions are not something that the human has to make. It’s something that the AI tools can help you with.

EMP: So right now we're talking about something behind the meter. But a lot of what your paper in the Conversation was about was the potential for all of the smart meters that we spent millions of dollars to help install throughout the country, and maybe getting them to be used for more than the utility just driving by and taking a meter reading. (laughter) You know, a lot of the conversation that's been happening in the electric industry is around data privacy and who owns the data. And that's true, whether you're talking about the Smart Meter, where there are a lot of entrepreneurs and competitive companies out there that want to compete with the utility, but they can't because the utility has control of the data and that's acting as a big impediment to better utilization of smart meters.

MK: I think this is a crucial point on such data being both the private good and a public good. Bhaskar, can you share your thoughts and some of our conversations on this?

BK: So the data that we're talking about here - so first of all, like it is the measurements taken by the smart meters that are often at a pretty fine granularity. Every 15 minutes or less, they can be reading your usage. And kind of collecting that statistical information across different households can help you understand what populations are very similar in their electricity usage. And then as you're learning more about your – the customers that you're serving, which ones have more flexibility in their demand, more elasticity. You're able to use the leverage this information coming from the smart meters. So that that's definitely a big part of it is to do data analytics on the information that's collected at a fine granularity. So that's certainly one piece of it.

MK: What I like about Bryan's point very much is, a health insurance company, if someone is sick, will try to raise rates on that individual. So, Bryan, a major topic in economics is whether it's fair to price discriminate - to charge different people different prices for the same good. And one of the things Bhaskar and I've spoken about is with an opt-in design, people would be free to choose. A libertarian should be comfortable with this, of opting in to face these dynamic prices. So we talk in our essay that we're not asking everyone to face dynamic pricing. We almost want a guinea pig effect of we want to identify those - we want to use machine learning to identify the subset of residential, commercial and industrial consumers likely to reduce their consumption if they faced dynamic pricing, and then to offer them a financial incentive to sign up. But they are free to choose. Milton Friedman is free to choose if he doesn't want to sign up for this. And it is the case what you're right about is the data scientist at the electric utility would then be able to trace out the demand curve for these individuals, as you see the same person's consumption under different pricing regimes.

EMP: We've had demand response for a while in the industry. And as I said it's pretty much only with the large customers - the industrial and commercial customers. By having demand response, you can have a real effect on air quality because when you're hitting peak demand, that's when you're running the dirtiest plants and the most inefficient plants. So if you can shave off that peak just by a slight amount you're creating a whole lot of value to this whole social engineering project that we're involved in. 

MK: Absolutely. Yeah, this is exactly, you know, part of that trifecta is clean energy. So by being more efficient, by passing these scarcity signals down to the customers in an opt-in manner, you're able to leverage that optimization to reduce that demand. And as you say, during peak hours, you can you know, turn off some of your dirtiest power production sources. And so it is very much connected to that whole drive towards more renewable sources as well, right, which tend to have greater fluctuation in the amount of power that they're providing. 

MK: We're talking to one California utility, who says if they don't engage in greater demand side management, they're going to have to build another power plant. And so do you deliver power by shaping demand or increasing supply? And so all these utilities, especially those who face land-use regulations have the right incentives to experiment more with demand-side tools.

EMP: So you talked about Jimmy Carter and his cardigan. You know, there's ways of doing this without really impacting the consumer, especially the retail consumer, in any real dramatic way. You spread out a 1% or 2% demand reduction across the customer base, you're lessening the need for a lot of electricity generation. Correct?

MK: Indeed, yeah, so that's definitely a big part there is, you know, reduce the need to build more power plants, build more, you know, sources of power that are coming from non-renewables.

EMP: And Bhaskar you I saw in your CV, that you've got some expertise in blockchain. Do you want to talk a little bit about what blockchain is? And there's been a lot of interest in blockchain and what it might pose for the electric industry.

BK: You know, when people talk about blockchain often what comes to their mind is Bitcoin, which technically speaking was the first implementation of a blockchain technology but for a very specific use case of cryptocurrency. But the core idea for blockchain goes well beyond cryptocurrency use cases. At the core of it, it allows you to send transactions and log transactions in an immutable manner, so that once you write this to a public ledger, it's very hard to erase it. And this can be a source of transparency. So that if you if you are entering any information on this public ledger, first of all, everyone can see it, and second, you can’t go back and manipulate your numbers behind the scenes. And further advances in blockchain technology have gone from just storing data in an immutable manner in a transparent manner to doing computations in a transparent manner so that you can run certain types of transactions and programs on those transactions, that are sometimes called smart contracts, in this very trustworthy manner, because you can see what's going on and if somebody manipulates it, you can observe it. So from a perspective of transparency, auditability, tracing back what happened, when, who did it? These types of questions tend to be potentially addressed by blockchain technology. So when it comes to the energy industry or the environment, more generally, you know, there are various use cases people think about where blockchain might play a role. Just to give you one concrete example, when we talk about CO2 emissions. If the information about CO2 emissions were logged in a public ledger, and they're hard to manipulate, and you have a way to kind of make sure that there is essentially the ability to ascribe the sources of CO2 emissions to, you know, who did it when, where, and if you had policies or regulations around it, they can consult the sort of “golden source of truth” to be able to apply whatever measures there might be. You know, from an energy industry perspective, as well, there's a lot of talk about setting up smart contracts to do energy trading for even dynamic pricing type-of-use cases where the policies that you're trying to put in place are implemented in code but in a transparent manner. So you know there's no manipulation happening behind the scenes. And so if you're entering into various agreements, that there's kind of a credible way to show what those agreements were, that they're being carried out or not being carried out, and various mechanisms you can implement on the blockchain to either reward or penalize the right behaviors, the wrong behaviors. And so, you know, so at a high level, there's a lot of potential here. Whenever I talk about blockchain to broad audiences, I like to tell them, it's a lot like trying to fly an airplane while we're still building it. This is the one domain where, you know, even though it's been around 10 years since Bitcoin showed up, the technology is still largely immature. There's a lot of ways in which the core protocols need to scale up to better use cases. They're still very expensive and slow. And there's a lot of security vulnerabilities still being worked out. But at the same time, I think, you know, if you look at maybe a five- to 10-year horizon, there's going to be a lot of use cases beyond cryptocurrency or we'll see blockchain play a role.  

EMP: You two concluded your piece for the Conversation by talking about using the clean-energy funding from the Inflation Reduction Act to study this and maybe try and get greater implementation of these sorts of demand-response programs?

MK: So This is a, Bryan, this is a very important point and thank you for your careful read. So in our proposal, we want to incentivize individuals to opt in to face dynamic pricing and we’re no dummies. You're going to have to offer an incentive but money doesn't grow on trees. So Bryan, a question arises, can you use President Biden's Inflation Reduction Act money - can a utility tap into Washington, D.C., to get a subsidy to launch this experimentation? Because Bryan, let me make a point. I am very interested in the green guinea pig effect. When we know that we don't know if an idea is going to be effective. We need someone to be a first-mover. The Wall Street Journal always mocks California saying those crazy grains are trying different things but people in Texas can benefit it California pilots an idea that actually works. So neither of you guys are laughing. We need to pilot more ideas to see if they work and so there is - even with my libertarian instincts - there is a reason for the federal government to subsidize things. To overcome the first-mover disadvantage. We need someone in the country to pilot new ideas to see if they work it because as we learn, these can be broadly adopted.

EMP: Rather than California setting an example for Texas, I think the industry guys in Texas would say, we've got something to tell California.

BK: Maybe another point there is that whatever schemes ultimately get deployed, they're going to have to be revenue. They can’t be at a loss for utilities, right? So if they want to offer some incentives, the money for that incentives has to come from somewhere. And initially, it might be helpful to use these government subsidies. But ultimately, the system that we design with the incentives has to be revenue neutral for the utilities. They can’t be losing money because they're trying to incentivize customers to cut back their demand. And so that is part of the ultimate goal of where we want to head. But I think these early government subsidies are just to help with the experimentation to help evolve the types of policies that are going to be sustainable in and of themselves without necessarily requiring long-term subsidies.

EMP: Well, that's part of the quandary is, you know, how do you incentivize the utility to convince their customers to use less of the product when they get paid by how many electrons they sell to their customer? I go back then to my earlier statement about how Texas, I think, got it right by quarantining the utility from the retail market, you don't worry about having to incentivize the utility. You've got entrepreneurs and competitive companies out there who are willing to risk the capital to conduct that sort of experimentation.

MK: So I love that. And I think that that's a very important point. Ultimately, I think our research agenda will inform that. But I like how you're approaching that. And that's, and it shows you the complexity of this industry. But Bryan, a question for you. With climate change, and with more severe weather, Bhaskar and I have discussed that it makes these ideas about experimentation even more important. So a question about Texas. With the Texas freeze of February of 2021, do you think that Texas has enough experimentation by the private sector such that the next Texas breach will cause less damage? Are we - is this sector more effectively adapting over time, because of dynamic because of the price signals that competition is bringing about?

EMP: Well, the reason the Texas grid crashed - and we've done a lot of post-mortems in the last couple of years - but there's two overarching problems that contributed to the outage in Texas. And that was that you had a what I believe was clearly a climate change-driven event where you had this once-in-a-hundred-year event happen 10 years apart. And the natural gas pipeline companies and the producers didn't want to expend capital to meet that one-in-a-hundred-year type of event like we saw. And I think there's been some mixed success by the Texas policymakers in terms of encouraging that sort of approach. But the utility industry, the utility sector, is getting their weatherization under control. And I'm not sure it's clear yet how much the natural gas sector is investing to be prepared for the next time we have a deep freeze. The other part of the problem was that a lot of heating in Texas like we have in Florida, too, is inefficient resistance heating, which is a big suck on electricity when they all go on at once. And that's what happened. And so that was part of the cascade that caused the grid to shut down in order to prevent damage to the system.

MK: So that's a very important point. I believe that if demanders faced dynamic prices, more of them would engage in weatherization. So we like peanut butter and jelly are complements in a similar way. If you're exposed to price spikes, you're more likely to invest more in such weatherization and so this is the complementarity that Bhaskar and I are trying to explore in our ongoing research.

EMP: Yeah, and if you've got a price signal that people can respond to, they're going to have an incentive to invest in a more efficient heating system for their home.

MK: And it's not just that they are incentivized to adapt, you know, to better technology. It's that the technology developers are then incentivized to develop new technologies, right? So it goes all the way down that kind of chain.

EMP: Well, I really appreciate you two coming on the podcast and helping us flesh some of this out. I really thought your piece in the Conversation was very thoughtful and covered a lot of points that we try to address on this podcast. Matthew Kahn and Bhaskar Krishnamachari. thank you very much. 

# # #