The economics of rooftop solar

Katie Luu: From MIT, this is the Energy Initiative. I’m Katie Luu. Welcome to the podcast. Today we’re talking with former Energy Fellow, recent PhD graduate, and MITEI researcher Scott Burger about energy poverty and rooftop solar. Welcome, Scott, and thanks for joining us today.

Scott Burger: Thanks for having me. I’m thrilled to be here.

KL: We often hear about rooftop solar as a no-loss solution for climate change. It saves money and reduces emissions. However, we’re learning that the story surrounding rooftop solar may not be so straightforward. Scott, your research and research that you point to shows that economic incentives for rooftop solar can actually work against low-income customers. What’s the story there?

SB: I want to start by taking a step back and saying that I started my career in rooftop solar. I actually went to work straight out of undergrad for a rooftop solar developer down in Texas. I was really enthralled with the industry, found it to be fascinating. It was growing incredibly rapidly and really changing very dramatically. I was a big believer in it. The story is that you install some solar panels on someone’s roof and they save money, because now they’re buying less electricity from the grid and that’s a win. Potentially the grid saves money because now this homeowner is buying less electricity from the grid. Maybe you need to buy fewer assets in order to meet their load in the future. You need fewer wires and you need fewer power plants to meet that customer’s load in the future. The customer saves money, the grid saves money, and on top of this, there’s a lot of research that shows that customers really like it. Customers are happy when they install solar and customers are demanding these products and that’s really cool. It seems like a win-win-win. On top of that, it’s green. You’re reducing carbon emissions. Awesome. When I got here at MIT and started digging into the economics of distributed solar a little bit more, what we realize that the story is maybe much more complex than we initially thought. There’s actually now a lot of evidence that shows that if you’re not careful, installing rooftop solar can actually drive up the costs of the distribution networks and transmission networks required to serve load. There’s some research out of California now from a professor named Frank Wolak at Stanford that estimates that two-thirds of the cost increases in the distribution networks for Pacific Gas and Electric in California between 2005 and 2016 are attributable to the growth in rooftop solar photovoltaics. If you believe his estimates, he’s saying that these rooftop solar systems are really driving a lot of cost in the system. That’s also consistent with some of the research from here at MIT. The Future of Solar study did a lot of modeling and simulation and showed that, if you’re not careful, rooftop solar can drive up the cost of distribution systems. There’s also some evidence from other types of what we call distributed energy resources. Resources like rooftop solar or home energy batteries and things like that, that are installed in the distribution system at customer loads, et cetera. There’s a lot of evidence that other types of distributed energy resources, or DERs, can cause problems. There’s some interesting evidence from California, where the California government was subsidizing home batteries and also commercial batteries. The goal was basically to reduce peak demand in the distribution level. But what they found when they actually looked at the data is because of how these batteries were charging and discharging, in some cases they were increasing peak demands. Which was really counter to what they wanted to do. They also found that in some cases they were increasing emissions. Because they were charging during times where maybe emissions were high and discharging during times when emissions were low. These programs were really not achieving the goals that they had. I think there’s this increasing recognition that the story around the economics, or the cost impacts, is much more complex than we initially thought. There’s increasingly a concern over the potential distributional impacts that distributed energy resource adoption has. When I say distributional impacts, I mean when I adopt a technology, how does that impact other customers? In particular, the other customers of different socioeconomic backgrounds. There’s really a lot of data now that’s emerging on the income trends of the individuals that are adopting these technologies. There’s a great paper from some researchers out in California at UC Berkeley, Severin Borenstein and Lucas Davis. They looked at the tax credits that the federal government provides for installing different clean energy technologies. They found that between 2003 and 2013, basically of the $18 billion dollars in clean energy tax credits, the bottom 80% of incomes only received about 10% of the credits. While the top 20% of incomes received almost 60% of the credits. It’s pretty shocking, the income distributions there for the different customers. More recently, there’s been a lot of evidence emerging about the income trends of rooftop solar adopters in particular. There’s a landmark study from Lawrence Berkeley National Lab a few years ago, which is some data that I’ve used now, that looked at the income trends of PV adopters between the year 2000 and the year 2016 across a number, I think 13 different states in the United States. What’s amazing is, 1) how consistent these trends are, and 2) how lopsided these trends are. About 80% of rooftops solar adopters are in the wealthiest 60% of society and only about 20% of rooftop solar adopters are in the lowest 40% of incomes. I guess the way to interpret that is basically someone in the top three income quintiles is almost four times as likely to adopt solar as someone the bottom income quintile. We wanted to take that information and look at, what are some of the potential impacts of that? Severin Borenstein and Lucas Davis looked at the distributional impacts of this federal program to support some of these clean energy technologies. But nobody had really looked into some of the distributional impacts of the other programs that are used to support clean energy technologies. In particular, one of the biggest programs that’s managed at the state level is net metering or rate design, which we’ll get into in a little bit. Which effectively determines how much you pay someone that adopts rooftop solar for the electricity that they generate. We chose to look at that income distribution data and also some of these state-level programs to dig into what that meant for the distributional impacts of rooftop solar.

KL: Great. And you mentioned PV, which I just want to define for the audience as photovoltaics, correct?

SB: Yes, yeah, PV is photovoltaics. Generally speaking, when I’m talking about solar in this conversation, I’m really referring to solar being installed on rooftops and typically residential rooftops. Homeowners, maybe some small businesses. You can imagine also solar being installed on the rooftop of, say, a gigantic Walmart building. Those installations can actually be quite large and you can take advantage of economies of scale. But when I’m talking about PV in the context of this conversation, I’m usually talking about rooftop solar, residential rooftop solar, where the systems are relatively small, and as a result tend to be relatively more expensive on a unit economics basis, on a dollar per unit of electricity that you generate basis.

KL: Great. You referred to rate design. How are rates designed? Are they determined at the state level or federal level?

SB: Taking a brief step back and thinking about, what are rates? Rates are something that we all interact with but probably none of us even think about. I study rate design as a researcher and don’t really think about rates other than in the academic context. Rates basically are how much you pay for electricity. When you consume electricity from the grid, how much do you pay for that? Rates also have another function under many state programs. The rates determine how much you get paid when you generate electricity from, for example, a rooftop solar system that you’ve installed on your site. When I consume one unit of electricity from the grid, I pay, for example, 10 cents per kilowatt hour. Where the kilowatt hour is the unit that we use to measure how much energy you consumed. Under a net metering program, if I generated one kilowatt hour, I would either, save that 10 cents, or if I exported one kilowatt hour, I would also get paid 10 cents or I would offset demand in some later period with that kilowatt hour that I generated. Rates do two things. They 1) determine how much you pay for electricity, and 2) determine how much you get paid for generating electricity with some of these DERs, these distributed energy resources. Coming back to some of the research we’ve done here at MIT, we in 2015 started a project called the Utility of the Future study. The focus of the Utility of the Future study was to think about how the power sector was evolving as we were incorporating all of these new types of technologies. These rooftop solar systems, home energy batteries, electric vehicles, smart thermostats, all of this cool stuff that’s happening in the power system. How does that change the economics of the system? One of the things that we realized was that rate design was really, really, really critical. It was one of the key levers that regulators and policymakers could pull on to ensure that all of these changes were happening in a positive way, were happening in a way that reduced system costs for everyone, reduced emissions, et cetera. The way I like to think about it is, again, as we mentioned earlier, DERs can cause some potentially undesirable outcomes. They can potentially drive up costs. They can also save a lot of costs. In the world where now we’re procuring less centralized generation, now we’re building fewer wires to meet demand, we’re saving a lot of costs. Rates are one of the key levers to make sure we get the good world and not bad world. Generally speaking, rates are regulated at the state level. It’s not entirely straightforward. In the structure of the power system you have, in many locations—not in all locations, but in many locations—you have competitive markets at the generation level. Different generators competing with each other to provide power to people who want it. Those generators, those competitive markets, they’re called wholesale markets for power, are typically regulated at the federal level. The Federal Energy Regulatory Commission regulates these wholesale markets. That obviously impacts rates. Because these markets are basically establishing the price that consumers need to pay for power in the regions that these markets exist. But typically, rates, the way we think about it for homeowners, they’re regulated at the state level. You have state utility commissions, public utility commissions, or public service commissions they’re often called, and they effectively determine how much costs some of our distribution utilities should be able to recover. These are the companies that own the wires that deliver power from power plants to your home. They also determine the structure of the prices, the structure of the rates, that you pay to consume that electricity. When I say structure, I mean, is this a charge that is per unit of electricity you consume? Is this a charge per maximum amount of demand that you had over the entire month? Or is this a charge that’s simply given to you at the end of the month that is independent of how much you consume or don’t consume, or a customer charge? That is almost always regulated at the state level.

KL: Could you talk a little bit about the problems with today’s rates?

SB: The overarching challenge is really that rates don’t align well with the economics of the power system in general. I know that’s a really esoteric comment but I’ll talk a little bit more about what I mean by that. In reality, the cost of energy, or the cost of, electricity really, can change dramatically by orders of magnitude throughout time. Throughout any given day or throughout different times in the year or between different locations on the grid, the power grid. This is due to basically the physical realities of how power flows over the networks that we use to distribute power from where it’s generated to where it’s consumed. There’s two or maybe three important factors for that. The first is that power lines can only really transmit so much power at any given time. The function for how much power a line can transmit is quite complicated, but you can think of it effectively as that sometimes they get full. You can’t really push more power from point A to point B. When that happens, if I want to consume power at point B, but the generator is located at point A, or the cheapest generator is located at point A, I won’t be able to move more power from point A to point B. What I might have to do is turn on a more expensive generator somewhere else. Maybe a more expensive generator is located at point B. What that does, what that dynamic does, is it creates different prices for power at different points in the system. If I consumed another unit of energy at point A, that’s great, because we can generate another unit of energy at point A and we don’t have to use the more expensive generator. But if I now want to consume another unit power at point B because that power line is, you can think of it as full, we’d have to turn on this more expensive generator. That’s really what’s called congestion in the parlance of power systems. There’s another aspect to it, too, which is that when you’re transmitting power over power lines, you lose some of the power along the way. Maybe the easiest way to think about this is like friction. If I kick a soccer ball, the soccer ball slows down as it rolls across the field. You can think of losses in the power system in the same way. If I generate a certain amount of power at point A, slightly less power reaches point B because we lose a little bit of it along the way. From an economic sense, imagine that due to losses I had to generate two units of power at point A to deliver one unit of power to point B. If, for example, it cost me five cents per unit of power to generate electricity, I would have to basically pay 10 cents at point B to consume power. Because I have to generate two units at point A. Those are what we call losses. Basically, due to congestion and losses, you can have pretty dramatic differences in the actual cost of energy at different points in the system. Now, rates almost universally, in the United States and many other developed power systems, places like Europe and Canada, the prices you and I pay for electricity don’t change over time or location. They’re pretty much constant. I pay, for example, 10 cents per kilowatt hour no matter when I consume it. No matter what time of day it is, no matter if demand is high or demand is low, no matter if I’m in a rural location or an urban location. As long as I’m served by the same utility, I pay the same amount of power.

Now, there’s a final complication with rates that I think is really, really relevant for the distributional conversation we’re having. The conversation about what are some of the potential deleterious impacts of rooftop solar. That’s that we allocate a lot of the cost of the system in charges that don’t actually reflect the reality of those costs. There’s a particular group of costs in the power system that economists call residual costs. I’ll talk about what I mean by this because it’s pretty wonky and technical. Basically, I just described how because of the physical realities of how power flows over these networks, you can get differences in prices of power across the system. In theory, you can think of the utility that’s operating that network as taking some of those rents associated with the differences in the prices of the power across the network. If at point A the price of power is five cents and at point B it’s 10 cents, you can think of the utility as buying power at point A and selling it at point B. They buy it for five cents, sell it for 10 cents. They’ve just made five cents off of basically the value of transporting that energy across the network. By doing that, the utility accrues a bunch of revenues. The residual costs are what you might expect them to be. Residual means left over. The residual costs are all of the costs of the system that are left over after the utility has accrued all of those rents, has made all of that revenue from charging these short run marginal costs, as we call them in the power system.

There are a set of costs that are associated with an operating system. Sometimes these are infrastructure costs. There’s the cost of networks, there’s the cost of generation, et cetera. Other times these are actually just the costs of policies or programs that, for example, a political body or the regulator has said that the utility has to engage in. These might be support costs for supporting energy efficiency programs or renewables investment programs, et cetera. In all of those costs, these residual costs can’t really be attributed to any one individual’s behavior. Because they’re not derived from the fact that you wanted to consume one kilowatt hour now. They’re just all the costs associated with the system. Typically, what we do in the power system is we allocate those on a dollar per unit of electricity basis.

That has a couple of problems. I’m going to dive into the problems now. There are really two key problems. One is what we might call a substitution effect and the other is a redistribution effect. I think the redistribution effect is really maybe the most important thing for this conversation about, again, these distributional impacts. But I’m going to touch on the substitution effect first. The substitution effect is that because these rates or the prices for power don’t actually reflect the true value of operating and installing a distributed energy resource like rooftop solar in any given location, we’re misconstruing how much value you can create by installing one of these systems and also misconstruing how much we should be paying for those systems. What this means is that in some locations, we’re effectively overvaluing how much we should be paying for rooftop solar or other types of distributed energy resources. Meaning we’re encouraging people to install resources in those locations even though they’re not necessarily actually reducing system costs or anything like that. That’s maybe a bad thing because you’re basically causing a substitution of maybe lower-cost resources for some of these higher-cost resources. Remember that rooftop solar is maybe three to four times as expensive as utility-scale solar and the same trends tend to hold for storage and other resources like that. Then in other locations, we’re actually undervaluing the potential value of distributed energy resources. There might be some really high-value locations and we’re not telling developers or other businesses or homeowners that if you install the distributed energy resource here, you could really create a lot value and we’d want to pay for that value because it’s going to reduce the cost of the system pretty dramatically. Again, maybe in that case we’re actually substituting a higher-cost resource where there could be a lower-cost resource. This is the substitution effect. Because these inefficient rates are there, we’re potentially driving in more expensive resources than we otherwise would like to. From a climate perspective, this is a bit of a tangent, but from a climate perspective, the overarching question is, what are you substituting for? If rooftop solar is substituting for a coal generator or a natural gas generator, then maybe that’s okay. We’re willing to pay a little bit more because these resources have these climate benefits. But if the rooftop solar system is substituting for a utility-scale solar system, meaning, it’s now generating one kilowatt hour from a distributed solar system, and we could have been generating one kilowatt hour from a utility-scale solar system, and we’re paying more for these resources than we otherwise would, we’re actually just making the cost of decarbonization more expensive. That’s potentially a challenge, that substitution effect. That’s really why we tend to care about these costs.

The other big issue is the redistribution effect that I mentioned. Because we’re recovering all of these residual costs, these costs associated with the infrastructure of the system and taxes and policies. Because we’re recovering these residual costs in dollar per unit of electricity consumed charges or dollar per kilowatt-hour charges, I can actually avoid paying for those costs by, for example, installing rooftop solar. I install rooftop solar, now not only do I consume less electricity from the grid, so I’m buying fewer kilowatt hours from the grid, I’m also avoiding paying for many of these residual costs. Because, remember, these costs are recovered in dollar per unit of electricity consumed, I’m now consuming less electricity from the grid, I’m paying for less of those costs. But remember these costs, these residual costs, they don’t go away. The costs associated with all the existing infrastructure, even potentially, like I said, policies, taxes. Important costs. The way I like to think about this is, imagine that I’m a homeowner and I live in an area where there’s a private school option and a public-school option. My property taxes, many of them go to paying for the public infrastructure. They go to paying for the public school and I decided to send one of them to private school. The costs of the school still are largely there. We still have to pay for the building, we still have to pay for the teachers. The one additional student that’s gone doesn’t really change dramatically the cost of the school. We would expect my property taxes to more or less stay the same. I would pay roughly the same amount. You can think about residual cost in the power system in the same way. The cost of the power system exists, they’re there. If I reduce marginally how much I am now consuming from the grid, that doesn’t necessarily change these overall costs. What that means is when I pay less, someone else pays more. I’m now consuming less electricity from the grid; I’m paying less towards the recovery of all of these residual costs. The utility and the regulator have to raise costs on everyone else to make sure that basically all of these costs are recovered. Now, remembering these trends, these income trends of DER adoption, what we know is that mostly the adoption is happening among wealthier individuals. Now wealthier individuals are contributing less and less to these shared costs and those costs need to be recovered. When costs are raised, lower-income customers tend to bear the brunt of that. That’s the dynamic that we modeled, that we assessed. We looked at a dataset of consumption and costs from the Commonwealth Edison service territory in Chicago. We modeled out solar adoption, taking into account the different income trends of adoption. Knowing that wealthier people are more likely to adopt solar than lower-income folks. It’s not that lower-income folks don’t adopt solar, it’s that they’re much less likely to adopt solar. We modeled this out and what we saw is that actually at relatively high penetrations of solar PV you could see really dramatic cost increases for lower-income customers. We saw for the bottom 20% of incomes as high as 80% increases in their annual expenditures on electricity as rooftop solar got some pretty substantial penetrations. These are some of the effects that people are really concerned about and some of the problems associated with these rates.

KL: You mentioned Chicago, but are there other areas of the country where we could see this disparity becoming really obvious first?

SB: This is going to be a problem pretty much anywhere that you have these inefficient prices. That’s almost everywhere in the United States. There are places that are trying to do better. New York has a really concerted effort right now, their Reforming the Energy Vision regulatory proceeding. They’re trying to move from a world where they’re paying rooftop solar adopters based on simply the average cost of the system to a system where they’re actually paying rooftop solar systems for the value that they create. They have this Value of Distributed Energy Resources proceeding and they’re trying to really get granular about how they’re doing it. But by and large most utilities around the country still have these inefficient rates. There are some good reasons for that and some bad reasons for that. Anywhere where you see this this kind of dynamic, and you have relatively high penetrations of rooftop solar, you’re going to start to run into these problems. We saw the problems starting to get really pretty bad around 25% of single-family homes having rooftop solar. That’s actually a penetration level that we see in some pockets of the U.S. today. Hawaii for example, a lot of homes there have solar. Now Hawaii is a bit of the unique example because the cost of electricity for anything else is really, really expensive. They have to burn oil and other stuff like that. But in certain parts of California, penetrations are very high, especially in the major urban areas. This could really start to be a problem in a lot of places around the U.S. relatively soon.

KL: What about solar in countries where the grid is just not going to get there anytime soon?

SB: That’s a great question. I think it’s important to draw a distinction between the types of conclusions that we’re drawing, which are really focused on developed power systems. Places where we have universal connection basically. Any customer that wants it in the United States has a connection to the grid. What has been emerging as a potential incredible life-changing, world-changing even, benefit of solar—particularly rooftop solar or maybe distributed solar—is that it’s actually a vehicle for providing access to a basic electricity services in locations where, or in countries where, maybe there’s not that universal connection. Places in really many parts of the world, various countries in sub-Saharan Africa and India and Indonesia and many places where the access to the more centralized grid is non-existent or very challenging because these communities are really remote. Solar home systems, panels that go on people’s homes and provide basic cell phone charging or maybe a fan or a radio or something like that, really can be life-changing. We’ve seen pretty incredible proliferation of those solar home systems and also now emerging are mini-grids. Whereas in the United States we basically have, across the entire United States, it’s actually pretty incredible, across the entire United States we have three grids. There’s a grid in the western United States, a grid in the eastern United States, and a grid in Texas. Texas has its own, it’s awesome. We have these three interconnections that cover this massive territory. That’s what you might consider a bulk grid. Mini-grids are a grid that serves 200 homes, or something like that. It’s a self-contained small system. Those mini-grid systems are now being used in certain parts of the world. Connect communities and provide electricity where these bulk grids don’t exist. That’s the work that some folks at the Energy Initiative are doing. Ignacio Perez-Arriaga and Rob Stoner, among others, are pioneering this work using electricity planning models to show that mini-grids, solar home systems, can be by far the most cost-effective way to provide a basic amount of electricity service to homeowners or individuals in many of these areas that lack access today. The story that we’re telling about some of these potential distributional impacts are really much more relevant in places like the United States where we have universal connection. It’s certainly not a universal story.

KL: Absolutely. Scott, what are some positive models of low-income community solar? I think there have been some successful programs mentioned in Massachusetts and California. Could you talk about those and provide some paths forward?

SB: Sure. Yeah. So I guess there’s… Community solar is one potential path for expanding access to solar to lower-income communities. If you think about the problem that we highlighted, the problem is we’re recovering these costs of the grid inefficiently. Now when I adopt solar, I pay a little bit less towards the total cost of the grid. Because solar adopters tend to be wealthy, the net effect is a shift of total cost from high-income to low-income customers. One way that you can try and solve that is increase access to solar among those low-income communities. That’s what a lot of community solar programs are trying to do. They’re recognizing this potential problem and saying, we’re going to expand access to solar in these communities. You mentioned there are state programs in Massachusetts and in California. There’s actually a recent announcement just yesterday. Sunrun announced a new program with East Bay Community Energy where they’re going to try and deploy solar and storage across low-income, single-family, and multi-family homes in the East Bay in California. There are a lot of these programs that exist. I think they’re great and I think they are trying to solve the problem within the system’s existing set of constraints. I think the ultimate problem with some of those programs is that they’re a band-aid. They’re not solving the underlying challenge of these inefficient rates. What we saw in our study is that it’s not just high-income to low-income cost shifts that’s happening. It’s really solar adapter to non-solar adapter cost shifts. There are cost shifts between solar adopters and non-solar adopters in the low-income communities as well. These programs, these community solar programs, are excellent but they might just exacerbate some of those cost shifts from some low-income customers to other low-income customers. Unless you have universal access to rooftop solar or universal installation of rooftop solar in these low-income communities, you might still end up with some customers that are left out and experience higher expenditures as a result. I should also mention that there are other types of programs to support low-income customers and get the basic amount of electricity that they need to live a modern life. These are federal programs like the low-income heating, or home energy assistance program, it’s called LIHEAP, or state-based programs like CARE in California it’s called. Many of these programs are designed to provide discounts on electricity for lower-income customers or provide rebates. Trying to make it more affordable for low-income customers to have these basic energy services. Those are excellent programs but there are some challenges associated with that, in that the enrollment is often quite low. There was a review of the enrollment in LIHEAP. This is the federal program to support low-income heating; home and energy assistance. What this review found is that only 22% of eligible customers were actually enrolled in LIHEAP. There are lot of low-income customers that weren’t receiving these benefits that could have been or maybe should have been. That could be due to a number of factors. Maybe it’s due to the fact that it’s difficult to enroll. There are transaction costs associated with actually getting access to the program. It could be due to the fact that information about the program is not widely disseminated. People just don’t know that it exists. While I think those programs are great and I laud them and think in many cases they should be expanded, it’s maybe not the best potential solution because a lot of people aren’t actually getting access to these programs that should. I should mention, we brought up the topic of energy property earlier today, this is not a distant future problem. This is a problem that’s really present for Americans today. There was a study from the Energy Information Administration, the EIA here in the United States. They do these surveys of home energy consumption every handful of years. They were analyzing data from the most recent survey. They found that roughly a third of Americans, about 31-33% of Americans, experience some sort of hardship in paying their energy bills. Maybe they either don’t heat their home in the way they would like to and they’re cold in the winter or maybe they forgo other types of expenses in order to pay their electricity or their energy bills. That’s really what we’re talking about when we talk about energy poverty. We’re talking about people not being able to afford to pay for some of these basic energy services. If you trust that data, that is really a pretty prevalent problem in the United States and I think when we’re talking about rate reform and making sure that some of these really cool new potentially impactful energy technologies don’t exacerbate that problem, that’s really what we’re concerned about.

KL: Scott, we’ve heard a lot of talk about the Green New Deal and how it relates to energy justice. How does your research play into that?

SB: That’s a great question. The Green New Deal is obviously very concerned about ensuring that historically disadvantaged communities benefit from the energy transition or the transition to a lower carbon or a more sustainable energy system. What we are really thinking about in this research when we’re thinking about how does some of these new energy technologies like rooftop solar impact some of these disadvantaged communities, I think it really is relevant findings that maybe some of our intuition about how these technologies are going to impact people isn’t actually reflected in reality. I think it’s particularly important as there are certain candidates for the presidency, at least in the Democratic side, that actually support national scale rollouts of some of these programs. For example, at different points in time, Bernie Sanders and Elizabeth Warren have claimed that they support national implementation of net metering. Which is, again, that program that basically guarantees that you get paid the same amount for generating electricity as you do as you pay to consume it. It’s one of the major support policies for rooftop solar in different parts of the country. I think that they’re incredibly well-intentioned. They basically want to expand access to and the deployment of clean energy technologies. But what we’re finding is that potentially that method or that program, that way of supporting clean energy technologies, has these potentially adverse distributional impacts. We’re finding that net metering does have this potential impact on low-income customers that is really hard to solve unless you fundamentally redesign how you’re paying for the electricity.

KL: How does technology like Nest, where people can tailor their energy consumption, factor into all of this?

SB: I think this question about new energy technologies, technologies like Nest and all of that, is the biggest factor driving a lot of the changes that we’re seeing today. Historically, customers didn’t really have very many options for how to respond to electricity. We also didn’t have very many options for understanding how much customers were actually consuming. We’ve had electricity meters, obviously, for a very long time but historically these meters have been kind of dumb. They can only tell you that over some period of time, maybe over the course of a month, you consumed a certain amount. Now we have really advanced technologies that can say, you consumed exactly this amount at exactly this time and here is exactly your peak demand and how that’s changed over time. That’s advanced metering infrastructure. Now we have all of this infrastructure to understand how much customers are consuming so that we can communicate that to customers and because of technologies like Nest or rooftop solar or home energy batteries or even potentially electric vehicles and all this cool stuff that’s proliferating, and maybe it’s all connected through your Amazon Alexa or your Google Home or something like that, all of these automated and flexible devices are giving customers a lot more options to respond to energy prices. My favorite example of this is in 1977 when Jimmy Carter wanted people to save electricity, he encouraged everyone to put on a sweater. Demand-response in the 1970s was wearing a sweater when it was cold. Today demand-response is millions of devices acting in an automated fashion to turn down heating or cooling demand during the highest demand days of the year to ease the load on the grid and doing all of these really cool sophisticated things. It has again potential to create massive benefits because we can really optimize the architecture of the grid. We can make sure that customers are saving money and we’re not spending too much on infrastructure that we don’t need. But it can also potentially create some of these challenges that we’ve highlighted where if we install these things and operate them in a sub-optimal way or maybe in a not coordinated or not well thought through way, we can actually drive up costs of the system. It creates this massive upside potential but also this potential downside that we want to avoid.

KL: What are you and others proposing for solutions to these challenges?

SB: There are a lot of folks that are concerned about this and that are working on this. I think there’s a general recognition of the need for rate reform. This is basically just changing how we pay for electricity on the grid. I like to take a step back and think about this in the context of some of the broader challenges that we’re facing in the power system. People are like, rates, what does it really matter? It seems esoteric and wonky and who really cares. I want to take a step back and say, I do think this really matters. When we’re thinking about the scale of the challenge of decarbonization, by some estimates we need to be spending a trillion dollars a year on new energy infrastructure to avoid some of the worst impacts of climate change. Every dollar counts. If we’re, like I mentioned earlier, maybe over-incentivizing some of these more expensive resources where we could be installing less expensive resources, we could be potentially wasting some very scarce societal dollars on inefficient ways to decarbonize. I really think it’s important that we are align how we’re paying for the services that these distributed energy resources provide with the economics of the system because it could ultimately be a boon for decarbonization and for the broader decarbonization effort. There are really two components to that. There is, first of all, actually aligning the marginal costs. The cost per kilowatt hour that you pay to consume electricity or that you’re paid to generate electricity with the overall economics of the system. We have pretty good models. We have pretty good economic models for understanding how much energy is worth at different points in time and different points in the system. These are facts that are not unknown to regulators and utilities and even some of these distributed energy companies. We know relatively well how much electricity costs. We should be charging customers based on those actual costs. Then, rather than recovering all these leftover system costs, these residual costs, in a way that really distorts how people are behaving, we should be recovering those costs in a way that actually minimally distorts how customers are behaving. There are a couple of different ways you can do that. You can recover those through fixed charges. Dollar per customer charges. The basic idea is that, because it’s a dollar per customer charge that doesn’t change no matter how much I consume, you’re not distorting my decision to install solar or decision to install storage or something like that. The classic concern that people have had is that if you take those costs and you “peanut butter spread” them across everyone and everyone gets the same dollar per customer charge, that could be really regressive. The reason for that is low-income customers tend to consume less electricity than higher-income customers. If you’re moving from a world where you’re recovering a lot of cost based on total volume electricity to a world by you’re just recovering all these costs no matter how much electricity you consume, you’re actually going to be raising costs differentially for lower-income customers. People are really concerned about that. What we’ve shown in our research is there’s actually really, really simple ways that you can change the design of fixed charges to ensure that they’re not regressive. Actually, you can make them progressive rather than regressive. There are basically a few different ways you can do that. You can look at the customer’s demand. You can look at the data and you can say, we know that these characteristics of demand, of electricity demand, look a lot like low-income customer characteristics and these characteristics of demand look a lot like high-income characteristics. You can use really advanced techniques if you want to do that, you can use machine learning and AI to identify it, to cluster customers into these different income buckets and do really advanced cool stuff. If you want you can come pay us to do that at MIT. I’m just kidding but only kind of. Or you can do actually really relatively simple methods like what we’ve shown. Once you’ve identified who you think is low-income and who you think is potentially high-income you design charges to make sure that low-income customers get lower charges. It’s relatively straightforward. You can also do that based on geography. Unfortunately, we live in a society where in many cases, a certain neighborhood will have a very high percent of low-income customers. You can design charges such that that area has a lower dollar per customer charge and the areas that have predominately high-income customers have higher dollar per customer charges. There are some challenges associated with all of these different ways of influencing it, but that’s one way you could do it. The last way you could do it is you could implement a system that looks like a progressive income taxation. You could look at someone’s income and this might be anathema to think of that the utility is going to be understanding what your income is, maybe you give them your social security number or something like that. That might seem scary to people, but it’s one way to do it. You could effectively have charges based explicitly on different customers income or different customers wealth. The long story short is there’s a lot of different ways that you can design these fixed charges that are progressive rather than regressive. That’s what we showed in our research. There’s one other thing that I think is really, really important. This goes without saying. It’s really important to value, the end of the externalities that are not currently valued. What I mean by that is to price carbon and other types of health impacts. When we generate electricity from fossil fuel resources, we’re not only generating carbon pollution and other types of greenhouse gases, we’re also generating other air pollutants that have health and other associated impacts. We should be valuing those. We should be pricing it based on the full social marginal cost, that’s the economics term for this. There are a lot of different ways you could do that. The classic economist argument is implementing a carbon price. That’s really hard to do. It’s been shown to be challenging to do politically. They’re not very politically popular. There are other ways that you can do that. What we’ve seen become much more popular is clean energy standards. Saying, X percent, maybe 80% or 100% of your electricity has to come from low- or zero-carbon resources. Those types of programs are great because they effectively incentivize the good as opposed to penalize the bad. That tends to be much more politically popular. I think it’s important to know that as the grid decarbonizes, it’s going to be important to structure those programs, those clean energy standard programs, around the amount of carbon that’s being released at any given time. That’s a nuanced design recommendation. All of these things impact the rates and how much we’re paying for electricity and how much we’re paying for generating electricity. There are really important solutions to this overall challenge of ensuring that these distributed energy resources deliver the benefits that they promised to deliver.

KL: We solicited questions for you from audience members before this recording. I’d like to bring those up to you now. One of the questions is regarding building integrated photovoltaics. This person is interested in resource recommendations and points to consider. Could you offer those?

SB: I think building integrated solar is a really exciting area. I won’t point to any equations today. But I will maybe provide some intuition for different areas for researchers in industry to push on. I have a paper with several colleagues, Jesse Jenkins being one, Ignacio Perez-Arriaga being the other, and Sam Huntington being the fourth author on that paper, in which we looked at the conditions you need to see to see distributed energy resources, small-scale solar or storage, be ubiquitously more valuable than utility-scale, large-scale solar or storage. How are we going to make a system such that the small-scale systems are always more economic than the large-scale systems? There are two factors that drive the consideration of which one is basically better for society. There is, first of all, how much additional value are you creating by the nature of being distributed? How much additional value are you creating by generating electricity close to demand? Reducing losses across distribution networks, reducing congestions. The things we talked about earlier. That creates value and, in many cases, that value can actually only be captured by being located in the distribution system. There’s additional value side of things. There’s also the additional cost side of things. Traditionally, at least in the United States, although this is not necessarily universally true in other countries, these distributed systems have cost a lot more on a dollar per unit basis than the centralized system. I mentioned this earlier. Rooftop solar in the United States is on average three to four times as expensive as utility-scale solar. You’ll want to have more distributed solar where the incremental cost is less than the incremental value that you get out of installing that system. Building integrated solar is one way to basically take the incremental cost down to zero, or down to a very low level. Because effectively you’re putting solar on the building as a byproduct of simply building the building. You have solar integrated into the windows or solar integrated into the roof. That could basically eliminate the incremental cost of installing distributed solar because, again, you’re just building the building. You’re building the system. If you’re thinking about building integrated photovoltaic, you need to be thinking about two things: 1) How much additional value you can get. How do you drive additional value in the system? 2) How are you eliminating that cost premium relative to the other ways of generating solar electricity? Basically, how are you eliminating the cost premium relative to utility-scale or larger commercial-scale systems?

KL: Another listener asked about recent research using a solar panel to collect water and generate energy. They want to know whether this is feasible and scalable.

SB: That’s a great question. I think it ties into this broader theme. There are a lot of cool things you can do once you have really cheap solar electricity. Generating water is one. There’s a startup called Zero Mass Water that’s commercializing these systems. The answer to the question is, yes, it’s possible, it’s feasible, there are people doing it today. There are really a whole host of other things that you could do with really low-cost solar electricity. You could potentially create hydrogen or other types of liquid fuels or gaseous fuels from solar. Now you’re creating renewable fuels. We could start to replace jet fuel or maybe marine fuel from fuel that we’ve generated from solar electricity. That’s really cool. As this technology continues to expand and develop at the exponential rate that it has been, or maybe the very rapid rate that it has been. It always irks me when people use exponential in a not precise way. As solar technology continues to develop incredibly rapidly and costs continue to fall and engineers and entrepreneurs continue to develop innovative solutions for how to use that solar electricity, I think we’re going to see a proliferation of uses, whether it’s generating water or other types of potentially useful outputs.

KL: Thanks so much for joining us today, Scott. I really appreciate it. It was great talking with you about this. People can follow you on Twitter @burgersb. Where else can people keep up with your work?

SB: You can follow any papers that I write. I have a Google Scholar page and I have everything linked there. We’ll also link to some of the papers in the show notes, I assume. But, yes, I would love to engage with anybody that’s interested in this type of research.

KL: Show notes and links for this episode are at energy.mit.edu/podcast. Share your questions, comments, and show ideas with us on Twitter @mitenergy, and subscribe and review us wherever you get your podcasts. From the MIT Energy Initiative, I’m Katie Luu. Thanks for listening.