Skip to content

Putting Solar in All the Wrong Places

High retail electricity prices, not economic value, are driving U.S. investments in rooftop solar.

If you were starting from scratch and could install the United States’ 22,500 MW of rooftop solar anywhere in the country, where would you put it?

One approach would be to put it in sunny states. Using this criterion, Arizona, Nevada, and New Mexico would be particularly good candidates.

Another approach would be to put rooftop solar in places where it offsets coal. Using this criterion, Midwestern states like North Dakota, Minnesota, and Wisconsin would be the most attractive.

So which is it? Mostly in sunny states? Mostly in coal states? A combination of the two?

The answer is none-of-the-above. Sun matters, but it is not the primary factor driving solar installations, and it certainly isn’t coal.

What drives rooftop solar in the United States is retail electricity prices. When you look at where solar is installed, it is all in the places with high retail electricity prices. But those high retail prices do not reflect high avoided costs for the system, or for society,  as Severin Borenstein and Jim Bushnell show in a recent paper.

Where is the Solar?

The map below shows rooftop solar capacity by state as of November 2019. This includes all “small-scale” (<1MW) solar installations, including both residential and non-residential, and to make these numbers comparable, I put everything in terms of capacity per capita.


As of 2019, the 22,500 MW of U.S. rooftop solar equates to 0.07 kW per capita. But the variation across states is large – from almost 0.50 kW per capita in Hawaii to less than 0.01 kW throughout most of the Southeast.

Top Ten

By this metric, the top ten states are:

  1. Hawaii
  2. Massachusetts
  3. California
  4. Arizona
  5. Vermont
  6. New Jersey
  7. Connecticut
  8. Nevada
  9. Rhode Island
  10. Maryland

On a per capita basis, Hawaii has twice as much rooftop solar as any other U.S. state. Hawaii is on the bleeding edge of renewables integration, with a goal of 100% renewables by 2045, not to mention great sun.

More surprising is the Northeastern states. Did you really expect Massachusetts to edge out California? I didn’t. Yet six out of the top ten are in the Northeast. These states rank near the bottom of all U.S. states for solar resources, and rely less on coal than most Midwestern states.


Net Metering

But, of course, it is easier to be green when it saves you money. The map below plots average retail electricity prices by state for November 2019. These prices are for all retail customers including residential and non-residential.


Look familiar? The pattern actually looks a lot like the pattern for rooftop solar. Prices don’t explain all of the variation in rooftop solar, but they explain a lot.

Hawaii is again number #1. Most of Hawaii’s electricity comes from oil, so Hawaii has by far the highest retail electricity prices of any U.S. state. It is not a coincidence that the state with the highest prices also has the most rooftop solar.

California? High prices and high levels of solar. Massachusetts? High prices and high levels of solar. Vermont? High prices and high levels of solar. You get the point.


Sometimes Correlation is Causation

I know, I know, correlation is not causation.

This pattern could also reflect “green” policies or tastes. New Jersey (#6), for example, offers 0% sales tax and other state-level incentives for rooftop solar.

Or it could reflect barriers to solar in less green states. For example, Florida and eight other states have restrictions against third-party power purchase agreements, which discourage solar leasing.

But come on! Isn’t it easier to convince yourself to do something good for the planet when it is also good for your bank account?


Too Much Incentive to Go Behind the Meter

This wouldn’t be a problem if retail prices reflected societal costs. But they don’t. In states like California and Massachusetts, the price consumers pay for electricity is much higher than social marginal cost.

This creates too much incentive for rooftop solar. In these states solar is a great private investment — it just is not a great social investment.

Can you really have “too much” rooftop solar? Yes! We’d all like to reduce carbon dioxide emissions, but these resources would go a lot farther if they were instead used for grid-scale renewables.

In these states with high electricity rates, one way to lower the per-kWh price would be to increase monthly fixed charges. This type of rate reform is not a panacea, but it would lower the overpayment for rooftop solar, reduce cost shifting onto non-solar customers, and encourage electrification.


Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.

Suggested citation: Davis, Lucas. “Putting Solar in the All the Wrong Places” Energy Institute Blog, UC Berkeley, February 3, 2020,


Lucas Davis View All

Lucas Davis is the Jeffrey A. Jacobs Distinguished Professor in Business and Technology at the Haas School of Business at the University of California, Berkeley. He is Faculty Director of the Energy Institute at Haas, a coeditor at the American Economic Journal: Economic Policy, and a Faculty Research Fellow at the National Bureau of Economic Research. He received a BA from Amherst College and a PhD in Economics from the University of Wisconsin. Prior to joining Haas in 2009, he was an assistant professor of Economics at the University of Michigan. His research focuses on energy and environmental markets, and in particular, on electricity and natural gas regulation, pricing in competitive and non-competitive markets, and the economic and business impacts of environmental policy.

75 thoughts on “Putting Solar in All the Wrong Places Leave a comment

  1. Lucas, I wouldn’t put rooftop solar anywhere in the country, but it’s not up to me. From a policy standpoint, it should be the option of individual homeowners whether to install it or not.

    I would, however, abolish net metering. The idea compensating home solar owners at retail electricity rates provides a net benefit to ratepayers or society is demonstrably false, for several reasons:

    • Cross-subsidization raises prices for non-solar ratepayers, among them the most financially-vulnerable customers;
    • Mid-day overgeneration is currently a physical and financial burden on the grid resulting in negative pricing. Solar owners who cannot curtail their generation should rightly bear all costs of negative pricing, not be rewarded for it;
    • Retail price compensation of home solar creates a non-competitive price advantage over other sources of clean generation (nuclear, large hydro).

    As many studies have shown, wholesale prices for utility solar are not indicative of retail prices, which includes many ancillary costs for all intermittent energy sources: frequency regulation, voltage regulation, backup generation, spinning reserve, etc. etc. And empirical evidence bears it out: in every location where utility-scale solar has been introduced in the U.S. retail prices have risen – without exception.

    So here’s what I would do: eliminate utility scale solar, eliminate net metering – there’s no justification for pawning off the added cost of intermittent renewable energy on anyone. But if home solar owners want to bear those costs, they should have every right to do so. Since you asked.

    • Since retail rates have risen everywhere in the US over the last decade, that’s not an appropriate standard.

      Additional costs from solar have not been substantial. Please provide a study done within the last 3 years that demonstrates that point (not an obsolete study from 2011).

      As for rooftop solar costs, the value of avoided transmission is substantial. While residential rooftop may not be completely economic, commercial solar installations are very cost competitive for this reason. I have previously provided citations showing these findings.

    • The cost of electricity has steadily increased over the long term, with or without solar generation. provide some study or justification for this erroneous claim.

      • actually many places electric prices have DROPPED due to low cost of nat gas….but liberal states…can’t allow that….so they have adders!

          • The EIA does not include the production subsidies paid to renewable energy by the taxpayer. The impact of forced mandates is also difficult to accurately assess. I suspect that the use of low cost natural gas is the real driver behind lower energy costs.

          • The issue I answered was a claim about retail rates. Retail rates contain all sorts of distortions, of which fossil, nuclear and renewable subsidies are but one type. I’ve already answered the issue of subsidies here. You are trying to move the goal posts.

          • I’m not sure how pointing out that your comment was off topic was being a “sole arbiter”. Moving the goals posts is a common tactic by Internet trolls and that behavior should be called out to clearly set boundaries.

      • No study necessary – since 2008, California retail electricity prices have risen seven times as fast as the U.S. average

        and solar and wind are solely to blame. Talk to any grid engineer – wholesale prices don’t begin to tell the story. Solar and wind are expensive to integrate, to balance, to regulate, to back up, to store. To prevent midday excess solar from taking down the CAISO grid, CA must either pay other states to take our excess electricity (“negative pricing”) or pay utilities to curtail (disconnect) their solar farms. In 2017, California paid ~$1 billion to get rid of excess renewable electricity.

        The Paradox of Declining Renewable Costs and Rising Electricity Prices

        “Wholesale prices are paid to the generators; the retail prices paid by final customers reflect the full cost of delivering electricity. Generation, though the largest component, only accounts for 44 percent of the total cost. The other main costs affected by renewables integration are transmission and distribution of electricity to its point of use, reliability costs to maintain stable voltage and frequency, maintenance needed to keep the system running, depreciation and taxes.”

        • You are confusing the increase due to direct procurement costs (which PG&E in particular has mismanaged as I show here: with the OPERATIONAL costs of integrating renewables. The latter costs have not changed significantly based on the annual review that I conduct on behalf of clients at the CPUC. The other source of the large increase in rates is in distribution and transmission rates, again a different story.

          • You are extrapolating a seven-fold increase in CA vs. US electricity rates based on purported mismanagement by the utility providing 39% of CA electricity. Nice try, but not really.

            “The other source of the large increase in rates is in distribution and transmission rates, again a different story.”

            Not a different story at all. In proportion to energy generated, transmission necessary to accomodate renewables is orders of magnitude more expensive than that for centralized generation. Re: your critique –

            “First, it’s entirely retrospective and then tries to make conclusions about future actions.”

            All statistical projections have a retrospective basis, except the paranormal ones commonly associated with renewables / / cold fusion / tarot.

            “the renewables are now cost competitive with conventional resources…”

            Again, you turn to your tiresome apples/coconuts comparison of wholesale vs. retail costs. As the paper makes abundantly clear, retail costs of renewables are significantly higher due to fundamental limitations of intermittent energy – and other sources predict the problem will only get worse.

            “The National Renewable Energy Lab reports that a large grid system with 30 percent VRE can operate with minimal system disruption. Going beyond 30 percent, however, can present new challenges. Lawrence Berkeley National Lab (LBNL) examined three scenarios (high wind, balanced VRE, and high solar) across several different US power markets. In each scenario, ancillary services costs to maintain reliability increase substantially.”

          • I highlighted PG&E as the worst offender. SCE and SDG&E have similar problems, which account for 75% of the state’s electricity supply. Rather than bother to respond to your comments, I just invite those who are interested to see my response in my blog.It goes into more detail about the shortcomings and misassumptions in the U. of Chicago study than I list here. The U. of Chicago report makes the simplistic mistake of confusing the past with the future.
            Given that CAISO ancillary services currently cost $0.85/MWh according to its 2018 Market Performance report or 85 thousandths of a cent per kilowatt-hour, a 100% increase, a doubling which would be substantial per the LBNL study, would increase California average retail rates about 0.5%. Meanwhile, PG&E’s generation portfolio costs have increased more than 40% over the last 6 years due directly to its PPAs, and not from its portfolio operations (other than running Diablo Canyon when its losing money for the year). (I have no idea where you get that I am confusing wholesale and retail, but I have testified before the CPUC almost 50 times on both retail and wholesale rates.)

          • “Rather than bother to respond to your comments, I just invite those who are interested to see my response in my blog.”

            Rather than being diverted to your blog, I invite you to come up with a concise summary here, and save everyone much time and digging with scant possibility of reward.

            “[ancillary services of regulation and spinning/non-spinning reserves] would increase California average retail rates about 0.5%.”

            So true. Now add negative pricing, backup generation, transmission / distribution to point-of-use, and decremental payments, and you might understand why solar is ~200% more expensive than solar advocates contend.

            “I have no idea where you get that I am confusing wholesale and retail…” See above. The oft-repeated claim solar is competitive with nuclear or large hydro only holds water in a comparison of wholesale prices. Intermittent renewables have no retail value but for avoided costs of dispatchable generation.

            “….other than running Diablo Canyon when its losing money for the year…”. Source, please.

            “…I have testified before the CPUC almost 50 times on both retail and wholesale rates.”

            Elsewhere, you claim CPUC is your client. What wouldn’t lead an independent observer to recognize a conflict of interest in being invited by a supposedly-impartial judge to testify, while being paid by that judge?

          • I will not be lured into another endless back and forth where you fail to acknowledge the facts that I put forth. You and others can go to my blog to discuss the U. of Chicago study further.

            “200% more expensive” – Unsubstantiated speculation. I put forward my evidence with citation. I expect the same of you.

            “Intermittent renewables have no retail value but for avoided costs of dispatchable generation.” It appears that it is you that has confused wholesale costs with retail rates. The costs you list show up on the wholesale side.

            I previously provided my citation several times on Diablo Canyon costs in my 2019 PG&E ERRA testimony. In fact, it looks like again in 2020 Diablo Canyon is a money losing endeavor that should be closed for at least the year until CAISO prices increase.

            While I have done work for the CPUC at times over the decades, we represent a wide range of clients, none of them energy IOUs. The CPUC is an adjudicatory body that has hearings much like court trials and parties put forward expert witnesses, me among them.

          • “I previously provided my citation several times on Diablo Canyon costs in my 2019 PG&E ERRA testimony.”

            Please provide a link here. It’s easy – copy/paste. No credible energy analyst states a claim with a reference to “google / my website / my previous work/my blog post from 2008.'” Again, I don’t have the time or inclination to do your work for you.

            “It appears that it is you that has confused wholesale costs with retail rates. The costs you list show up on the wholesale side.”

            Wrong. Take your fight to Brian Murray at Forbes:

            “Wholesale prices are paid to the generators; the retail prices paid by final customers reflect the full cost of delivering electricity.”


            “While I have done work for the CPUC at times over the decades, we represent a wide range of clients, none of them energy IOUs.”

            Well now.

            You probably are aware CPUC President Michael Peevey, erstwhile President of Edison International, sketched out the terms for the closure of SONGS at a secret 2013 meeting in Poland. It sounds like CPUC represents a wide range of clients, and at least one of them is indeed an energy IOU.

            “The CPUC is an adjudicatory body that has hearings much like court trials and parties put forward expert witnesses, me among them.”

            Were you there? If so, were you serving as Edison’s expert witness, or on behalf of CPUC? It certainly doesn’t sound much like a court trial, huddling together at a hotel restaurant in the dark. But I hear it was a beautiful hotel.

          • “we represent a wide range of clients, none of them energy IOUs”
            Edison is an “energy IOU” as anyone who has knowledge of the industry knows, so I have clearly NOT worked for them. Please don’t misrepresent who I am working on behalf of as you try to sully me.

            Transmission charges including ancillary services and other load balancing are part of the wholesale market regulated by FERC. FERC isn’t allowed to regulated retail rates.

            I have provided to you personally in my replies the link several times. If you’re didn’t review the testimony then why should expect that you will look at it now. If you are truly interested, you can find it either linked in the recent comments, or you can find it on PG&E’s regulatory filings page under the 2019 ERRA case file for the Joint CCAs. And this is a comment in a blog, not a regulatory proceeding. I don’t need to produce documents for you, but I at least have given you the citation to find the analysis, unlike many of your assertions.

  2. So, if there were a carbon taxassociated with electricity generation, then electicity prices would correlate more highly with sociatal costs. And, solar and storage insillation would more closely track socital benefit. Remind me again, why do energy producing companies not have to include the cost of their polution in the price of their products?
    Bill Zwick

    • Bill, I’m with you. As we sales guys always say, “Compensation drives behavior.” The fact that we don’t impose a direct cost for pollution means that those costs are invisible to decision makers, whether the decision maker is the home owner or the utility or anyone else. To enable market forces to work better, we need to impose a cost (I like a broad-based carbon tax) that better represents the impact of carbon on our common welfare.

      • Costs attributable to pollution can be derived from added costs to build power plants. If regulations are changed to lower allowable emissions, then more equipment must be added, resulting in higher build and operating costs. Alternatively, something else must be built. In any case, there is a readily identifiable cost differential associated with an emission or discharge from the plant.

        In point of fact, the differential is visible to decision makers in the form of higher energy costs.

        For instance, if there is a mandate to use X% renewable, then new green energy plants must be built and the cost to operate the existing plants goes up (the plants run less often). The before-and-after-cost differential is the value of reducing CO2. This type of analysis requires that true and quantifiable costs (including subsidies, carbon taxes, etc.) be identified. The analysis is not that hard to produce, but you cannot rely on data from the government, universities, trade organizations, etc., otherwise you end up with meaningless results. The underlying assumptions used by the various groups are generally completely different, depending on what the group is trying to push.

        • You are correct on one way to calculate externality costs up to the point about changes in operating costs. First, operating costs are far from the only costs the must change to accommodate reduced environmental damages. I, and others, have pointed out the fallacy of relying on short-rum marginal costs as the metric for full direct economic costs, and I won’t repeat them here (and I have a more comprehensive discussion that I’m currently working on.) You need to include full long term incremental costs. But further, for GHG emission reductions, fossil fuels must be replaced in their entirety to achievea the needed reductions, so partial operational changed cannot reflect the true costs. It is the cost of alternative technologies that become the externality cost benchmark, which is the case of using solar and wind.

          The issue then is what is the best tool for reflecting those costs. You can argue that subsidies are not as good as standards or taxes or permit trading, but that’s a second order issue after agreeing that the externality needs to be addressed and then narrowing the value benchmark to either alternative technology cost or realized benefits from reduced damages. (And both have been used in the legal system for many issues beyond environmental regulation.)

  3. “Societal cost”? What complete nonsense. Nothing more than a gimmick used by charlatans whose favored product cannot compete in a free market. Just out of curiosity, is there a “societal cost meter” available somewhere?
    Solar energy should be located where the energy resource is cost effective. Further, no subsidies or mandates.

    • Glad to hear that all fossil fuel producers should return all of the subsidies that they’ve received over the last two centuries to return us to a level playing field for all technologies. Somewhere among these blog posts by Severin Borenstein is an accounting of some of the more recent subsidies.

      • So should green energy return their over $50 billion in subsidies?
        The past cannot be re-created.
        The only way subsidies should be applied is if they are evenhandedly applied. That being said, I doubt that is possible.
        Going forward, no subsidies, no mandates. Period, and yes that includes nuclear and fossil energy.

    • You might want to familiarize yourself with the work Ronald Coase did on market externalities. He spent a substantial portio of his career at that bastion of liberal thinking, The University of Chicago. Granted, this may be way beyond what you were exposed to in your Econ o.1 training. But don’t despair – the math is easy, and most of it can be explained with pictures!

      • “Externalities” invariably degenerate into subjective judgement peeing contests. As such, their value is near zero in the real world, although academics no doubt spend vast amounts of time and money debating issue. Just a waste of time, in my view.

        I do have advanced degrees in both engineering and management as well as over 50 years of experience in the energy industry.

        • If you believe that, you clearly have not lived in LA for several decades where addressing the externality of air pollution has led to a remarkable improvement in air quality. There are many, many other examples. This is a very well established branch of economics. The only ones who deny the importance of the externalities are those who are just trying to protect their own self interests at the expense of the rest of us in society.

          • You can measure ozone and NOx levels and set targets based on quantifiable analysis of health impacts.That does not require externality analyses unless there is some other agenda at work. Say for instance creating massive levels of bureaucracy whose minions will fund some particular political party who will then insure the minions receive vastly better pensions than the working folks.

          • Setting those standards to achieve environmental goals is a measure of externalities. Attempting to make a semantic distinction does not counter this fact. The only issue is what type of instrument should be used to address that externality–standards (per your example), subsidies for new technologies (per the example in this post), permit trading or emission taxes.

            As for public employee unions, environmental staffing is a tiny percentage of the entire government workforce. And the federal government workforce has been shrinking for some time. To claim that these regulations are created to increase government jobs is a wild claim that requires proof.

          • It is not semantic. Rather the issue is using data that can be readily straightforwardly quantified, versus speculated (i.e. subjective judgements).
            Mischief inevitably arises once you leave the arena where judgements and decisions are made based on what can be actually be kicked-and-counted. The whole “climate change” debacle is a good example of the mess created when feelings are substituted for clear-eyed logic.

          • Externality values are rarely based on subjective evaluations. Most are derived from quantitative data that is as valid as many of the engineering models in use, e.g., modeled 2D vs actual 3D flow dynamics. You are focused on a single element in the use of contingent valuation that relies on surveys, but those are a small segment of externality valuation that has been done.

          • BTW, do you insure our house against fire? How many times has your house burned down? How many of your neighbors or friends? Yet you likely still have insurance despite not “being able to kick it or count it”. Risks and threats are real even if they aren’t immediately present. If you’re climate change denier, then you are simply not familiar with the full set of threats, the modeling that has been done to date, the evidence of change, and the calibration of model results with actual data. I was a skeptic in the early 1990s (and you can find my publications on it then), but have now accepted that the risk is real and large.

          • These are the principles of risk mitigation and management. They are real costs that are not as concrete as you want them to be. If you don’t understand them, I’d be concerned about the management advice that you have provided. Risk management is the core of the climate change issue.

          • Classic response from those intermingling facts, opinion, conjecture, and faith.
            As one moves from facts to opinion, uncertainty grows and must be accounted for in a decision. To claim otherwise indicates a potentially deliberately hidden bias.

          • I’m at a loss. Are you saying that we should never plan for the future because we have to base those plans on opinion and conjecture? And saying that the future will be just like the past is based on opinion and conjecture–such actions are unavoidable.

          • The vast majority of climate models are not fit for the intended application. The underlying mathematics are deeply flawed. The issue lies with the fundamental structure of highly non-linear partial differential equation. You end up with vast numbers of solutions. No way to tell if any particular solution is proper. The uncertainties are so large that the results are simply not believable.

          • “Sure it works in practice, but does it work in theory?” NASA just released a study showing that the model forecasts have tracked climate metrics.

            And perhaps of greater concern is the acidification of the ocean, which could lead to ecological collapse of that food chain.

    • coal gets ~50% of all US energy subsidies. I agree, end subsidies for all energy. Tax carbon. get excited, the future of power gen will be rad and much more portable.

  4. And sometimes UCEI’s institutional bias against DG is causation. This sloppy analysis doesn’t account for high levels of distributed generation in most of the southwest, for example.

  5. At the core of this issue is a desire by customers to avoid volatile energy costs. Home ownership is desired to avoid rising rents; solar is being installed for much the same reason. Utilities now prefer to shift their risks on to customers. In California, the IOUs have overprocurred long-term contracts and PG&E is reluctant to use its bankruptcy to abrogate the most expensive PPAs. So under state law ratepayers bear all of costs of poor portfolio management. I’ve written on my blog about this problem.

    To compete with rooftop solar, the utilities will have to offer attractive rates such as long-term fixed price rates for 10 or 20 years. I wrote here in 2014 about the utilities’ lack of creativity in rates:

    As to the societal marginal costs that you refer to, I have critiqued those extensively on that site as have others, so those should not be accepted as gospel. (I’ll have a better developed response on that issue soon.)

  6. Codctemps increase pv output outputcis shoutcdouble in summer vs einter cduectonlonger sunshomev housrs.
    Outputvin summervcorrespondsctonpeak demand duevtobairccondittioning use in rescand nonnres buldings.

    Key is storagec dutinng long. Ssunnlesss. Periodds. Baatteries arr thevfrontier..


  7. One can play all sorts of economic games and come up with different answers. We can subsidize, net meter, or any of a number of other schemes. That is hardly the issue. What is the social cost of global warming? If you would like to put a number like $1 trillion per ,01 degree of wardming, you get a very different answer than all of the verbiage that you have posted here, which I think is mostly useless in the face of climate change reality.

  8. Interesting post! In our working paper “The Heterogeneous Value of Solar and Wind Energy: Empirical Evidence from the United States and Europe”, Kenneth Gillingham and I bring together many of the things you mention. Among other things, we also find that in almost all regions feed-in tariffs (variable retail rate in case of net-metering) are higher than rooftop solar’s estimated value (avoided social cost).

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: