Rooftop solar is now competitive with utility-scale power in Hawaiʻi


By Michael Roberts

We need to transition away from fossil fuels quickly, and with inexpensive renewables and batteries, we can do so in a cost effective manner, so long as we employ a sensible plan and the right policies. But what are the tradeoffs involved with how we do it?

In most places, rooftop solar is considerably more expensive than utility-scale solar and wind. In the continental United States, it probably makes more sense to build large utility-scale solar projects in the deserts of California, Nevada, Arizona, and Utah (among other places), and wind on the Midwestern plains and perhaps off shores. A roughly 100 x 100 mile remote desert area could be covered with solar cheaply and power the whole continent. Despite substantial costs involved with transmission and storage, economies of scale in such projects are so dominant that rooftop solar makes little sense compared to grid-scale alternatives. And, given the way rooftop solar has been subsidized, it mainly benefits a rich few at the expense of many.

Here in Hawaii, however, the tradeoffs between grid-scale and rooftop solar are different. Land on our remote islands is more scarce and unique than on the continent. Oahu has six times the population density of California and sixty times that of Nevada. While we have enough vacant land (mostly former agricultural land) to power Oahu and the neighbor islands with solar and wind, we may prefer to use our scarce land in other ways, such as in food production or native habitat restoration. Rooftop solar can preserve these valuable land resources.

At the same time, utility-scale solar in Hawai’i is expensive to develop, two to three times mainland costs, while rooftop solar is more competitive with mainland pricing. And now, Tesla is offering rooftop solar, installed, for just $2 per watt (unsubsidized), the same low price as the mainland. Other solar companies are likely to drop prices in order to compete, if they haven’t already. At this low price, and especially with Federal subsidies, rooftop solar is fairly competitive with grid-scale solar, at least in Hawai’i.

How much power can we potentially generate from Oahu rooftops? Quite a lot, actually. Using google maps Matthias Fripp has estimated a potential rooftop solar capacity of about 4000 MW island wide, which is easily enough to power the whole island. But this would require filling all available rooftops with solar. Rooftops of single family homes would need to provide surplus energy to help power the hotels, office buildings, and apartment buildings in Waikiki and downtown Honolulu, which do not have enough rooftop space to self generate.

Unfortunately, most of those putting solar on their roofs install only enough to satisfy their own needs, because tariffs for residents with rooftop solar, generous as they are, do not normally compensate surplus generation provided to the grid. [1] As a result, only a small fraction of the available rooftop space gets used. Our home, for example, has about 1/6 of the rooftop covered with solar panels, and we’re pretty close to net zero. To make use of more rooftop space, home and building owners would need to be compensated for the power that they can supply to others.

It turns out that a mechanism to do this may already exist. It’s called Schedule Q. Schedule Q compensates small (< 100KW) generators at the avoided cost associated with electricity provided to the grid, a requirement under PURPA. [2] Right now this compensation is just 7.68 cents per kilowatt hour, but it varies month-to-month depending on oil prices, which are low today. This roughly amounts to the fuel costs saved by HECO when more electricity is fed into the system, mainly by generating less from its largest remaining power plants.

Is 7.68 cents per kilowatt hour enough to finance rooftop solar at $2 per watt? Discounting at today’s interest rates, such as a home equity line of credit at about 4%, the answer is yes. An approximate net-present value calculation is provided in the box. Add Federal and State subsidies, plus the fact that oil prices and avoided-cost compensation are projected to rise, and the rewards could be fairly substantial. Indeed, the returns appear substantial enough that the State subsidies seem hard to justify at this point. Most single-family rooftops could probably accommodate a system several times the assumed size, which would further increase home or business-owner returns.

Note that the calculation does not consider solar power installed for the household’s own consumption, because the return for that is considerably greater under other tariffs, and is clearly worthwhile to homeowners, if not for the system as a whole.

Rooftop with solar panels
Underutilized Rooftops. The author’s rooftop has only 10 solar panels, which covers only a small share of roof space. Might schedule Q be used to make more effective use of rooftop space for solar, and thereby save land resources?

The value of installing rooftop solar and selling to HECO under Schedule Q.

Up front cost for 16.32 kW of rooftop solar at a price of $32,800.00
Annual revenue of 0.0768 x 16.32 x 5 hours per day at 365 days per year = $2,287.41
Present value of revenue over 25 years, 4% discount rate = $37,163.47
Annual monthly connection fee of $20 x 12 = $240
Present value of monthly connection fee over 25 years = $3,899.27
Federal tax credit: $8,528.00
State tax credits: $5,000*

Net Present Value: $37,163.47 – $32,800.00 – 3,899.27 = $464.20
$13,992 with tax credits.

* The state tax credit may be larger or smaller depending on whether and how the customer installs solar for its own electricity consumption, and possibly on other factors.

Under former net metering agreements, and even new solar tariffs that limit compensation for back feeding electricity, customers often save the retail price of electricity, which is typically over 30 cents per kilowatt hour, which far exceeds HECO’s avoided cost of 7.68 cents per kilowatt hour. Thus, most of the customers’ savings are not savings to society as a whole; mainly they just pass costs onto other customers.

The remarkable situation that we have come to now is that the all-in, leveled cost of new rooftop solar is comparable to the fuel costs alone from conventional generation that it would displace. This calculation does not include subsides or even count the unrecovered capital costs of Kahe and Waiau, the existing plants whose power would likely be displaced by new rooftop solar. Rooftop solar is even competitive with grid-scale solar purchased by HECO in recent competitive bidding processes, although it is a little difficult to make an apples-to-apples comparison since the recent purchases include batteries. Moreover, compensating rooftop solar under Schedule Q is unlikely to have a negative impact on other customers, as other rooftop solar tariffs do.

It is not clear to me that anyone is currently seeking credit for rooftop solar under Schedule Q. It is also unclear whether solar companies besides Tesla are offering prices low enough to make it worthwhile. If readers know of any examples, I would be interested in learning about them. While this channel for compensating rooftop solar will eventually pose some problems, as renewable penetration grows and avoided cost calculations become more complex (they would need to account for the time-varying value of energy), it might be an efficient way of accelerating renewable energy in the State. And it would do so in a way that saves scarce land resources for other purposes. Hopefully this channel or one similar to it, with appropriate adjustments over time, will enhance competition and help us make better use of rooftops for solar.

[1] Under grandfathered net-metering agreements, which are no longer available to new subscribers, homeowners receive a credit for surplus net generation in any given month that can be used in subsequent months, but any surplus remaining at the end of the year is forfeited. Under the current Customer Grid Supply tariff, customers receive a credit for electricity provided to the grid that equals roughly half the retail rate, but any surplus credit not applied to the bill in the current month is forfeited. Under other tariffs, back feeding any energy at all receives zero credit. Hawaii also has a Feed in Tariff program with very limited capacity that appears to have been fully subscribed as of 2011.

[2] PURPA is the Public Utilities Regulatory Policies Act. It is a Federal law established in 1978, revised and reinterpreted since, and requires avoided cost compensation for Qualified Facilities connecting to the grid to provide power. Schedule Q applies only to the smallest of such facilities.


10 thoughts on “Rooftop solar is now competitive with utility-scale power in Hawaiʻi”

  1. Very interesting. Why do you suppose nobody seems to be seeking the Schedule Q credits? I’ll posit lack of awareness (first time I’ve heard of this), but that begs the question of why it’s not heavily marketed huh.

    Aside, I assume you used 25 years for the NPV calculation because I think that’s the useful life of the system, but wouldn’t it be more appropriate to use the median length of stay in a residence since that’s the duration during which benefits accrue? Think that’s about 15 years or so for HI.

    1. Michael Roberts

      Hi Laron,
      Thanks for your comment. I am not entirely sure why. But some guesses include:
      1) This has not been economic until very recently.
      2) The private returns to homeowners and solar suppliers are far greater under self-supply.
      3) Getting approval from HECO for the connection (the requirements are vague) could be excessively costly. HECO may have an incentive to keep this barrier high, but I suspect that would be a relatively easy thing for the PUC to nudge along if it become a serious issue.

      As for length of stay: I don’t think this is an issue. The value of the panels would likely be capitalized into the property value. In fact, evidence in research by Sherilyn Wee suggests that solar panels are over capitalized into property values, possibly reflecting the transactions costs of dealing with HECO and the solar companies.

      1. Thanks, those all make perfect sense. Anecdotally, #3 and the associated time to approval is something we hear from every so often from property owners.

        That’s funny that you note Sherilyn Wee’s work; I actually sent the abstract to a couple of realtor friends awhile back. I had asked a handful of realtor/brokers about how PV impacts a home’s price and most said it wasn’t a factor at all or was negligible at best. That didn’t make sense to me then and it still doesn’t now.

        But anyway, I hear what you’re saying but I’d say the NPV example is still misrepresentative since it doesn’t include the marginal gain from sale at the end of the life (or stay). This could just be a difference in academic vs. business NPV methodology though. (Just to be super clear since this is the internet, I don’t mean that in a condescending way. Nothing wrong with similar methods being applied a bit differently.)

    1. Michael Roberts

      Hi Jim,

      I think this is an important question, one I ask all the time, and I’ve never received an especially satisfying answer.

      1) The cost of land may be higher here, but that should be a very small differential—this is land that has few practical alternative uses. The rental rates for land that I’ve heard suggest it can’t be that much.

      2) Solar companies complain about complexity and strict requirements that HECO places on independent contractors, and the slow processes for approval from the PUC. But when I look at the details of requirements and bidding processes, they look quite similar to those in other states getting record low PPAs (e.g., Nevada).

      3) HECO already receives a performance payment for eliciting solar PPAs below a threshold price—they get a share of the difference. Also, the PUC worked hard to approve the last PPAs in record time.

      4) I think it’s possible that new PBR will enhance HECO’s incentives, but that’s not clear to me. They must fear having their rate base in Kahe and Waiau rendered obsolete, disallowing further return on their capital (reduction from 9.5% to 0). Pertaining to this: they were able to push through fairly substantial upgrades to these plants in recent years, growing their rate base, all without any prudence review by the PUC. The PUC hadn’t realized that HECO could do this under the old decoupling rule, and it took a number of years to shut down the resulting growth in rate base. That’s a powerful incentive to slow renewable integration keep these power plants relevant. Of course, this is not something that is talked about in frank terms by very many people. My understanding is that HECO is angling for a bailout, sometimes called “securitization” where the state simply pays HECO a fixed sum equal to their remaining rate base and then finances this with much lower interest rate using state bonds. This might be a reasonable compromise, especially with today’s low interest rates, and would eliminate a potentially powerful disincentive on HECO’s part. Most utilities complain that such a payment wouldn’t be enough, which of course makes completely transparent the fact that their allowed rate of return is too high.

      State law does now require a true decoupling between the utility’s incentives and the allowed rate of return. We’re not there yet. I hope this sticky problem is dealt with when the PUC makes its final decision on the new PBR rules.

      Maybe the new PPAs will show dramatically lower prices. We’ll see. I think that will be a real test. But if rooftop solar can get streamlined approval for avoided cost pricing, I think it may both aid competition for utility scale PPAs, save land, and generally accelerate the renewable energy transition. It seems to the that this would serve the public interest on all fronts.

  2. James Roumasset

    Thanks for that detailed answer, Michael. Most states, including Hawaii have antitrust and consumer protection laws complementing Federal antitrust statutes and enforcement. My impression is that these
    enforcement agencies often have broad broad jurisdictions including public procurement and complementary utility regulation. I wonder if locating Hawaii’s Office of Consumer Protection in the Department of Commerce limits that agency’s independence. That was long a problem in Thailand, which has recently reorganized an independent competition authority. I believe California’s very active antitrust and consumer protection enforcement efforts are organized under the office of the attorney general. This might be an interesting governance issue for UHERO, the Public Policy Center, and the Dept of Public Admin in CSS to look into in conjunction w/ the UH Law School.

  3. There are several (I think quite critical) issues here that your analysis ignores.
    1) You have presented the cost of installation of solar on rooftops with (I assume) some minimal battery backup for overnight household supply. Longer term backup – say, for one week of no-solar – is, in this model, dependent on utility power generation, but would be substantially more expensive than the values cited if sufficient battery capacity alone was the source. This is particularly so if that rooftop supply is powering our major economic sector of tourism. (there is a further factor in this cost since the battery backup does not have a useful life anywhere near that of the pv generation system itself – more on that later – and therefore there will be significant periodic ongoing costs for that that are not included).
    2) Your model does not include the decline in efficiency of the pv system over that 25 years and, as some have learned to their great grief, less expensive pv cells have a much shorter useful life than 25 years. If, after 10 or 15 years, the efficiency of the pv cells is down to 70% of their “new” generation capacity, then value of the power generated should be appropriately adjusted.
    3) There is a significant risk that is routinely ignored as penetration of pv increases – both utility scale as well as rooftop pv systems are vulnerable to weather damage – clearly in Hawaii that damage is from winds. Having experienced Iwa and Iniki in Honolulu, it doesn’t take a great deal of imagination to be concerned about large scale losses of generation capacity. Whereas loss of transmission lines is devastating to operations of a modern society, the generation capacity was minimally impacted. Transmission capacity – sufficient for more or less physical recovery of the hurricane damages – was restored within a few weeks; how long would it take to replace all the damaged rooftop and grid scale generation that could be damaged by a severe storm?

    I find the narrative that solar will solve all our renewable supply problems to be very troubling – any analysis of the economics of solar needs to incorporate these issues in the analysis.

    1. Michael Roberts

      Hi Donald,

      Thank you for your comment! I will try to address your points in turn.

      1.) In this blog post I am trying to get at a fairly narrow issue within the broader context of the renewable energy transition—the role of rooftop solar versus grid-scale solar, and how we ought to compensate rooftop solar. I am not considering batteries in this analysis; we have a lot of grid-scale batteries coming online, and from a system-cost perspective, grid-scale batteries make a lot more sense today. Individual households and businesses are installing a lot of batteries too, for their own backup, and because current tariff structures and subsidies reward this generously even though they provide little value to the larger grid.

      Solar is, I believe, a very big part of the broader transition. I have provided links to analyses that do more comprehensive analysis of the broader transition and associated costs.

      Briefly: in the near term, we can backfeed a lot more solar from rooftops and be fine. Longer-term, as I allude in the end, the avoided cost from rooftop solar will change and will require a different compensation scheme.

      2) You’re correct about the decline in efficiency. But then I assume zero efficiency after 25 years, and an average of just 5 hours of sunlight a day. In West Oahu, you can average 7 hours according to data from NREL. In Monoa or Kanehoe, these numbers are far too generous, however. Mileage will vary. On my own home, the panels have averaged 22% of nameplate DC capacity during the 5th year of use, or 5.3 hours/day. Capacity declines less than 1% per year. I think 25 years is a typical assumption today. I also have a mortgage at 2.6%, which would more than compensate for an extra 1% loss. Anyhow, I’m just trying to keep things simple. I don’t think it’s too far from the mark for an unshaded rooftop, even with this and other omissions, like inverter losses.

      An old solar calculator at can help readers consider how much these factors may matter. Note, however, that net-metering tariffs are no longer available, so that calculator no longer applies to any homeowner’s calculation today.

      3) Yes, I think Hawaii faces a lot of risk from hurricanes and other kinds of natural disasters. It’s not clear to me whether solar is more or less risky than other forms of generation. Puerto Rico has learned a lot about this in recent years and they are working to develop a more robust system. My understanding is that they plan to make their next system more resilient, in part by using a lot more solar, not less, and use a system of interconnected microgrids. Solar is mounted flat on the ground instead of rooftops or in more efficient but more vulnerable tracking systems raised off of the ground. I agree that this is an interesting and important consideration, one that’s connected to the whole infrastructure of our islands and how hurricane risks may be changing.

      On your final point: A number of us are working hard on more comprehensive analyses. We can’t put all of that in a blog post. But I do think that solar is a big part of our future — it’s cheap, clean, and much easier to site than wind. Storage is getting cheaper very fast too. This is going to change the world. At this point, the larger challenges are more institutional than technical: utilities and utility regulation are poorly suited to distributed resources and the intermittency of renewables. We need to change the rules so that they better enable the needed technologies.

      Of course, new forms of geothermal, hydrogen and other forms of renewable resources may eventually show broader viability, but that’s a lot more speculative than solar is today.

      1. Hi Michael,
        Thanks for your reply. I’m happy to hear that you and UHERO are working on a more comprehensive analysis – I have been advocating with DBEDT for a while that a comprehensive economic/engineering analysis needs to be done in order to find the optimum path forward – but have encountered only resistance to that pursuit.

        I understand that the costs of both pv and battery storage are coming down, but, here again, a realistic analysis of how those costs can be expected to change over the next several decades needs to be done to make anything approaching a realistic assessment of the best path forward: at what point does scaling up demand for lithium exceed what the market can provide and result in a reversal.

        And please understand that my concern is not that pv is somehow bad – but I have seen other base load renewable alternatives come under attack (e.g. Hu Honua/biomass, and geothermal in general) by those who claim that all we need do is double down on solar while ignoring the actual costs and risks of complete reliance on pv.

        With respect to risks – I would disagree about the risks solar is subject to: I can’t recall any time that a hurricane (which I believe represents Hawaii’s highest annualized hazard loss over the long term) has done substantial damage to a centralized power station. The distribution grid is clearly at risk to hurricane hazards; to the same degree that other distributed infrastructure is. But my concern is that you could lose both the distribution grid and a significant fraction of the power generation capacity at the same time were we overly dependent on pv – recovery from that would be extraordinarily difficult. After Iniki, when Kauai was completely dark, they had some serious challenges restarting conventional power generation, even though their generation facilities were not substantially damaged.

  4. Michael Roberts


    What you are suggesting with regard to optimizing has been done. The real intellectual leader here is Matthias Fripp (UHERO and Electrical Engineering) — he has developed a state-of-the-art capacity expansion model that is designed for renewables, intermittency and storage. It is called “Switch 2.0,” and his modelling appears to be having a marked influence on policy in the state. A few years back his model helped to push HECO to revise their Power Supply and Improvement Plan. HECO’s plan is still flawed, in my view, but it moved a little in a positive direction. Elemental Exellerator later hired Rhodium Group to use his model to develop a statewide plan for all of the islands. I linked to that at the very beginning of the blog post. And in just the last few months, he has been working with Ulupono on the current PBR docket that will hopefully change HECO’s business model to encourage them to optimize going forward. The version of his model reported in that docket (look up 2018-0088 at This new version of the model has been updated with latest projections for costs going forward. The best part of all this effort is that the model is open source, which means anyone can scrutinize assumptions and build from it. See

    Matthias presented this model and how it is being used in state policy a couple of weeks ago in WEER. You can find the slides from that talk on the WEER webpage, here:

    I have been fortunate to be able to work with Matthias. With former graduate student, we developed the paper, also linked above, that folds a more sophisticated demand side to estimate how much this can help deal with intermittency. We are now we are building these models on much larger, national scales, and integrating natural gas networks, in an effort to build open-source decarbonization plans.

    I think the issue of resiliency in the face of potential disasters is important, and it ought to be addressed, but it is not a priority of ours right now. I do know others are working on this, especially focused on Puerto Rico. I don’t see strong prima-facie evidence that solar is worse than other alternatives, and I think there are good reasons to believe it would be part of a more resilient system—it is, at least so far, in Puerto Rico.

Leave a Comment

Your email address will not be published. Required fields are marked *

The University of Hawaii Economic Research Organization (UHERO) welcomes online comments to stories that are posted on our website or social media pages. Comments are intended to be a forum for open, respectful, and family-friendly discussion. UHERO reserves the right to remove anything posted on our website or social media pages that is deemed inappropriate. All comments are moderated and will therefore have a delayed post time.
Some guidelines (not an exhaustive list) we use when moderating/approving comments include:

  • Do not bully, intimidate, or harass any user.
  • Do not post content that is hateful, threatening or wildly off-topic; or do anything unlawful, malicious, discriminatory or defamatory.
  • Observe confidentiality laws at all times.
  • Do not post spam or advertisements.
  • Observe fair use, copyright and disclosure laws.
  • Do not use vulgar language or profanity.

UHERO may amend this policy from time to time.