The Lawrence Berkley National Laboratory released several studies this summer that explore what comes next in the evolving power market landscape of the United States. The application of academic literature can often be limited in the world of investing; however, we find Berkley's analytic rigor combined with thoughtful questions useful for contextualizing some of the second and third-order effects of the energy transition.
Two papers stood out to us. The first examines the benefits and costs associated with a utility ownership model for residential rooftop solar. The second study examines the impact of high renewable energy penetration on demand-side decision making, with a particular focus on theoretical hydrogen producers operating industrial-scale electrolysis assets.
It would be accretive to utility shareholders, but effects on non-solar customers are not clear.
This is an interesting question, particularly considering the operational challenges recently faced by California's grid operator CAISO. Rapid growth in rooftop solar1 poses fundamental challenges to the financial stability of regulated utilities, displacing traditional utility earnings opportunities and/or creating unsustainable cost-shifting among different groups of utility ratepayers.2 Utilities have responded in different ways, with some looking at taking direct ownership of rooftop solar.
Leaving aside the legal implications (of which there are many), this is not a theoretical question anymore. Arizona Public Service, Dominion Energy, and Duke Energy have all initiated pilot programs to this effect in the last two years.
Berkley Lab finds that utility-owned rooftop solar is accretive to utility shareholders, boosting earnings by 3.4% on a present value basis over 20 years. This is in contrast to a 1.7% reduction in shareholder earnings should the same level of solar be built out but owned by the homeowner (or third party). Under both structures, additional rooftop solar defers capital expenditures for gas combined-cycle plants, a net loss for shareholder earnings. Should the utilities begin to own and operate rooftop solar, the earnings erosion from capital deferral is more than offset by the earnings gain on the capital investments in rooftop solar.3 Under the homeownership structure, utility shareholder earnings erode even further as a result of reduced retail sales and capital deferment.
So far, the premise is straightforward: when distributed energy resources increase, regulated utilities lose earnings power. Any plan to own more of it is likely accretive to future earnings and shareholder value. Yet who bears the cost?
The impact on retail customers is mixed, particularly for those that choose not to install solar. As rooftop solar increases, non-solar customers are likely to see electricity bills increase by ~2%, regardless of whether solar assets are owned by the utility or homeowners.4
An important outstanding question (which leads to the mixed result conclusion) is the monthly payment provided to a participating homeowner as compensation for the use of their rooftop. How much would your utility need to pay you to install, own and operate equipment on your roof?
Utility pilot programs have thus far offered a $30 monthly payment to participating customers (APS 2019). Larger amounts may be required to motivate high levels of participation. If true, non-solar customers will experience a disproportionate increase in electricity rates to fund this scheme. Shareholder earnings are not affected as the program would be an operating cost recovered by a direct pass-through to retail rates.
It's not that simple.
The statement above is likely what you walk away with if you listen to quarterly earnings calls from PLUG, NFE, BE (pick your hydrogen company of choice). Sell-side research largely echoes this claim. In theory, if the levelized cost of hydrogen can drop below $1/kg it becomes an economically viable alternative to other sources of power.
The energy costs to split water into hydrogen and oxygen (electrolysis) are a significant component of total overall costs. The variable cost of renewable energy is often $0, so it is convenient to draw the parallel between large scale renewable buildout with low hydrogen costs. It turns out even simplified modeling5 shows this is not necessarily the case. Berkley Labs set out to study whether private, and public sector decisions that are made based on assumptions of a low level of renewable energy penetration still achieve their intended objectives in a high renewable energy penetration scenario, the grid being 40-50% wind and solar.
For large energy consumers, which firms producing hydrogen via electrolysis would be, the availability and price of energy is a significant factor in decisions on capital investments, operations, and siting of assets. Increasing renewables on the grid should lower wholesale market prices. For example, Berkley Labs modeling shows that in a high solar penetration scenario in Texas, wholesale prices may drop below $5/MWh for ~20% of the year. (Prices averaged $38/MWh in 2019).
The increasing availability of cheap electricity should thus motivate industrial energy users to invest in equipment that allows energy use flexibility: switching from direct fossil fuel consumption to electricity low price periods or modifying production processes to use electricity to make products that compete with those derived from fossil fuels.
Berkley Labs finds that in a high renewable energy penetration scenario (~40% of installed capacity), the levelized cost of hydrogen production decreases between 13% and 40%. In several optimal run-time scenarios, Berkley finds hydrogen production costs drop to $1.5/kg, but fails to reach the desired $1/kg level.
It's important to note how wide the spread in cost reductions are (a 27-percentage point difference). High levels of renewable penetration can have dramatically different impacts on the level and variability in wholesale prices in different parts of the country. This also means cost reductions associated with identical increases in renewable energy are disproportionate (again, primarily based on location).
The failure to reach $1/kg (and the uneven distribution in cost reduction) is not the primary issue though. The optimal run-time associated with the low-cost hydrogen is. The base case scenario used by Berkley Labs dispatches the electroylzer such that it minimizes the levelized cost of hydrogen produced. For states like California, which have steep price curves, minimizing cost means running the equipment less.6 Most industrial uses of hydrogen, however, are subject to minimum run-time constraints.
Green steel, a theoretical use case for green hydrogen, is an excellent example of the problem addressed by the subsequent scenarios in the Berkley study. Steel mills, regardless of the energy source, are not intermittent operations. Steel mills are operations for which a continuous flow of hydrogen is often necessary (start-up and shut down costs can outweigh any cost-benefit resulting from a volatile production profile). Under the constraint that hydrogen needs to flow 24 hours a day, Berkley finds that the levelized cost of hydrogen would likely increase by 15% in some states. If we assume the 15% increase occurs on the average optimal outcome in their base case, we are looking at the cost of hydrogen rising to about $1.75/kg. At that hydrogen price, the Columbia Center on Global Energy Policy estimates that the cost per ton of steel would increase by anywhere from 10% to 30%.7
Low-cost energy is an important step to enable economically viable hydrogen production. The variance in price distribution from an increase in renewable energy is uneven across states, suggesting that theoretical cost reductions are also unevenly distributed. Utilization constraints based on end-use applications are also not uniform. The idea that cheap and plentiful renewables = competitive hydrogen economics is not entirely accurate.
We would encourage investors that are interested in the energy and industrial businesses navigating the energy transition to take advantage of the growing body of research (immediately applicable or otherwise) in this space. We find the level of discourse in the financial community to be shallow in an otherwise rich area of innovation and changing markets.
The two papers discussed above present an excellent example of what is available. The first paper discussed some of the potential weaknesses of the existing utility business model as distributed energy expands across the grid. But it also highlights some of the ways that utility management teams can shift from ceding market share in a distributed energy environment to growing earnings, an obvious benefit for any investor. As investors, we endeavor to see around corners. Doing so in the current energy environment requires a nuanced perspective of how business strategies can and will change, even for business as staid as utilities. In-depth operational research can help us accomplish that goal, especially for companies with as much operating leverage as energy and industrial firms.
When you read about a utility management team proposing a strategy shift that includes ownership of rooftop solar, remember, there are cases in which it might produce a meaningful boost to earnings. But ask yourself, how do the management team's proposals compare to the case studies presented in the paper discussed above? How might the differences impact the bottom line or make success more or less likely?
The second paper paints a picture of the operating context for the nascent but currently hot hydrogen producer market; several of the hydrogen firms we follow are up more than 100% this year despite all having never made any money. The management teams at several of these companies suggest the time for hydrogen is now a great story for sure, but one that appears to be on less stable footing then many investors are being led to believe. The simplicity of the stories belies the complication of execution.
As investors in real things, we are once again confronted with the reality that there are constraints on what can be accomplished as a result of time, resources, and science. Hydrogen and its potential, as described by management teams, stirs the imagination of investors, but matching the story with the complex reality of building a company that is not only environmentally sustainable but also economically sustainable is another matter altogether. Disruption produces winners and losers; the ability to bridge the gap between narrative and reality will largely determine who survives and prospers and who does not. We cannot forget no matter how great the story, businesses going forward must be both environmentally sustainable and economically sustainable. Hydrogen producers appear likely to fail that test for now.
For those interested, we would recommend paying a visit to the visualizations and tools put forth by Berkley Labs. We are of course more than happy to strike up a conversation to talk more in depth about the technical matters in this energy transition.
1And distributed energy resources more broadly.
2Impact of Residential PV Adoption on Retail Electricity Rates. Energy Policy, Cai et al, 2013. The economic effect of electricity net-metering with solar PV: Consequences for network cost recovery, cross subsidies and policy objectives. Energy Policy, Eid et al, 2014.
3This is largely because rooftop solar is more capital intensive and has a lower capacity factor than the generation it is deferring.
4The homeowner compensation for selling excess rooftop solar back to the utility is a big question moving forward. The Berkley Lab counterfactual assumes full retail rate net metering, “an historically accurate by relatively generous compensation scheme that leads to high impacts on both shareholder earnings and non-participant bills”. The relative value of utility ownership may diminish should utilities move away from this type of rate design.
5While the modeling Berkley does is mathematically sophisticated, there are several theoretical simplifications. Berkley assume industrial end use customers pay only energy and capacity costs. The model includes price effects of an expanded transmission system to integrate new resources the model chooses to value externalized environmental costs, demand change induced price effects, reserve requirements, and fuel price risks.
6This also means that capital expenses constitute a larger share of total levelized costs compared to a lower renewable build out scenario. The implication here is that in certain parts of this country, the decision to produce hydrogen via electrolysis may hinge less on the potential renewable energy pathways and more on the total reduction in capital expenditures.
7Low Carbon Heat Solutions for Heavy Industry. S. Julio Friedmann, October 2019.