For most of my career, electricity prices have been dull. For more than a decade, US inflation was very low, and so too was electricity price inflation, regardless of the use class.  Homeowners, small-and-medium businesses, and even major heavy industrial consumers could expect price increases in a tight band centered on zero. While the market was not immune from price action, it was not the kind of trend that required customers to take their own action in response.  

And then inflation happened. At its peak in 2022, US inflation was more than nine percent.  Electricity price inflation, however, peaked higher than 10%, and in 2022, industrial power price increases spiked to more than 20% year over year. And while electricity rates (in particular, industrial rates) fell again in tandem with declining inflation, they are starting to creep up again.

What does the future hold for US electricity rates? I love a good market narrative, and I do not mind a theory of change either (on occasion). We can look at forward curves for commodities and get one theory of how prices will move; we can add in predictions or expectations of demand changes, and expand that theory further. This is what traders do in creating their understanding of the market.  

But utilities and regulators also practice this theory of change — though I doubt that they would describe it that way! Utilities say that they expect prices for inputs, commodities, labor, and so on to move in one direction; they add in their expectations for the cost of capital to build new things, and then include their expectations for how much demand will grow. The combination creates a number: the amount by which a utility needs to increase its rates in order to meet its expected return on equity while investing for the future.  

But it’s more than just a number; it’s also an argument in a way, because regulators are (and should be) in the business of making their own arguments about the same variables. Settling that argument gives both parties their parameters (regulators: what they are willing to accept; utilities: what they are able to get) and ideally, unlocks capital to invest while doing so at the least cost. And in a period of rising inflation expectations, with high costs for key inputs like transformers and booming demand from hungry customers like data centers and factories, you would be right to think that the general utility argument is “rates need to increase.” 

Fair enough — but the magnitude of those increases is striking, and it exists all over the US. I mentioned two in my most recent look across the market. One is in Idaho, where Avista is proposing a 14.7% increase in residential rates. In addition to every other macro factor, Avista also must account for the uncertainty in the cost of its fuel imported from Canada, a stochastic process that does not play well with deterministic needs. 

Another is in Wyoming, where Montana-Dakota Utilities is requesting an 11.7% rate increase — but a much higher (more than 15%) increase for its residential customers. And when I look across the US, I see similar utility requests everywhere, and not just for electricity, but for water and natural gas utilities too. Some small utilities (such as a water utility in Ohio) are requesting increases of more than 30%. This is napkin math territory: a 30% increase is about 10 times greater than current inflation, and for ratepayers with lower or fixed income, it could be hard to manage. 

Managing begins with measuring. For stakeholders active in their own market, that might be relatively easy to do; for those looking across many markets, it becomes increasingly difficult to track data points, and also to track arguments as well. Difficult, but something which large language models and data science are increasingly able to address — not just the numbers, but the reasons behind them.  

Earlier this month, my colleague Alex Klaessig described some of the work that we’ve completed via Halcyon Helpdesk (HHD), our managed service offering that helps energy professionals supercharge their research projects. A lot of the research projects HHD has helped complete have been centered around understanding rate case changes within and across different utilities and different regions. Rate cases, in particular, are challenging for a few reasons: they are largely table-centric and contain various forms of structured data, which can be challenging for OCR (optical character recognition) technology. Understanding rate cases also requires cross-referencing other documents and testimony to understand what was proposed, what was accepted, and why — defining this universe is tough for an LLM, even if the subsequent analysis is more straightforward. 

These challenges are why HHD employs a human-in-the-loop approach. If LLMs are the supplements that help supercharge information analysis, then data scientists are the nutritionists that prescribe how they should be used, and when. Over time, we will encode their knowledge into our systems (which is a key benefit of verticalized AI relative to more general foundation models), but for now, as Alex says, having subject-matter experts steer the wheel “is a feature, not a bug.” 

If you’d like to get a better handle on rate cases, or have questions about how uncertainty in US economic policy is impacting energy markets, HHD can help answer complex questions faster and more affordably than you’d expect. Give us a shout today: helpdesk@halcyon.io

Comments or questions? We’d love to hear from you - sayhi@halcyon.io, or find us on LinkedIn and Twitter