In May, one of us (Mr. Loewen) published an opinion piece in Utility Dive suggesting that there could be an issue with the widely used levelized cost of energy (LCOE) metric — specifically that the metric does not appropriately capture the time value of money.

Given the widespread use of the LCOE metric, and intrigued by the notion that there could be something incorrect about it, the other two authors of this letter (Pieter Gagnon and Trieu Mai, energy economics analysts at NREL) reached out to Mr. Loewen to explore the issues more fully.

After working through the problem together, we were able to reach a conclusion: there is not an issue with the formulation of LCOE. It properly captures the time-value of money, allowing for projects of different lifetimes and cash flow patterns to be fairly compared. Therefore, LCOE can be used for rank-ordering electricity projects by their ability to deliver energy at least cost.

Working to complete the public record, Mr. Loewen has detailed the technical aspects of our discussion in The Electricity Journal. Independently, Richard McCann of the M.Cubed consultancy also published a response in the same journal defending the formulation of LCOE.

However, there is a caveat: what we just said applies to LCOE if calculated in real terms (i.e., adjusted for inflation). What had originally caught Mr. Loewen's attention — and what none of us fully recognized until working through the problem together — is that, if you calculate LCOE in nominal terms, and then compare projects with unequal lifetimes, the values produced by LCOE can incorrectly rank-order technologies.

Nominal LCOE can, in effect, "penalize" longer-lived projects, making them seem more expensive relative to short-lived projects. The problem is particularly acute if you are comparing a technology against the current cost of grid electricity: the use of nominal LCOE will make the project seem relatively more expensive than it actually is.

Fortunately, calculating real LCOE is no more difficult than calculating nominal LCOE; real LCOE simply utilizes a different discount rate to generalize the effects of inflation, making the resulting value comparable across projects. NREL products, such as the Annual Technology Baseline and Standard Scenarios, generally report LCOE values in real terms. Nominal LCOE can be useful in some specific circumstances, but real LCOE is less susceptible to issues with unequal lifetimes.

Alternatively, if you wish to compare projects with unequal lifetimes in nominal terms, Richard McCann points out that you can use "replacement chains" to make the projects have equal lives (for example, two sequential 30-year projects can be compared against three sequential 20-year projects).

Although the LCOE metric — when calculated in real terms — can appropriately rank-order the costs of energy supplied from different technologies, we want to make it clear that the metric also has its limitations. Most notably, it only summarizes the costs of projects, and does not capture their benefits. Given that the value of electricity can vary hour-by-hour or across even shorter intervals, this flaw is particularly significant when comparing technologies that have different dispatch profiles, e.g., fossil fuel-fired generators versus renewable energy technologies "fueled" by variable wind and solar.

Final project decisions would generally want to consider both the costs and benefits of a project, ideally through well-understood metrics such as net-present value or benefit-cost ratio. But in the same way that price tags will continue to be useful — despite also only containing information about costs — we believe that LCOE will continue to be a useful metric despite these well-recognized limitations.

We will conclude by emphasizing our main point: LCOE, when calculated in real terms, is an appropriate metric for rank-ordering projects by their costs. When calculated in nominal terms, LCOE can give misleading results when comparing alternatives with unequal lifetimes.