- Despite widespread use, the levelized costs of energy metric has several key limitations.
- By simplifying costs into one number, it blurs the distinction between capital and operating costs, obscures the importance of interest rates, and implies only one optimal economic outcome.
- Other operating, economic, and policy assumptions that are critical in forming the metric can significantly impact its outcome.
- LCOE should not be the only metric used to compare the economic attractiveness of different energy sources – more holistic approaches are needed.
Many professionals across the energy policy space uncritically use the levelized costs of energy (LCOE) metric to judge the economic competitiveness of different technologies.
LCOE is attractive because it appears simple – by combining all present and future costs you have one number that represent the economics of energy supply. Accordingly, people continue to use LCOE to compare the attractiveness of different energy supplies, in energy models, and in other applications.
However, any user of the metric needs to recognize its very real drawbacks. While not exhaustive, you should consider the following 9 limitations the next time you read about or use LCOE.
A quick note: this post is focused on LCOE as an energy metric, not a project finance calculation, where levelized energy costs are use with other metrics in developing a new project.
- LCOE blurs distinctions between marginal, fixed, and capital costs
By design, LCOE is a simplification – worryingly, that simplification obscures the critical differences in cost types and cost structures of different energy sources. Most important are upfront (capital) versus operating (primarily fuel) costs.
Nuclear, hydro, solar, and wind have very high capital costs with low operating costs. Coal has moderate capital and operating costs. Natural gas has very little capital costs but very high fuel costs.
EIA’s Levelized Cost of Electricity for Fuel Sources in 2020
These distinctions are critical – even if a high capital cost technology has the lowest LCOE of any fuel source, the upfront costs could be too much to overcome.
Although rarely addressed, this is one of the larger challenges facing the developing world. In many countries coal is the primary centralized electricity source and diesel is the primary decentralized electricity source because electricity purchasers cannot afford the upfront costs of higher capital technologies. On a LCOE basis, diesel in particular is very costly but it is selected as the fuel choice because there it has low initial capital costs.
Even if it is cheaper in the long term to choose a capital-intensive technology, the lack of access to financing prevents widescale adoption in many countries. It is telling that one of the largest growing sources of renewable energy in the developing world is hydroelectricity financed by outside countries or institutions.
Meanwhile, in developed countries, calculating LCOE based off capital costs is misleading because….
- Interest rates are made up
Okay, not quite made up, but also over-simplified.
Although the level of capital costs vary between technologies, all sources of electricity generation have capital costs that need to be financed. To finance construction of a new project, project developers need to either internally finance the project, with a corresponding return on investment, or they need to secure outside financing, which usually has interest costs.
Most calculations of LCOE assume a set interest rate, regardless of the viability of the project. Lazard’s LCOE, one of the most rigorous and respected in the field, assumes that almost all projects have 60% debt at 8% interest and 40% equity at a cost of 12%. Meanwhile, the U.S. EIA uses a flat 6.1% cost of capital.
In reality, interest rates vary considerably based on literally hundreds of factors. Overall macroeconomic conditions are critical. For example, the low interest rates of the last several years have been a critical and underappreciated factor in financing both the clean energy and shale booms.
Other major factors include the type of technology, the creditworthiness of the power project developer, the interest needs of the lender, and the location of the project.
Changing specific assumptions for interest rates can dramatically change the estimated LCOE of an energy source. Choose a high enough interest rate and a capital-intensive project like hydro could look worse than a low-capital cost natural gas plant. Choose a low enough interest rate and solar and wind could blow away the competition.
Both the graph below and this link provide good examples of how different interest rates affect LCOE calculations.
Lazard’s LCOE – Sensitivity to Cost of Capital
At their core, interest rates reflect the financing risks of different projects, and so vary between technologies and projects (even if LCOE does not account for them). Speaking of risks…
- LCOE ignores project risks
Although not often described this way, decisions about electricity supply planning are primarily risk-based decision making. Every major choice that we face when planning and operating an energy system has certain elements of risk. Examples:
- Construction issues
- Power price issues
- Fuel price issues
- Generator availability
- Radiological/other safety risks
- Regulatory risks
Many of these factors are project, company, technology, and region specific. And they are utterly ignored by LCOE calculations.
Sure, building a nuclear power plant might make sense on a LCOE basis, but it carries a near guarantee of severe cost and time overruns. This is why all new nuclear power plants in the U.S. are being built in rate-regulated markets – utilities can rely on ratepayers to cover (most) costs when the project goes over budget.
Building a coal plant today makes sense on a LCOE basis, but its horrible environmental profile and market factors have virtually killed new coal builds in the U.S.
Risk factors are also important in light of capital and marginal cost differences. Wind and natural gas may look competitive on a LCOE basis, but natural gas potentially faces higher risks. Build a capital-intensive wind project now and you lock in most of your costs. Data from Lawrence Berkeley National Laboratory indicates that the average solar power purchase agreement actually has declining prices over time.
Generation-Weighted Average PV PPA Prices Over Time by Contract Vintage
Conversely, build a natural gas plant now and most of your costs are still uncertain in the future – they are dependent on prevailing natural gas prices, which are notoriously volatile. LCOE rarely, if ever, accounts for volatility in natural gas or other fuel prices.
These risk factors are not reflected in LCOE calculations, but can be as important to decision making in electricity. Speaking of which….
- LCOE assumes one decisionmaker
If a casual observer were to listen to the rhetoric around energy choices, they would assume everyone is trying to convince the one ‘chief energy decisionmaker’. While this is a bit of a hyperbole, the way we analyze and discuss energy usually obscures the many different actors in the energy space.
LCOE is a great example. The entire point of calculating one levelized cost of electricity is to financially compare the costs of different energy sources so that ‘someone’ (a utility, the market, the god of electricity) can figure out what is the cheapest way to get electricity.
In reality, the way that we now choose energy supplies is closer to a chaotic brawl loosely controlled by inconsistent market and regulatory guidelines, influenced by thousands of market players with different costs and perspectives of the future, and continuously disrupted by rapidly changing technologies.
Thirty years ago, things may have been different – vertically integrated utilities (overseen by state commissions) dominated electricity supply choice.
Today, the US has seven different regional wholesale electricity markets with varying levels of competition, a rapidly increasingly number of market participants, and (critically) an increasing number of potential electricity suppliers. In practice, this means that thousands of decisions are being made either explicitly or implicitly by actors with a varying level of sophistication who favor different cost structures and project risks.
Map of North American RTOs/ISOs (Competitive Wholesale Electricity Markets)
Source: Sustainable FERC
Electricity consumers can choose renewable energy to guarantee fixed electricity costs and mitigate price volatility risk. Or a utility could make a simple calculation that assumes a low and flat natural gas price through 2030, leading to them constructing a natural gas plant. Or they could take a massive risk by building the nation’s first commercial carbon capture facility and make it more difficult by trying to also make it one of the world’s first IGCC coal plants.
LCOE seems to imply that we have one decision maker that sees the world in exactly the same way in all parts of the country and acts accordingly. Reality is a lot messier when it comes to financial guesses about the future by thousands of entities.
Oh, and those decision makers are not just making financial determinations…
- LCOE ignores non-financial factors influencing electricity production
The purpose of comparative LCOE is to calculate which energy sources are the most attractive economically. However, the prevailing method of calculating LCOE ignores many types of externalities.
Most notably, the environmental costs of energy production are very large constraints that are increasingly influencing the power sector.
Coal is the worst, with horrible air emissions, major land use impacts from mining, and high rates of water withdrawals as well as water pollution issues. Natural gas is better but nonetheless emits significant amounts of CO2 and has major open questions regarding the environmental impacts of fracking. Even nuclear (very high water usage and uranium mining) and renewables (land use impacts) have environmental costs which affect fuel choice decisions.
Although the risks of a meltdown are very small, nuclear plants also face serious safety challenges. The current generation of nuclear reactors also produce a substantial amount of nuclear waste that still does not have a permanent repository.
All of these issues influence power supply planning but are not factored into LCOE numbers.
Similarly, LCOE generally ignores energy trade questions. Importing a fuel could be an economically attractive proposition on a LCOE basis – however, countries generally want to minimize their dependence on energy imports, so might choose a more expensive energy source.
- Operating and Technical Assumptions Critical for Calculating LCOE
To calculate the LCOE for comparative analysis, one needs to determine a specific lifetime and capacity factor for each type of power plant. Objectively, the best way to do this is (probably) to use a range of lifetimes and potential capacity factor.
Most LCOE calculations generally assume a specific lifetime for a plant – while these assumption generally correlate with common financing terms (i.e. 20-year lifetime for a 20-year loan), power plants generally operate beyond assumed lifetimes.
More worryingly, capacity factor assumptions are not always defensible. Several of Lazard’s LCOE assumptions are certainly questionable – the LCOE for an IGCC coal facility is exactly 75% while a new supercritical coal plant is 93%. EIA’s LCOE estimates make similar questionable assumptions, including:
- Coal plants: 85% (too high)
- Natural gas combined cycle: 87% (too high)
- Natural gas combustion turbine: 30% (too high)
- Onshore wind: 36% (both too high and too low)
In general, the higher the assumed capacity factor, the lower the LCOE. If you ever want to challenge a LCOE calculation, capacity factor assumptions are probably the quickest way to find a problematic assumption. Or you could point out that…
- LCOE is calculated for the generic MWh
In comparing cost effective choices for electricity, LCOE compares on the basis of a unit of energy delivered, commonly cost per megawatt-hour or cost per kilowatt-hour. However, one megawatt-hour does not always equal one megawatt-hour.
Electricity is not like other commodities – because you cannot currently store it, a megawatt-hour has a time characteristic that dominates its financial profile. The cost of supplying one megawatt-hour at night in the spring is very different than the cost of supplying one megawatt-hour during a very hot afternoon at the height of summer.
Average Day-Ahead Electricity Price in PJM, January through September 2014 versus 2015
Source: PJM Market Monitor
In practical terms, this means LCOE is misleading as it obscures how competition between energy sources actually works.
Baseload power plants, like nuclear, coal, and natural gas combined cycles, have very different economic imperatives than peakers like combustion turbines or (increasingly) solar. Wind, which most generates at night, is most appropriately compared to the most efficient baseload units. Solar, on the other hand, is most appropriately compared to the least efficient peaking units.
The location of electricity generation also significantly affects its price. Transmission constraints, as well as significant variations in regional fuel availability, influence energy supply choices in all sorts of locations and lead to very different per MWh costs. Using one set of LCOE numbers, even if they use a range, obscures the geographic dimensions of choosing the best electricity source. Or put another way…
- LCOE is not a measure of system costs
Many critics of renewable energy decry the integration costs of renewable energy. They argue that the need to integrate renewable intermittency and build transmission lines greatly increase the system costs of using renewables beyond what project developers actually pay. Conversely, renewable energy proponents point out that renewable energy can also reduce system costs, whether through the merit-order effect, deferred transmission investment, or other means.
While this debate is most intense with renewable energy, it hints at a deeper truth about energy supply choices: the all-in cost of supplying electricity is more important the generation costs for one unit of electricity. Any type of electricity generation can either decrease or increase system costs.
New nuclear energy might be the cheapest power source on a LCOE basis – indeed, existing plants are key in keeping overall system costs down. But it would not make sense to build a new nuclear plant to meet peak demand requirements. Rather, one would build a natural gas peaker even if it has a higher LCOE.
Conversely, building a natural gas plant in Pennsylvania might make perfect sense because infrastructure is readily available – building that same plant in New England could impose major system costs by worsening that region’s already growing dependence on natural gas generation.
In choosing LCOE as the metric to judge the competitiveness of energy supplies, one is narrowly framing the decision to exclude these system costs. In fact, this hints at the cross-cutting problem in relying upon LCOE at all…
- LCOE is a result of qualified assumptions, not objective truth
Underlying all other problems, LCOE is a troubling metric because it is like all other energy models and metrics: it often reflect the values and judgments of the creator, not true objective analysis. For whatever reason, having a number seems more objective or truthful than making a qualitative assessment. Energy models and metrics are prized because they supposedly tell us objective truths.
The reality is that energy models and metrics are qualitative judgments expressed numerically. To make neat and tidy numbers from a messy and complex world, we need to make many simplifications and assumptions. Sometimes these numbers are the best we have – but to use them properly we need to recognize their inherent limitations.
The best way to use LCOE is as one of many factors in an analysis. When comparing the cost effectiveness of different technologies, use LCOE as a starting point to explore more nuanced issues related to cost structures, future risks, and system impacts. Finally, if you are calculating LCOE or using LCOE in an energy model, make sure that you are as transparent as possible about assumptions.