In the USA, publicly-owned nuclear is actually increasing…
As a follow-up to my recent post tracking a century of public ownership of generating capacity in Canada, in this blog I turn my attention to our southern neighbour, the USA, and do a deep dive on the ownership question in the nuclear generating sector.
US Nuclear Overview
As the graph from the Energy Information Administration (EIA), shows, nuclear generating capacity took off in the early 1970’s and plateaued at around 100 GW starting in 1990. The graph also shows that generation has increased more than capacity because average nuclear capacity factors increased over the last 50 years; starting at an average of 43% in 1970, it increased about 1.2% points/year to hit 67% in 1990 and then increased by 0.9% points/year to 2021, when it reached an average of 93%.
Generating capacity peaked at 101.9 GW in 2012, after which it has declined to 94.1 GW in late-May, 2022, after the closure of Palisades NPP in Michigan.
Public Ownership of Nuclear in the USA
My research question is whether ownership “mattered” in the evolution of nuclear capacity over the last 10-15 years.
I look at two years, 2008 and 2022, to see how public vs. private ownership has changed.
For 2022 I take the semi-compiled data provided by the EIA to September, 2021 that includes capacity and ownership shares data for 94 reactors, including Indian Point 3 and Palisades. I update the dataset to late May 2022 by deleting those two no-longer operational reactors, for a revised total of 92 reactors. I identify each of the named ownership entities and classify them as either “private” or “public” (see discussion below). The EIA also provides information on reactor operators, but that is a different matter - my interest is on ownership.
Why 2008? Because this is the last year for which the EIA provided ownership data in a semi-compiled basis. Prior to that would require much more additional detailed compilation based on EIA-860 Survey Datasets (see below). For 2008 I take the similarly-compiled data from the EIA, this time from the Wayback Machine because the EIA does not seem to maintain an archive of older data. That dataset includes capacity and ownership share data for 104 operational reactors. I do not revise this dataset.
These are the results of initial research. If there was interest, it may be worth compiliing more data to have observations from other years, including before 2008, including all the way back to 1990 or before - that would require a deep dive into EIA-860 Survey Datasets/Reports.
Two important aspects with respect to terminology.
There appears to be different uses/definitions of the term “public” power in the USA. For example, the American Public Power Association (APPA) uses a relatively narrow definition of “public” restricting it to local, county or state-owned entities. It does not include co-ops or federal agencies under “public power.” That is not the approach I take. I use the approach I took for Canada in the previous blog, which is that under “public” I include any entity owned by any level of government as well as all co-ops. In this context, in the USA, “Private” includes the private utilities – the investor-owned utilities (IOU) – and all non-utility generators (NUGs) – that is, independent power producers (IPPs) or merchant generators.
The other matter is that the USA has three types of reactors when it comes to ownership structure. Some reactors are 100% privately-owned, either by IOUs or NUGs. A second type of reactor are those that are 100% publicly-owned, including those owned by the Tennessee Valley Authority (TVA) and other public entities. A third type is the mixed ownership reactor, where some percentage is owned by both private and public interests.
The table above shows that there was a total of 104 reactors in 2008 with capacity of 100.8 GW, compared to 92 reactors with 94.7 GW as of late May, 2022 (after the closure of Palisades NPP). That is a decline of 12 reactors (-11.5%) and 6.1 GW (-6.0%), respectively, over 14 years.
The difference in percentage reductions between reactors and capacity is because the 13 reactors that closed in the 14 years were smaller than the average, and the one reactor that started operation during the period was larger than the average. Further, for the 91 reactors that were present in 2008 and 2022, capacity increased on average by 3.6% due to uprates.
Public versus private Nuclear in the USA over the last 14 years
There were marked differences in which reactors closed and which remained open in the USA.
Of the net closure of 12 reactors, fully 10 were 100% privately-owned. One reactor that closed was of mixed ownership. And a twelfth was 100% publicly-owned.
The capacity numbers tell a starker story. Of the net loss of -6.1 GW over the last 14 years, fully -6.5 GW was due to 100% privately-owned reactors shutting down. 100% publicly-owend reactors actually increased capacity slightly, while mixed-ownership reactors increased 0.4 GW.
Further, within the mixed-ownership reactors, public ownership increased by 1.0 GW, while private ownership decreased by -0.6 GW.
Taking all of this together, as set out in the table above, privately-owned nuclear capacity decreased by 7.1 GW from 2008 to 2022, while publicly-owned nuclear capacity increased by 1.0 GW.
In percentage terms this means that public ownership of the nuclear sector increased from 17.4% in 2008 to 19.6% in 2022.
By way of context, the table below shows my aggregation of APPA-compiled EIA data for public/private ownership of all generation technologies in the USA (for 2019). Overall, publicly-owned capacity accounted for 21.8% of all capacity. Note that public ownership is close to that average for some technologies (gas, coal, nuclear and oil). But publicly-owned generation is much higher than that average for some technologies (hydro) or much lower for others (wind and solar).
Discussion
These findings are stark. I am unaware of any other researcher parsing out the evolution of nuclear reactors in the USA over the last decade and identifying such a potentially important explanatory variable - ownership.
There could be other drivers for why some reactors closed and others stayed open, including the State-level regulatory environment under which they operate (traditional vs. restructured vs. Federal carve-outs), whether the 100% private or mixed-ownership reactors were owned by IOUs or NUGs, or whether the type of public ownership matters (Federal (TVA) versus other public entities). Further statistical inference testing may be required to be able to separate these and other effects and to assess whether a binary public/private ownership categorization matters as much as this initial research suggests.
In the meantime, this is a promising hypotheis, based on which we can make some observations looking backwards and forward.
Looking back, the counterfactual would be whether fewer reactors would have closed had they been 100% or partially publicly-owned. This is not a moot point in the USA: Indian Point 3 was publicly-owned (by the NYPA) until it was privatized in 2000. It was closed in 2021 as a 100% privately-owned reactor, by a NUG. This initial research suggests that Indian Point 3 was less likely to have been closed had it remained publicly-owned.
Looking forward, this initial research suggests a preventative approach to reduce the risk of future closure.
For example, the average public ownersip of the 24 mixed-ownership reactors in 2008 was 28.2%. The public ownership share of the 20 reactors that continued operation into 2022 was 31.0% in 2008, while those of the 4 reactors that closed was 12.1% in 2008. That suggests that the higher the public ownership share, the lower the risk of subsequent closure. In this manner, it could be the case that a partial purchase in a 100% privately-owned reactor, or an increase in the ownership share of a mixed-ownership reactor, by a public entity, could reduce the chances of that reactor closing in the future.
Does that hypothesis extend all the way to 100% publicly-owned? Maybe. But such a transaction did not occur during the 14 year study period - one would have to go back to the 1970s to capture such an occurrence. On the other hand, changes in partial ownership shares did occur in the study period, so they are incorporated into the data.
If this hypotheis is validated, the policy prescription is clear if the objective is to reduce the risk of closures of existing nuclear reactors: 1) advocate for some partial public ownership in 100% privately-owned reactors and 2) advocate for an increased share of public ownership in the mixed-ownership reactors.
That is preventiative and would constitute an incrementalist approach to what appears to be occurring organically. The objective therefore would be to drive this forward as an explicit policy objective. This appears to be feasible within the current political economy of the USA.
To a non-American what does not feel politically possible right now, however, is the option of public ownership to reverse an imminent or announced closuse. That is a different matter than the preventative approach discussed above. As far as I am aware, this did not occur during the 14-year study period, including most recently in the case of Indian Point 3 and Palisades: neither the Federal government nor the respective New York or Michigan governments or any other public entity considered taking an ownership stake in these reactors with the objective of keeping them open. My US friends and colleagues are better placed to answer whether in the future that political calculus could change due to an impending or actual electricity crisis, as the grid becomes less reliable because more clean, baseload nuclear is shuttered by its private owners.