I’m two weeks late on this one, but the EIA’s Uranium Marketing Annual Report, released on 26 May 2020, has a lot of great tidbits in it. It’s an absolute treasure trove of information for members of the uranium community without access to paid services. Here are a few data-driven takeaways from my initial read-through of the report.
High-level: what’s in the report?
The report has an exhaustive amount of data on the quantities, the geographic origins, the type of seller, and the realized uranium price ($/lb U3O8 equivalent) for the transactions. Because the US uranium market has become increasingly dependent on foreign-origin uranium, the letter W, for “withheld,” abounds as there is so little domestic production right now that disclosing many details on domestic contracting would expose confidential information on some producers. There are three main things I pay attention to in this report:
- The realized price of uranium in each calendar year (i.e. what did US utilities pay per pound of uranium?) and how it compares to the then-current spot price
- Reported forward contracting activity (how much supply was contracted for future years?)
- The “octiles” – distribution of US uranium supply into eight groups by realized uranium price – and the “quartiles” – the distribution of end users into four groups by their average uranium price
What the report doesn’t disclose, however, is the sell-side counterparty. Uranium might be from Uzbekistan or Kazakhstan, for instance, but it does not imply that a US utility had a contract directly with Navoi or Kazatomprom. So don’t pay too much attention to the specific geographies.
Realized Price and the Octiles
Before digging into the EIA data itself, I want to share a graph from Mike Young of Vimy Resources. [Ed. note: Mike and his team at Vimy, an emerging Australian uranium producer, are very engaged in the market and extremely friendly. Worth following them on Twitter if you aren’t already, if only for Mike’s great sense of humor.] Mike is correct in pointing out that many in the space (new and old) can become overly focused on the spot price, and that year-to-year there is significant variation in the realized price of uranium, as delivered.
I’m a terrible golfer, but I think the analogy holds up. Finesse in long-term contracting can determine the long-term success of a uranium producer because, historically, the realized delivery price of uranium doesn’t track spot.
The spot price gets a lot of attention because a significant amount of publicized business happens at that price, but even in busy years, the spot market has only accounted for a quarter of the pounds consumed by end users. That’s not the whole story, though:
- Some number of sell-side organizations use the spot market to buy the pounds which they provide on long-term contracts.
- The delivery price in many contracts is spot-referenced (the exactly ratio of spot and fixed prices are determined in the terms of the contract).
My only gripe about the above graph is that it doesn’t dig quite deep enough to tell the whole story. Given the length of some vintage long-term uranium supply contracts, it shouldn’t be shocking to anyone that some utilities got nervous and booked some long-term supply at the pre-Fukushima prices of 2010 and 2011 and those contracts are still delivering to this day (and if you listen closely enough, you may even hear the tortured screams of nuclear fuel buyers past). But how do we dig deeper? By graphing the octiles, of course!
Displayed this way, the graph tells us a few more things than just the range of prices:
- The most expensive pounds in a bull market are bought at market price; the cheapest pounds in a bear market are bought at the spot price. But imprinted on each market are long-term contracts from the previous pricing regime.
- The average price is right in the middle of the two lines on the Vimy chart, but creeping downwards towards spot as older contracts fall off. This shouldn’t be a surprise: as these older contracts retire, mines have been placed offline because producers aren’t getting paid 2010 prices for 2018 pounds.
- Many producers did not capture the upside of the uranium bull in 2006 and 2007 until several years later.
It goes without saying that my graph here is far less visually-appealing than Vimy’s original and doesn’t really tell that different of a story, so I think the point stands.
Future Contracting Data
The other important piece of data from the UMAR is reported future contract commitments. In other words, utilities report to the EIA what commitments they made for future purchases in 2019. In recent reports, this is shown in Table 9 of the report (it’s Table 18 in very old reports).
Between W (withheld) information and the considerable quantity flexibility offered in some contracts (look at 2027!), this data only has limited use. But if you were looking to verify the “future coverage” rates of utilities reported by other market reporters, you can construct a forward book of utility contracts by using this Table 9 data from many years of EIA reports. For instance, here was the estimated coverage of 2019 using EIA reports from 2009-2018, plus the 2019 number for spot:
An astute reader will point out that there were actually 48.3 million pounds of U3O8 delivered to US utilities in 2019. Maybe there was a tendency to flex up on contracts, or maybe there were optional deliveries signed in earlier years which were not reported to the EIA. Maybe deliveries from past years were deferred, or future deliveries advanced. In any case, being within 10% of the actual number with such a simple data set is good enough, in my book, for my next trick…
Today’s Spot Price Matters…Just Not Today
We can cobble together a pretty good picture of when the pounds delivered in 2019 were purchased using only the EIA UMAR reports. I decided to do an experiment: could I estimate the realized uranium price in 2019 using only the contracting data and a few well-reasoned assumptions? The actual average realized price was $35.59/lb, so it seems that I can get close.
Let’s talk about my assumptions and methodology:
- For each projected year, take the linear average of minimum and maximum as the purchased quantity of uranium.
- Between 2009 and 2015, assume all forward pounds are sold into traditional long-term contracts that escalate from (spot + $3) at a rate of 2.5%/year. Assume further that the average long-term contract has a 25% market component (with 3% discount to market) at time of delivery.
- For pounds purchased in the bear market from 2016-2018, assume that they were carried forward by traders and financials at an escalation rate of 5.5%/year.
- Then, volume average the realized prices, including the market component for the long-term deals, into a realized average price ($36.71).
This is a very quick and dirty way of doing things, but it’s surprisingly accurate. I decided to carry forward the same assumptions for 2012 and 2018 to see how accurate the assumptions were.
For 2018, using a similar LT/carry trade split as for 2019, the results hit the mark. Not perfect, but using all freely-available data I am happy with the result.
2012 looked considerably less accurate, even with some minor corrections to the LT premium (e.g. term traded at a discount to spot in 2007 and at spot in 2008). Considering some of the “hidden” things going on in the market between 2005 and 2012, I’ll consider this one a win given the low quality of my half-assed nuclear forensics.
So why do we
give a f*** care about this?
Because, in addition to making sense of the multi-year formation of delivery prices in the past, projecting the delivery prices from different years also gives us the ability to look at future contract years and where realized prices may go.
With additional assumptions of $32 average spot price for 2020 and about 10 million pounds of volume, the realized price in 2020 might be around $33.11/lb. Feel free to come back here in 11 months to check if I was accurate.
I really want to bring this last point home: between 2003 and 2012, only about 1 in 6 pounds were purchased by utilities on the spot market. That number has crept to 1 in 4 in recent years. Reported coverage levels for 2021 (at the end of 2019) are only 46% of estimated 45 million pounds of annual requirements, and inventory levels are nearing all-time lows. One could project a potential 15 million pounds (or more!) or spot market demand. A historically-significant number of pounds still must be contracted for the years 2021 through 2023 as compared to previous forward coverage rates…all of which will, in some way, be leveraged to the market price when they are purchased. And if prices go up…the average realized price will quickly increase along with it.
We can compared the reported coverage of delivery years 2020 and 2021. Even with below-replacement levels of supply and inventory drawdowns…there are a lot of pounds left to buy. The picture does not get better when looking at current forward coverage rates of future years. To compare, 2019 was 30% covered by the end of 2014, so there appears to be a lag in future coverage even in the deep out-years.
Anyways, I hope this data is helpful. There’s a ton more in the report to look at, especially by using previous years’ reports to construct an open-source compendium of uranium knowledge. Feel free to leave a comment on Twitter (@808sandU3O8) or at 808sandU3O8 at gmail.com. Thanks for reading!