Now You See It, Now You Don't - Is Propane Production Increasing Or Decreasing?

Understanding whether propane production is up or down over the past few months is a bit more difficult than you might think, depending on which set of EIA numbers you choose to look at. The U.S. Energy Information Administration provides monthly numbers on the last day of the month lagged by about two months, and weekly numbers on Wednesdays, lagged by only five days. Both time series are closely watched by the propane market to assess the availability of supply for retail customers, petrochemical feedstock demand, and exports. Usually, these two sets of numbers move in tandem. But not always. The monthly numbers show production down by about 70 Mb/d from April to June, which is what you would expect given what was happening with crude and gas production at that point in time. Yet EIA weekly production numbers showed production increasing by about 90 Mb/d for the same period. So which way is propane production really trending? If you want to understand what’s going on, and you don’t mind delving into some deeply wonky NGL analytics, hang on for today’s blog.

In recent weeks, we’ve done a few blogs about the changes that the U.S. propane market has been undergoing over the past few months. In our Hold on to Your Hat blog series, we talked about how propane exports from the Ridley Island Propane Export Terminal (RIPET) in British Columbia have tightened up surplus supplies that previously moved out of Alberta to U.S. markets via rail to the Midwest and Pacific Northwest — as well as through the U.S. to Mexico. It’s a similar story on the East Coast, where growing exports out of Energy Transfer’s Marcus Hook terminal near Philadelphia have been ramping up following completion of the Mariner East 2 pipeline from the Marcellus/Utica production area to the terminal in December 2018. Until then, surplus propane from the region had been fanning out by rail to the Conway, KS, NGL hub and across propane-consuming territory, all the way down to Florida. Like Alberta surpluses, exports have since dried up those Marcellus/Utica surplus rail volumes too (see Caught in the Balance).

​

All of which makes understanding the trajectory of production growth — or decline — critically important as the market prepares for the winter of 2020-21. After all, it might actually get cold this year.

Before getting into the different stories that weekly and monthly numbers are telling us, we need to do a deep dive into what makes these two time series different, both in terms of what gets included and what the data sources for the numbers are.

Weekly vs. Monthly EIA Propane Statistics – What’s In and What’s Out?

First some basic background. Propane is produced at natural gas processing plants and at refineries. As shown in Figure 1, since the onset of the Shale Revolution, essentially all of the increase in U.S propane production has come from the processing of natural gas. While gas plant production has increased from 630 Mb/d in 2011 to 1.7 MMb/d in 2020 year-to-date, refinery production has been relatively flat, averaging 290 Mb/d over that nearly 10-year period.

Figure 1. U.S. Propane Production by Gas Processing Plants and Refineries. Source: EIA

That’s an important thing to keep in mind when looking at EIA propane statistics. The weekly numbers are published each Wednesday at 10:30 a.m. Eastern Time as part of the Weekly Petroleum Status Report.  EIA provides estimated production for PADDs 1 (East Coast), 2 (Midwest), 3 (Gulf Coast), and 4+5 combined (Rockies and West Coast). These production numbers represent “Refiner, Blender, and Gas Plant Net Production of Propane and Propylene.” In other words, the values combine (and do not break out) production from gas processing and refineries, and they throw in propylene from refineries to boot (but not propylene from chemical companies). So the weekly numbers reflect more than just propane produced from gas processing, and you must take that in to account when using the numbers in analytical models.

EIA weekly propane/propylene production stats are plotted in the left graph in Figure 2, with the last update posted on Wednesday, August 26. Production of propane and refinery propylene averaged just over 1.9 MMb/d in early May and increased to 2.3 MMb/d in July. Two weeks ago, production dipped to 2.2 MMb/d before bouncing back to almost 2.3 MMb/d last week.

Figure 2. Weekly and Monthly Propane/Propylene Production. Source: EIA

In contrast, the monthly numbers, which come out at the end of each month two months in arrears (June 2020 numbers were posted yesterday, 8/31/20), break out all the pieces of the puzzle: propane from gas processing, propane from refineries, and propylene from refineries. The data is more useful because it is split out and thus easier to interpret, but it’s less timely because it’s two months old. The good news is that refinery production of propane and propylene usually tend to be relatively stable, at ~290 Mb/d and ~285 Mb/d, respectively, on average. Consequently, we can add monthly propane from gas processing, propane from refining, and refinery propylene and come up with a value that is generally comparable with the weekly production number.   

That’s what is shown in the right graph in Figure 2. On this basis, monthly and weekly propane production stats generally tracked each other between January 2018 and January 2020. Things get a bit more volatile after that, with weekly production stats dropping from 2.4 MMb/d in January of this year to 2.0 MMb/d in May, before shifting into growth mode, and moving up to average 2.3 MMb/d in August (dashed purple circle). Monthly production, in turn, was 2.3 MMb/d in January and bounced around that level through April before falling to 2.0 MMb/d in May, then moving back up to 2.1 MMb/d in June.

​

Weekly vs. Monthly EIA Propane Statistics – Where’s the Data Come From?

So why is the data reported monthly not simply the accumulated average of the weekly data? Therein lies one challenging issue with EIA propane data, and a number of their other statistics — namely, they come from different sources. And, in the case of propane, quite different sources.

Monthly propane data is sourced from the revered (at least by NGL analysts) EIA-816 survey form. U.S. law [15 U.S.C. §772(b)] requires that the operators of all facilities that extract liquid hydrocarbons from a natural gas stream (e.g., natural gas processing plants) complete and file this form within 20 days after the end of the reporting month. Gas processing plants produce a mixed stream of NGLs (y-grade), but the form requires processors to report NGL components as determined by chemical analysis (i.e., ethane, propane, normal butane, isobutane, and natural gasoline). This means that while the propane production reported to EIA on form 816 is not propane ready for your BBQ grill (see The Long and Winding Road), it will be eventually, once the y-grade is fractionated and the propane is moved to market.   These production numbers are rock solid, since statistically the 816 data is a complete census (meaning all plants fill out the form) and not just a representative sample like EIA is sometimes forced to take due to the cost of data collection.

Unfortunately, the EIA-816 form is only a monthly thing. In order to come up with a production number for weekly reports, the EIA uses a sample. But it’s not a sample of gas processing plants. Instead, EIA uses a form filled out by fractionators and refineries: the EIA-800 Weekly Refinery and Fractionator Report.  The EIA-800 survey form includes a field titled “propane/propylene production,” which explains why the weekly report aggregates the products together. Each week, the EIA uses this number in a statistical modeling process to back into a production number, derived from the fractionator and refinery data input.

So what’s the problem? After all, over time, the propane produced at gas plants as a y-grade component gets fractionated, right? So the weekly and monthly numbers ought to converge. That’s true. But in any given week, or over several weeks, the trajectory of the numbers can differ significantly. There are lots of reasons why. Let’s consider one example. Say midstream companies that operate fractionators build y-grade inventory due to fractionator capacity constraints or other limitations on fractionator output. In the weekly EIA numbers, that would show up as lower production, even if actual gas plant production was ramping up. Conversely, if midstream companies have cranked up fractionator utilization and are working off inventories, weekly EIA statistics — again, collected from fractionators and refineries — would indicate high levels of production, even if gas plant production of propane was actually declining.

Worse yet, EIA’s weekly survey of propane inventories also requires that holders of those inventories (like Mont Belvieu midstream companies) report y-grade inventory by component, not as a separate y-grade volume. So there’s no way to tell the difference between propane held in storage as a component of y-grade (which can’t be used to heat a home or run in a petchem plant) versus propane that has been fractionated and is ready for shipment to market. The combination of fractionator-based production estimates and the inability to distinguish between usable propane and propane entrained in y-grade inventory makes the propane market considerably more opaque than it appears from the outside looking in.

You might ask, doesn’t EIA reconcile the weekly numbers to the monthly numbers after the monthly numbers are posted? The answer is no. As with all EIA time series data, the weekly numbers live in their world, and the monthly numbers in another. For the most part, there is no process to synchronize the two sets of numbers after the fact. Of course, EIA is always improving its models, and what the agency learns from the disparities between the weekly numbers and the monthly ones can improve its processes. 

A case in point. There is a proposed new survey form: the EIA-806 Weekly Natural Gas Liquids Report. It would go a long way toward fixing the problem and give all segments of the industry a better understanding of propane market developments. But for a variety of reasons too abstruse to cover here, the new form has not been rolled out.

So for the time being, we are stuck with weekly EIA propane production numbers that may not be an accurate reflection of production. This situation can also result in other statistical issues that can sometimes make the weekly numbers even less representative of true propane supply and demand. In times of market stability, such aberrations are usually not a big deal. But, right now, the markets are anything but stable. 

The moral of this story is that you really need to understand the EIA statistics that our industry relies on to make decisions about producing, stocking, transporting, and using propane. Otherwise, you are at risk of assuming the market is moving up, when instead it is moving down, or vice versa. 

Now that you’ve made it through all the wonky NGL analytics, what’s the bottom line? Is propane production increasing or decreasing? We think there is a pretty good chance that propane production declined during the April-May timeframe, along with associated gas from crude oil plays and wet gas from other basins, including Marcellus/Utica. As producers have brought shut-in wells back on, gas production has increased, as has propane. So production likely did move up in June and is at least flat — if not up some — since then. But we are not out of the woods. With the rig count and well completions down hard, the natural decline of most shale basins is dragging down gas production and propane along with it. So as we move into the winter, there is a good chance that production will again decline, and probably not recover until mid-2021. There are a lot of variables that can make a big difference, not the least of which is the price of natural gas, which despite Monday’s slight decline of $0.03/MMBtu is still up 60% over the past month. Just about anything could happen. So strap on your seat belt. This is shaping up to be a wild and woolly propane heating season.

​

"Now You See It (Now You Don't)” is credited as being written by Ozzy Osbourne, but in fact was co-written by Bob Daisley and Ozzy Osbourne. The song appears as the third tune on Osbourne's third solo studio album, Bark at the Moon. Although all of the songwriting credits on the album are attributed to Ozzy Osbourne, bassist Bob Daisley and guitarist Jake E. Lee have said that they wrote most of the lyrics and music on the record. They claim that they signed a buyout contract with Ozzy's wife and manager Sharon Osbourne forgoing any songwriting credit on the record. Personnel on the record were: Ozzy Osbourne (vocals), Jake E. Lee (guitars, backing vocals), Bob Daisley (bass, backing vocals), Tommy Aldridge (drums), and Don Airey (keyboards). 

Bark at the Moon was recorded at Ridge Farm Studios in Rusper, England, in early 1983. Produced by Ozzy Osbourne, Bob Daisley, and Max Norman, the album was released in November 1983. It went to #19 on the Billboard Top 200 Albums chart, and has been certified 3X Platinum by the Recording Industry Association of America.

Ozzy Osbourne is an English singer, songwriter, and television personality. He came to prominence as the lead singer and front man of British heavy metal band, Black Sabbath, and has also had a successful career as a solo artist. He has released 10 albums as a member of Black Sabbath, as well as 12 studio albums, five live albums, seven compilation albums, five EPs, and 63 singles as a solo artist. Osbourne has won one Grammy Award, and is a member of the Rock and Roll Hall of Fame with Black Sabbath, and the UK Music Hall of Fame as a member of Black Sabbath and as a solo artist. In the early 2000s, Osbourne became a reality television star, appearing as himself, along with wife Sharon and two of their three children, Kelly and Jack, on the hit series, The Osbournes. He still records and tours, and is currently at work with producer Andrew Watt on the follow-up album to his February 2020 release, Ordinary Man.