On April 30, Delta Air Lines (DAL) bought a 185,000 barrel per day (bpd) oil refinery in Trainer, Pa. from ConocoPhillips (COP). Delta's goal for this transaction is to mitigate risk stemming from the "crack spread" (the price difference between crude oil and jet fuel) and thus generate cost savings. Delta's refinery purchase has been met with a wide range of responses. Most analysts in the airline industry have been cautiously optimistic about the decision, while most analysts in the oil/gas sector have condemned Delta's move. However, with the Trainer refinery re-start coming soon, I think the purchase looks like a smart decision. Elevated crack spreads will make it possible for the Trainer refinery to earn a sizable return on investment.
In a presentation given Thursday morning, Delta president Ed Bastian pointed to crack spreads as the fastest growing cost for Delta. There have been several recent refinery closures in the U.S., particularly on the East Coast, as well as extended downtime due to various accidents. These factors, combined with some demand growth since the Great Recession, have sent crack spreads soaring. On slide 12 of his presentation, Bastian discloses that the crack spread paid by Delta tripled from 2009 to 2011. For the month of April (when the deal to purchase the Trainer refinery was negotiated), Brent crude averaged $119.75/barrel, while Gulf Coast jet fuel averaged $3.226/gallon, which works out to $135.49/barrel (data sourced from the EIA). The April crack spread was thus $15.74, though it apparently varied quite widely over the course of the month. Delta hopes to use the Trainer refinery to produce jet fuel at a lower cost than it would otherwise pay on the market. Since the company consumes nearly 4 billion gallons of jet fuel per year, a small decrease in fuel costs could have a dramatic impact on the bottom line.
Back in April, Delta announced that it expected the Trainer refinery purchase to reduce its annual costs by $300 million. However, jet fuel crack spreads have actually widened since then. While the average spread was $15.74 in April, the crack spread between Brent crude and Gulf Coast jet fuel was up to $21.50/barrel as of Wednesday. If this additional premium of nearly $6/barrel were to remain, Delta's annual benefit from the Trainer refinery could double. (I will present some more detailed calculations below.)
Critics of the decision argue mainly that the Trainer refinery was not viable for ConocoPhillips, and that there is no reason to believe an airline could run a refinery better than an oil company could. In a very interesting piece published back in May, Gregory Millman argued that for three reasons, the logic behind Delta's refinery purchase was flawed. First, by focusing on jet fuel production, Delta would be unable to switch between different product combinations based on market conditions. Adopting a rigid production target would depress the refinery's long-term return on investment. Second, the Trainer refinery was uneconomical to begin with, which is why ConocoPhillips shuttered it last year. Third, Millman argues that Delta has not factored in $100-$200 million of necessary working capital in its estimate of the total refinery investment.
There is some merit to these arguments, but they do not capture the full logic of Delta's decision. The issue of working capital is least important: if the refinery produces a $300 million or better annual savings for Delta, it will be a good use of capital regardless of whether the investment is $250 million or $450 million. The other two points boil down to the argument that the Trainer refinery will be uneconomical for Delta, just as it was for its previous owners. However, my own analysis of the cost breakdown shows that this is not the case.
Millman states that the cost to transport Brent crude to the refinery is $2/barrel, and the cost of refining a barrel of oil is $5.25/barrel. He also claims that the value of capital investments works out to roughly $1-$2/barrel. I will round this off to $9 total transportation/refining cost. At a Brent price of $113.32, this amounts to $122.32/barrel of finished products. However, only 32% of this mix will be jet fuel. The other products will be swapped with BP (BP) and Phillips 66 (PSX) for jet fuel in other U.S. markets. Jet fuel tends to be more expensive, and so Delta would not quite get 1 barrel of jet fuel for every barrel of refined product it sent to BP and Phillips 66. I will adopt Millman's conservative estimate of 1.04 barrels of refined product per barrel of jet fuel. As a result, for every barrel of crude Delta purchases and processes (at a cost of $122.32 as of 9/5), it will receive only 0.974 barrels of jet fuel. The effective cost of jet fuel would be $125.59/barrel or $2.99/gallon. The crack spread would amount to a little over $12.
By contrast, Gulf Coast jet fuel was selling for $3.21. At a 22 cent discount to the market price, Delta would save nearly $700 million annually after accounting for all of its shipping and refining costs. Even if the market crack spread were $18 (as it has been for much of this summer) rather than yesterday's level of $21.50, Delta's savings would still exceed $400 million.
On Thursday, Delta also confirmed that it is looking into the possibility of switching the refinery's crude oil sourcing (at least in part) from imported seaborne crude to domestic Bakken crude. Due to a shortage of shipping capacity, crude from the Bakken region has traded at a significant discount to Brent crude over the past couple of years. This discount is expected to remain in place for a few more years as pipeline capacity tries to catch up to production increases. In the meantime, BNSF recently announced an expansion of railroad shipping capacity to move Bakken crude to refineries by rail. If Delta can buy Bakken crude and ship it by rail to the Trainer refinery, its costs could be perhaps 10-15 cents lower per gallon of finished product, even after accounting for the higher cost of moving oil by rail. The possibility of alternative sourcing provides further upside for Delta, with no real downside (if it's not economical, Delta can continue buying Brent crude).
So why did the energy analysts get it wrong? Quite simply, Delta is a net consumer of oil products, whereas ConocoPhillips and now Phillips 66 are producers. This has two impacts. First, adding a refinery creates risk for a producer of oil products. If refining margins drop, operating another refinery would increase the losses of a refiner like Phillips 66. The increased risk makes oil companies err on the side of having less refining capacity. By contrast, for Delta, a drop in refining margins would hurt the refinery subsidiary, but benefit the company as a whole. The refinery thus acts as a natural hedge and de-risks the company. Therefore, it is unsurprising that Delta would be willing to operate a refinery that was marginal for an oil company.
Second, energy analysts assume that the market is in equilibrium, when this clearly is not the case for refiners. As Millman notes in the article I discussed earlier, U.S. refiners are earning more than twice the return on capital of U.S. airlines. As risky as the refining business is, the airline business is even riskier! The average returns should thus be reversed (in theory). It is always possible that Delta's entry to the market will cause the crack spread to narrow to $10, in which case Delta would be paying higher than the market price for jet fuel. Even so, the alternative would be not entering the market and buying at a higher market price (because without the Trainer refinery, there would be less supply). This would just mean that other airlines such as United (UAL) would benefit more from Delta's move into refining than Delta itself.
In short, I see no evidence to support the bear case regarding Delta's refinery. With the recent widening of the crack spread (probably due in part to disruptions from Hurricane Isaac), the move looks even more prescient. Furthermore, other airlines may benefit from the expansion in refining capacity, considering how tight supply has been recently. This underlines my overall bullishness on the airline sector.