Streetwise Professor

July 16, 2018

Oil Spreads Go Non-Linear (Due to Infrastructure Constraints), To the Chagrin of Many Traders: The Pirrong Commodity Catechism in Action

Filed under: Commodities,Economics,Energy,Exchanges — cpirrong @ 3:59 pm

When I wrote about the demise of GEM Trading a few weeks ago, I hypothesized that sharp movements in various spreads had been its undoing.  A story in Reuters says that GEM was not the only firm rocked by these changes.  Big boys–including BP, Vitol, Trafigura, and Gunvor–have also suffered, and the losses have caused traders their jobs at Gunvor and BP:

The world’s biggest oil traders are counting hefty losses after a surprise doubling in the price discount of U.S. light crude to benchmark Brent WTCLc1-LCOc1 in just a month, as surging U.S production upends the market.

Trading desks of oil major BP and merchants Vitol , Gunvor and Trafigura have recorded losses in the tens of millions of dollars each as a result of the “whipsaw” move when the spread reached more than $11.50 a barrel in June, insiders familiar with their performance told Reuters.

The sources did not give precise figures for the losses, but they said they were enough for Gunvor and BP to fire at least one trader each.

The story goes on to say that binding infrastructure constraints are to blame, which is certainly the case.  But implicit in the article is a theme that I have emphasized for literally years (I recall incorporating this into my class lectures in about 2004).  Specifically, bottlenecks imply that marginal transformation costs (e.g., marginal costs of transporting oil between Cushing and the GOM) tend to rise very steeply when capacity constraints are reached.  That is, when you are operating at say 90 percent of capacity, variations in utilization have little impact on marginal transformation costs, but going from 95 to 96 can cause costs to explode, and basically go vertical as capacity is reached.

This has an implication for spreads.  Another part of the Pirrong Commodities Catechism is that spreads equal marginal transformation costs, and are essentially the shadow prices on constraints.  The behavior of marginal transformation costs therefore has implications for spreads: in particular, spreads can be very stable despite variations in the utilization of transformation assets, but as utilization nears capacity, the spreads become much more volatile.  Moreover, and relatedly, small changes in fundamentals can lead to big moves in spreads when constraints start to bind.  The relationship between fundamentals and spreads is non-linear as capacity constraints become binding, and well, here spreads have gone non-linear, to the chagrin of many traders.

Put differently, spread trades aren’t always “widowmakers” (as the article calls them)–sometimes they are quite safe and boring.  But when bottlenecks begin to bind, they can become deadly.

There is one odd statement in the article:

“As the exporter of U.S. crude, traders are naturally long WTI and hedge their bets by shorting Brent. When the spreads widen so wildly, you lose money,” said a top executive with one of the four trading firms.

Well, why would you hedge WTI risk with Brent?  You could hedge your WTI inventory by selling . . . WTI futures.  The choice to “hedge” WTI by selling Brent is effectively a choice to speculate on the spread.  That brings to mind the old Holbrook Working adage that hedging is speculation on the basis.  The difference here is that most, say, country grain elevators about which Working was mainly writing had no choice in hedging instrument (at least not in liquid ones), and perforce had to live with basis risk if they wanted to eliminate flat price risk.  Here, BP and Gunvor and the rest had the choice between two liquid instruments, and if the “top executive’s” statement is correct, deliberately chose the one that exposed them to greater spread (basis) risk.

So this isn’t an example of “sometimes stuff happens when you hedge.”  The firms chose to expose themselves to a particular risk.  They took a punt on the spread, which was effectively a punt that infrastructure constraints would ease.  They lost.

In my 2014 white paper on commodity trading firms (sponsored by Trafigura, ironically) I noted that to the extent that they speculate, commodity trading firms tend to speculate on the spreads, rather than flat prices, because that’s where they have something of an information advantage.  But as this episode shows, that advantage does not immunize them against risk.

This also makes me wonder about the risk models that the firms use, which in turn affect the sizes of positions traders can put on, and where they put them on.  I, er, speculate that these risk models don’t take into account the non-linearity of spread risk.  If that’s true, traders would have been able to put on bigger positions than they would have been had the risk models accurately reflected those risks, and further, that they were incentivized to do these trades because the risk was underpriced.

All in all, an interesting casebook study of commodity trading–what can go wrong, and why.

Correction: Andrew Gowers, head of corporate affairs at Trafigura says in the comments that (a) Trafigura did not suffer a loss, and (b) the company had told this to Reuters prior to the publication of the article.  I have contacted the editor of the story for an explanation.

Print Friendly, PDF & Email


  1. Hi Craig, nice piece (again)! Just for the record, Trafigura was incorrectly cited by Reuters as one of the firms that suffered losses. we did not, as we told Reuters prior to publishing. You could ask them why they chose to ignore this guidance! Best regards. Andrew

    Comment by Andrew Gowers — July 17, 2018 @ 3:51 am

  2. What do you make of Putin/Trump and oil? The anti-Trump people were unhinged. Garry Kasparov is saying he is a threat to the Republic and former communist voter John Brennan wants high crimes and misdemeanors for the President. I didn’t like one of Trump’s statements in the press conference, but Chris Wallace interview with Putin was telling at the end. Russia didn’t join NATO because they thought there was an agreement not to spread past West Germany. If NATO countries are smart and strategic (saying a lot when it comes to geopolitics), they’d cut purchases of Russian oil and buy American.

    Comment by jeff — July 17, 2018 @ 5:32 am

  3. Hi Craig – quick question: since BP is undoubtedly long of Brent then wasnt being long WTI and selling Brent against it simply an expression of preference to be diversified into (a slightly larger, or) partial WTI length?

    Comment by David Gerard Stack — July 17, 2018 @ 6:23 am

  4. @Andrew-Thanks so much. Great to hear from you. I will correct the record at my end. I will ask the Reuters reporters! Cheers.

    Comment by cpirrong — July 17, 2018 @ 10:01 am

  5. Prof,
    As you well know, pops in vol on spreads are brutal because of the illiquid nature of the notional amount of each individual leg getting introduced into the marketplace upon unwinding.

    Nick Leeson, LTCM, Amaranth, etc.

    A firm getting its face ripped off on a spread trade isn’t trader’s fault. Spread traders understand all spreads revert to the mean. Give any spread trader a market materially outside the mean and he will jump on it.

    The person getting fired shouldn’t be the spread trader (unless he violated his position limits). It should be the individual responsible for managing the overall book who failed to take away the punch bowl when the party was starting to get out of hand.

    Traders are paid like they are because they are identified as being the one-in-a-thousand capable of handling the intellectual stresses of the job. Senior traders are paid what they are because they are identified as being the one-in-a-million capable of managing traders.

    If you want a successful trading organization, you don’t fire the trader and retain the senior trader when the senior trader is the one who failed at his job.

    Comment by charles — July 23, 2018 @ 1:37 pm

  6. Risk Quant here,

    I know this article is a little old, but I thought I’d chime in on the following:

    “I, er, speculate that these risk models don’t take into account the non-linearity of spread risk. If that’s true, traders would have been able to put on bigger positions than they would have been had the risk models accurately reflected those risks, and further, that they were incentivized to do these trades because the risk was underpriced.”

    In addition to parametric VaR, most trading houses do run monte carlos for complex portfolios and additionally run historical stress tests that should inform the risk associated with these spreads. I would imagine the traders knew the risk they were taking. The trades went awry but I doubt the risk blindsided anyone.

    Comment by Q — November 8, 2018 @ 11:48 am

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress