An official selling price is a differential rather than a price [6], and that is what makes the $11.50 a barrel between July's premium and September's discount [12] the cleanest figure in the sector this month, because the whole of it lands in the crack spread on every Asian-lifted Saudi barrel without any product price having to move at all. Shinhan Securities' Lee Jin-myung makes the same point from the other direction when he says margins come from relative changes in crude and product prices rather than from the level of oil itself [10].
The equity tape is messier than the differential. GS at 45.17% against SK Innovation at 5.79% [2][3] is a ratio of 7.8 to one [13] between two companies buying crude off the same OSP sheet, so most of the GS move is GS, which is to say distribution subsidiaries sitting outside the refining business and dividend expectations [2], set against the financial burden of SK Innovation's absorption merger with SKIET [3]. S-Oil's 17.20% [1], 14.27 points clear of the index [14], is the closest thing on offer to unlevered exposure to the spread itself.
Which leaves the ordering question, and this is where the trade is thin. The cheap-crude leg can be reset by one seller, monthly, at the stroke of a pen [5]; the tight-product leg depends on physical units coming back on somebody else's timetable [11]. Those are not symmetric risks, and the tape is pricing the second one as though the first were fixed.
Two other readings deserve a hearing. If the inventory position reflects years of under-building rather than a handful of units down for maintenance, the gap holds for several quarters and holders get paid to wait for a convergence that keeps not arriving. If instead the restocking that analysts expect to absorb any incremental output [8] is done quickly, the same low stocks that put a floor under diesel and jet become the mechanism that takes it away.
This desk's view, offered as probably wrong on timing: buying Korean refiners at these levels is buying an outage, and the variable worth tracking is a repair schedule, not consumption. The falsification is dated and observable. If product inventories rebuild through the autumn while cracks hold where they are, then scarcity was not doing the work, demand was, and these become businesses to own on a multiple rather than a spread to rent.
