Something’s got to give in the oil marketMay 3, 2016·Stefan Wieler
Something’s got to give in the oil market
Crude oil time-spreads have completely dislocated from inventories. Historically, such dislocations have proved to be short lived. We expect that either spot prices will sell-off again or the back end of the curve will move sharply higher. As per our proprietary gold pricing model, the latter would be very supportive for gold prices.
View the entire Research Piece as a PDF here...
Oil prices have rallied sharply from their January lows. At the time of writing, ICE Brent and NYMEX WTI front month prices are up 64% and 71%, respectively. While the media and commodity analysts have focused mainly on the recent rally in the spot price, in our view the more interesting development was in the curve structure. The entire move up happened in the front end of the curve. Longer-dated oil prices have remain almost unchanged. This has led to a sharp rally in crude oil time-spreads; 1-60 month ICE Brent time-spreads moved from -USD20.64/bbl in January to -USD8.82/bbl at the time of writing. In commodity markets, the shape of the forward curve is primarily a function of inventories. In our view, the inventory time-spread mechanism is the strongest and most robust relationship in commodities. Any divergence presents a physical storage arbitrage opportunity which will inevitably be exploited quickly by the market. Hence any deviation between spreads and inventories is typically very short lived. And this is where the oil market is now completely out of balance in our view. The rally in front month prices combined with the lack of any price action in the back has pushed time-spreads roughly 20-30% above the levels they should normally be. This is the largest discrepancy between time-spreads and inventories we have witnessed over the entire time-horizon in which 5-year forward prices are available. In our view, either near-dated crude oil prices will sell off again or longer dated prices appreciate.
The two possible outcomes described above point to an interesting opportunity in gold. Gold prices are driven by longer-dated energy prices while changes in oil spot prices have little to no effect. Hence a renewed sell-off in the front end of the curve would in our view have little impact on the gold price. A move higher in the back end however would be positive for gold prices. We estimate that in order for time-spreads to move back in line with inventories, either front end prices have to sell off by USD10-15/bbl or the back end has to appreciate USD15-20/bbl. Given the parameters of our gold pricing model, the latter would imply roughly a USD100-150/oz rise in the gold price.
Oil prices have rallied sharply from their January lows. At the time of writing, ICE Brent and NYMEX WTI front month prices are up 64% and 71%, respectively. Interestingly, the whole price rally took place even as petroleum inventories continued to rise. Total global commercial petroleum stocks are now at an all-time high, up 285 million barrels year-over year and 550 million barrels above the 5-year average (see Figure 1).
The market however focuses on a host of production outages in Iraq, Venezuela, and Nigeria. This hasn’t led to any decline in inventories though. Arguably more recent high frequency storage data indicates that the inventory build has slowed down on the back of a sharp decline in US crude oil production. More specifically, over the past two months, stock builds in North America, Europe and Japan have been roughly in line with the seasonal average and Singapore stocks actually drew compared to the seasonal average. But because seasonal inventories tend to build well into summer, stocks continued to build in absolute terms.
Some market participants would argue that the market is looking through current fundamentals and is pricing in future deficits. However, oil, like any other commodity, is not a financial asset. Changes in prices and price relationships have physical, real world implications:
- Higher prices encourage production. Rising spot prices will help producers, particularly in the US shale industry, to survive longer, meaning the declines in US output could be slower. Some might even bring production back. Pioneer Resources announced in their recent earnings call that they would add rigs once oil prices reach USD50/bbl.
- Crude oil time-spreads (the delta between the spot price of oil and the futures price) are a function of inventory levels. High inventories require storage holders to be compensated, which happens though a steep enough contango term structure (more about this below). The lack thereof leads to barrels coming out of storage.
While there can be a lag (and potentially a very long one) between supply and price, the interplay between inventories and time-spreads tends to be much more instantaneous. In our view, the inventory time-spread mechanism is the strongest and most robust relationship in commodity markets. Any divergence between the two presents an arbitrage opportunity which will inevitably be exploited quickly by the market (unless there is government intervention to subsidize or penalize market participants).
While the media and commodity analysts have focused mainly on the recent rally in the spot price, in our view the more interesting part happened in the curve structure. The entire move up happened in the front end of the curve. Longer-dated oil prices have almost not moved at all. This has led to a sharp rally in crude oil time-spreads; 1-60 month NYMEX WTI spreads moved from -USD20.54/bbl in February to -8.10/bbl at the time of writing. Brent time-spreads had a similar run from -USD20.64/bbl in January to -USD8.72/bbl as of now (see Figure 2).
In commodity markets, the shape of the forward curve is a function of inventories. In a market with very low inventories relative to demand, consumers - for example a manufacturer of a consumer good - of that commodity will be willing to pay a premium for immediate delivery. The alternative is to risk running out of the commodity that is an input good in the manufacturing process. The result is that the consumer (manufacturer) would have to shut in production, which represents the worst case scenario. Hence the consumer is willing to pay a higher price for the commodity if needed. For example, a battery producer needs copper in the production process. He can sell one battery for USD100 and production costs are USD90 given current prices for raw materials, labor etc. The copper that goes into one battery costs USD10. Now imagine on a global scale, copper consumption exceeds production and inventories are declining. The battery producers begins to realize that he might not be able to buy enough copper in order to keep battery production steady. There might be new copper mines being built to alleviate the shortage, but those will add new copper supply in 3 years from now while he needs copper now. So he is willing to pay more for the immediate delivery of copper even if the price for copper delivered in 3 years is still USD10. In the end, he knows that if inventories continue to fall, not all the demand can be met. Hence, the lower the stocks are, the larger the premium for immediate delivery as consumers compete for the available inventory. The battery producer will be willing to pay up to USd20 for the copper all else equal, at which point his profit become zero. But another consumer might have a higher utility from the copper and is willing to pay even more. In such a market, the spot price will be higher than the future price for delivery one month out, and that price will trade above the price for delivery two months out and so forth. This forms a downward sloping futures price curve which is called ‘backwardation’ (see Figure 3).
The opposite of backwardation is ‘contango’, where the front month (or spot) price is trading at a discount to the forward price. Contango price curves occur when inventories are high relative to demand. In a commodity market with high inventories, the likelihood for a commodity consumer to run out of the input good is very small. Hence there is no incentive for a consumer to pay a premium for immediate delivery. Quite the contrary, storing commodities costs money (to rent the storage capacity, cover insurance and the opportunity costs of money). Hence the holder of the commodity needs to be compensated for these costs. This compensation comes in the form of contango. In a contango market, an owner of storage capacity can buy the commodity at a discount now, store in and sell it forward simultaneously at a higher price. Similarly to backwardation, the higher the stocks are, the larger the discount. This is due to the fact that the cheapest storage is filled first. As inventories continue to rise, more expensive storage capacity has to be utilized. For example, when crude oil inventories rise, tank farms and refinery storage tanks fill up first. But once all that capacity is exhausted, alternative forms of storage have to be used. For example oil tankers that normally would be used to move oil from oil producing regions to oil importing regions are rented out by traders to store oil, which is significantly more expensive than storing oil in onshore tanks. And not all tankers are equal, so the more oil that has to be stored, the more expensive tankers will move to storage.
Inventories are clearly the most important driver for the shape of the forward curve. And this is true for almost all commodities (As a side note, it has typically no significant impact on the gold forward curve, as the available supply of gold is so stable. That is circumstantial evidence that gold is not a commodity; rather, it is money).
There are other drivers that impact the shape of the forward curve as well which go into our model:
- 1. The size of storage capacity: higher capacity requires less contango given the same amount of inventory.
- 2. Inventory seasonalities: Inventories follow seasonal patterns. For example, when oil stocks decline in winter, the curve doesn’t necessarily become more backwardated as long as the decline is in line with seasonal pattern.
- 3. Positioning in the futures market: Higher net speculative positions are consistent with stronger time-spreads. The common explanation is that speculators on aggregate form an opinion about future inventories and position themselves accordingly. An increase in net speculative positioning therefore would have an anticipatory effect on time-spreads, meaning that in some instances spreads can move before inventories move. When stocks subsequently decline, net speculative positioning should decline as well (because the market no longer expects even lower stocks in the future) which brings time-spreads and inventories back in line. However, we find there is an alternative explanation for the relationship between net speculative positions and time-spreads. The market on aggregate is neither long nor short. For every producer that is short, there has to be a speculator that is long, or vice versa. An increase in net speculative length can also be interpreted as producers getting shorter. If producers increasingly hedge their forward production because they expect more production next year, they will add pressure on the back end of the curve. Hence the curve strengthens, not because inventories are declining or speculators expecting them to decline, but because there is more producer hedging activity.
- 4. Interest rates: As interest rates change, the opportunity cost of money – and thus discounted storage costs – changes. This effect is relatively small in comparison to all other drivers
- 5. Credit availability / financial stress: In 2008, crude oil time-spreads crashed to levels beyond what inventories alone could explain (20% in first to second month at some point). The reason was that the credit crisis severely impacted commodity traders to take advantage of the apparent arbitrage opportunities as they found themselves cut off from financing. Credit became scarce for everybody, but even more so for traders of a good that had just dropped 80% in value in a few months. Hence, in such a situation, the limit contago is set by how low spot prices can fall until producers shut in supply.
By using these drivers in a regression analysis, we can predict the level of crude oil time-spreads extremely well. We find that while the other variables help to explain some discrepancies, inventories remain the main driver of spreads. Any deviation between spreads and inventories is typically very short lived. And this is where the oil market is completely out of balance in our view. The rally in front month prices combined with the lack of any price action in the back has pushed time-spreads roughly 20-30%, or roughly USD10-15/bbl, stronger than they should be, (see Figure 4). This is the largest discrepancy between time-spreads and inventories we have witnessed over the time-horizon where 5-year forward prices are available.
Such a dislocation happened almost exactly a year ago with reversed signs. Crude oil time-spreads crashed at the end of 2014 to levels inconsistent with stock levels. At the time, the market gave a price signal in form of spreads that inventories are so high that crude oil had to be stored on oil tankers, similar to what happened in 2008-2009. But that was simply not the case. Inventory capacity had increased substantially over the past 5 years, and there was actually plenty of storage capacity left. With the crash in spreads, crude oil spot prices had crashed too. And as spreads recovered and moved back in line with fundamentals, oil spot prices rallied. Later that year spot prices plunged again and reached new lows as inventories eventually built to a level consistent with those spread levels.
The situation this time is exactly the opposite. Crude oil time-spreads have rallied while inventories hit all-time highs. Clearly the market is expecting a tighter crude oil market in the near future given all the production outages. However, this alone does not warrant the move in spreads. It seems that the move up in oil prices was exacerbated by a general move up in commodity prices overall. Since the lows in mid-February, prices of copper, aluminum, zinc, nickel have all been moving up in a similar fashion to oil (see Figure 5).
It thus appears that investors have been pouring money indiscriminately into commodities of late. This is visible also in positioning. Open interest in WTI has reached an all-time high. Net speculative positions and open interest in WTI and Brent are near all-time highs (see Figure 6 and 7). In our view, it is investor flows that have pushed time-spreads up, hence time-spreads will ultimately revert back to fundamentals (inventories).
We can speculate about the rationale of the buyers. Some would argue that ever-more dovish central bank policies in Europe and Japan, as well as the clear shift away from the promised hawkish path in the US have heightened inflation concerns again. Arguably this should have the exact opposite effect on commodity forward curves though, as this should drive up longer-dated prices while spot prices should remain unaffected by inflation expectations.
However, while correct in theory, one has to look at how the market can express an inflation view in practice. Trading longer-dated futures is something typically reserved to hedge funds. The bulk of asset managers such as pension funds are only allowed to invest in securities, and thus if a pension fund or an insurance company seeks exposure to commodities they typically have to buy a commodity index product. Plain vanilla commodity indices are very simple products. Being long a commodity index is equivalent to being long the most near-dated future of a basket of commodities and roll the position to the next month as that contract nears expiry. While commodity index products have become more complex over the past years, allowing exposure that is further out in the curve, a large move into commodities will still mean that most money will end up in the front of the curve. In our view such an investment strategy completely defies the purpose of trying to hedge a portfolio against inflation. The performance of a plain vanilla commodity index has often very little to do with the price of the underlying commodity. This has not so much to do with the fees investment bank charge – decades of fierce competition have pushed them down to almost nothing – but more with the principal mechanics of indices (mainly negative roll yields as investors are heavily exposed to very volatile first-to second month spreads). For example, NYMEX natural gas prices are roughly at the same levels as they were over the past 20 years ago. Yet the inherent index lost 99.78% since inception. There is no arguing that this is not an efficient hedge for rising gas prices (see Figure 6).
However, that doesn’t stop investors from buying these products to express bullish views on inflation. This pushes front month oil prices higher, which is what we think has happened over the past couple weeks and what is partly behind the sharp rally in front month crude oil prices. But why is the back end of the oil curve not moving? If market expectations for inflation are shifting higher, why is the smart money that can buy the back-end of the curve not doing it? The answer is that they actually might be doing it, but there are enough sellers of the back end to keep a lid on prices for now. US shale oil producers are in a dire state. Bankruptcies are accelerating and those who have so far survived are struggling to maintain financing. Banks demand that producers hedge their production in order to keep credit flowing. So any demand for longer dated futures is currently met by producers more than willing to sell.
So what gives?
We are confident that this discrepancy between inventories and time-spreads will self-correct in the near future as well. However, the question is how. There are two ways how stocks and time-spreads could realign, either inventories drop sharply or spreads will weaken dramatically again.
We strongly doubt that inventories can be the correcting factor in the short run. While there are multiple short term production outages around the world, there has yet to be a real impact on inventories. We ran dozens of models using different type of inventories as input variables (total petroleum, excluding NGLs, just crude etc.) and they all come roughly to the same result. Total industrial petroleum inventories would have to be more than 250 million barrels below current levels to justify current spreads. Crude oil inventories (without NGLs) alone would have to be about 100 million barrels lower. Even should current production outages as well as the declines in US output persist, inventories are unlikely to fall to these kind of levels until well into 2017.
In our view, the risk to the global oil balance is that stocks decline a lot slower, or don’t decline at all. So far the focus has been on the supply side. Demand growth was very strong over the past two years, thanks to sharply lower prices. But supply growth was simply too strong, hence stocks built up. More recently the low prices have finally led to a decline in US shale oil production, one of the main sources of the oversupply (the other being OPEC). But demand growth is no longer boosted by lower prices, as prices are now basically flat year-over-year. In addition, global economic growth is stuttering. Even in the US, which was the only bright light in the global economy, there is increasing evidence that the US economy is slowing down materially. (Q1 GDP growth was a mere 0.5% annualized)
Hence, in order for time-spreads and oil inventories to realign, we believe time-spreads will have to weaken again. But there are two ways how this can happen as well. Either via a renewed sell-off in front month prices or a move higher in the back end of the curve. In our view, both can happen:
The back end of the curve is too low in our view in order to encourage enough investment in replacement production. At roughly USd50-55/bbl, there aren’t enough oil project sanctioned to ensure that future demand can be met. According to Wood MacKenzie, an energy consultant, oil companies put on hold or scrapped oil projects worth USD380 by January 2016. This accounts for 27 billion barrels in oil reserves. To put this into perspective, the US Energy Information administration (EIA) states that US crude oil reserves increased by only 15.7 billion barrels from 2009 to 2014 (17.7 including lease condensates) thanks to shale oil. And given the price collapse, a large junk of these reserves are likely no longer economical viable. The oil from these abandoned projects will be missing at some point in the future. Hence ultimately we expect longer-dated oil prices to move higher so that some of the global projects become economical again. However, what speaks against a sudden appreciation in the back end is the need of producers to hedge their forward production. Thus, any immediate rally in the back end will likely be sold. So that would suggest that the reversal in spreads will be driven by a renewed collapse in prices in the front-end.
Hence either outcome is possible in our view: either near-dated crude oil prices will sell off again or longer dated prices appreciate. What does that mean for our readers? Not many people have the ability or feel comfortable trading crude oil time-spreads. But the two possible outcomes described above open an interesting opportunity in gold. As we outlined in our framework report: Gold Price Framework Vol. 1: Price Model, gold prices are driven by longer-dated energy prices. Importantly, we found that changes in oil spot prices have little to no effect on gold. Hence a renewed sell-off in the front end of the curve would in our view have little impact on the gold price. A move higher in the back end however would be hugely positive for gold prices. Hence, the expected reversal in the crude-oil time-spreads creates an optionality for gold prices. If oil spot prices sell off, nothing happens with gold, if longer-dated oil prices go up, gold most likely goes up.
We estimate that in order for time-spreads to move back in line with inventories, either front end prices have to sell off by USD10-15/bbl or the back end has to appreciate USD15-20/bbl. Given the parameters of our gold pricing model, the latter would add roughly USD100-150/oz to the gold price. In the end, we think something has got to give in the oil market and a clever way to take advantage of this move is to build a long position in gold.
View the entire Research Piece as a PDF here...
The views and opinions expressed in this article are those of the author(s) and do not reflect those of GoldMoney, unless expressly stated. The article is for general information purposes only and does not constitute either GoldMoney or the author(s) providing you with legal, financial, tax, investment, or accounting advice. You should not act or rely on any information contained in the article without first seeking independent professional advice. Care has been taken to ensure that the information in the article is reliable; however, GoldMoney does not represent that it is accurate, complete, up-to-date and/or to be taken as an indication of future results and it should not be relied upon as such. GoldMoney will not be held responsible for any claim, loss, damage, or inconvenience caused as a result of any information or opinion contained in this article and any action taken as a result of the opinions and information contained in this article is at your own risk.