A Closer Look At The Oil Glut
February 02, 2016
Crude oil prices continue to fall with the WTI benchmark price recently dropping below US$30 per barrel, a level unfathomable just 18 months ago when WTI was trading over US$100 per barrel. We all know the touted reason for the price collapse: oversupply. According to the International Energy Agency (“IEA”), the world is producing approximately 1.5 million more barrels of oil per day than it consumes. If so, those excess barrels must be going somewhere. In this commentary we try to find them.
SECTION I: HALF OF THE WORLD’S EXCESS PRODUCTION IS UNACCOUNTED FOR
SECTION II: WHERE IN THE WORLD ARE THE MISSING BARRELS?
SECTION III: WHAT IF THE IEA’S DATA IS WRONG?
SECTION IV: PARALLELS AND LESSONS FROM 1996-1998
SECTION V: CONFUSIONS, CONVICTIONS, AND CONCLUSIONS
For context, the world consumes approximately 95 million barrels of oil per day. The chart below shows that supply started to exceed demand in 2014, according to the IEA.
Zooming into the quarterly data for the last few years illustrates the extent of the IEA calculated oversupply situation that materialized in 2014 and has been persistent ever since.
820 million barrels of excess supply accumulated since 2014 and counting
Approximately 1.5 million barrels per day might not seem like a lot of extra barrels in the context of global consumption of 95 million barrels per day, but keep in mind this is per day. On a cumulative basis since 2014, average excess supply of 1.3 million barrels per day equates to 820 million barrels as of Q3’2015 (Figure 3.) This total will exceed 1 billion barrels when Q4’2015 data is released.
The difference between supply and demand should equate to a change in inventories. If supply exceeds demand those extra barrels must be stockpiled somewhere. The IEA is able to track inventory levels in OECD countries, as well as Floating Storage / Oil in Transit (“OIT”). The plug in the IEA’s model to balance demand, supply, and inventories is Miscellaneous to Balance (“MTB”). The IEA claims that MTB represents unreported OECD inventories and non-OECD inventories. While this is true, the plug also incorporates any errors in the IEA’s estimates. For example, if the IEA underestimates demand, MTB will be too high and will ultimately be revised later or simply left “as is” in the MTB category. More on this later in Section III.
Half of the excess supply accumulated since 2014 is unaccounted for
Here is the provocative part: of the 820 million barrels of excess supply since 2014, the IEA can only account for half of it with 407 million barrels going into OECD stocks plus Floating Storage / OIT. The rest is simply accounted for as MTB.
The obvious question: where did the other 412 million barrels go? That is a lot of “Missing Barrels” represented by the MTB plug in the IEA’s model as of Q3’2015.
Why is this important? If the 412 million MTB barrels (and growing) are going into storage, that is a lot of excess inventory (above and beyond the known OECD inventory builds) that will have to be drawn down over time. This would constrain the magnitude of any price rally and extend the downturn for the oil & gas industry. If on the other hand, the 412 million barrels, or at least a significant portion thereof, are actually a function of underestimated demand, overestimated supply, or structural events such as strategic reserve builds, then oil market fundamentals are tighter than official reports and oil prices could move higher and faster than most forecasts.
The two main possibilities worth exploring to explain the large number of MTB barrels are:
- The oil went into inventories in non-OECD countries, which include China, Russia, India, Singapore, Brazil, the Middle East, and others that comprise approximately half of the world’s consumption.
- The IEA’s supply/demand estimates are inaccurate.
For now, assume the IEA’s data is correct (hint: it isn’t) and consider the possibility of oil going into non-OECD inventories. The data is limited but we can piece some things together.
It doesn’t make sense for most net exporters to be storing production
A net exporter is unlikely to produce oil just to store it because leaving it in the ground is the best and cheapest storage around. That eliminates the Middle East, the world’s largest aggregate producer and also a significant consumer of oil (8 million barrels per day) from being a significant hoarder of oil. It also eliminates the Former Soviet Union (FSU) countries which account for 14% of the world’s oil supply and 5% of the world’s oil consumption. OPEC’s data verifies that the FSU’s net exports essentially equate to production less consumption (Figure 5). An analysis of the financial statements of the two major Russian integrated petroleum companies, Rosneft and Gazprom, also reveals no indication of a significant inventory build.
There are only a few significant non-OECD net importers of oil
Of the non-OECD countries, only China, India, and Singapore have net imports of more than 1 million barrels per day. The rest of the non-OECD countries are fairly small and we can reasonably assume import oil as needed on a just-in-time basis given most of these countries are developing nations with limited resources.
China is building its strategic oil reserves, but…
We can determine that China is not stockpiling oil commercially beyond normal levels. China is the second largest consumer of oil at 11 million barrels per day or 12% of world consumption. According to OPEC’s data, China’s commercial stocks are up 42 million barrels from December 31, 2014 to September 30, 2015, however that includes a seasonal impact with an inventory drawdown in the fourth quarter of each year and a large build at the start of the year (this seasonal affect appears to hold true for Q4 of 2015 with November 2015 data showing a 22.9 million barrel decrease compared to the end of September 2015). Further, China’s oil consumption has increased by ~10% since 2013 and its refining capacity continues to grow, implying that the corresponding increase in oil stocks over this time frame is likely a function of normal working storage for refineries.
What about China’s strategic oil reserves controlled by the government? Numerous reports and articles cite China buying up some of the world’s excess supply for its strategic reserves. By reconciling OPEC’s numbers for China’s demand, supply, and net imports, we can calculate 102 million barrels since 2014, or 160,000 barrels per day on average, of implied stock builds (Figure 7). Note however that the implied increase in China’s inventories has tapered off in recent quarters despite the world’s touted excess supply situation intensifying.
We can at most attribute 102 million barrels of the 412 million MTB barrels going into China’s inventories. However, with most going into China’s strategic reserves this is price bullish because strategic reserves are not a temporary means of storage. These barrels won’t come out and flood the market when oil prices recover; rather they are structural barrels that won’t ever see the light of day unless there is a severe global oil supply disruption or some sort of crisis in China.
India is also building strategic oil reserves but on a much smaller scale with no meaningful stockpiling
India had three projects in the works in 2015 to add 39 million barrels of strategic reserve capacity, however only one of the projects with 9 million barrels of capacity was completed in 2015. The other two are reportedly not “anywhere near completion” (source: New India Express; January 2016). Also from the same source:
According to Amit Sharma, oil analyst with CommTrade, “Though steps are taken by the government for strategic reserves and setting up of underground crude oil storage facilities, […] we are not equipped today.” He also said that refiners are already suffering from inventory losses and they will not try to speculate the price.
Oil in Singapore’s storage facilities has modestly increased
Singapore is a major oil transport/refining/trading/storage hub with 129 million barrels of storage capacity (source: International Enterprise Singapore). According to OPEC, Singapore’s product stocks were 55.3 million barrels as of September 30, 2015, a 17.3 million barrel increase since the beginning of 2014.
Oil being stored offshore in tanker ships likely doesn’t explain a meaningful amount of MTB barrels
Where else could the excess oil be going? One touted theory is the oil is being stored on oil tanker ships or “floating storage”, perhaps as much as 100 million barrels (source: FT; November 2015).
There is no doubt that offshore tankers have been used to store oil at times over the last couple years. For example:
At anchor just a few kilometers off the coast of Malaysia lies the TI Europe, brimming with about 3 million barrels of oil destined for China. -CNN; June 4, 2015
Traders said that Vitol, the world's top oil trader, has booked two 90,000 tonne tankers off the coast of Britain to store diesel volumes for an extended period of time. Royal Dutch Shell has also booked a 90,000 tonne tanker to store product outside the New York Harbour, according to traders and shipping brokers. -Reuters; October 9, 2015
[…] seven cargoes, accounting for 13 million barrels of oil, [are] sitting in in the Arab Gulf, suggesting that either storage is short, or there are not enough willing buyers. […] In Singapore, Reuters reported that oil traders took advantage of contango by contracting out seven VLCCs in September for fuel oil storage, at a time when supply was up and demand was weak. -The Fuse; October 22, 2015
The level of crude oil stashed at sea is nearly double what it was earlier in 2015. “Onshore storage is not quite full but it is at historically high levels globally,” David Wech of JBC Energy told the FT. “As we move closer to capacity that is creating more infrastructure hiccups and delays in the oil market, leading to more oil being backed out on to the water.” […] at least 10 oil tankers from Iraq are set to arrive on U.S. shores in November, a delivery of around 19-20 million barrels. -Oilprice.com; November 12, 2015
Also supporting the floating storage theory is the fact that tanker rates increased in 2015.
Unfortunately, the exact number of additional barrels being stored on tanker ships is not readily verifiable due to the complexities involved in the shipping industry and a lack of transparent data on a global scale.
The IEA does track Floating Storage / Oil in Transit in its model with a calculated increase of 27 million barrels since 2014. Note that this is outside of the MTB plug.
While it is quite possible there are some additional barrels on tanker ships beyond what the IEA tracks, perhaps in the millions or even tens of millions of barrels, there are other factors to consider in the context of there being a significant amount beyond 27 million barrels since 2014 being accumulated in floating storage:
- Other than for brief periods and in certain cases, storing oil at sea for extended periods of time for arbitrage profits has not been justified due to the high cost of storage (increased tanker rates) and the relatively small contango in the oil price forward curve (the price of oil for delivery in the future compared to the current price).
- In regards to tanker rates, many factors go into pricing including regional supply and demand of ships, travel speed, length of routes, overall demand for oil, economics, etc. Of note:
- World oil consumption increased by approximately 1.8 million barrels per day in 2015 (source: IEA OMR; December 2015), which was a lot stronger than initially expected. Not all of the increase is transported by sea, however a significant portion is, and this increases the demand for tankers. Note that for 1 million barrels per day to be transported by sea for a year, 30-40 VLCC (very large crude carriers) ships are required given that each VLCC can hold ~2 million barrels and each ship can only lift 5-6 cargoes per year given the long distances and slow speeds the ships travel at. Only 19 net VLCC ships were added to the fleet in 2015 (source: EURONAV; January 2016), implying that at least some of the tanker rate increase could be due to a tighter VLCC market as a result of increased global oil demand.
- OPEC production increased by 1.2 million barrels per day in 2015, much of which is transported by sea, and its end markets are relatively far away, translating into increased tonne-mile demand for tankers (source: Teekay; January 2016).
- The FT article that reported 100 million barrels in floating storage included 40 million barrels of oil offshore of Iran. However, Iran’s offshore storage volumes are not new. As of October 2013, Iran had 37 million barrels in floating storage (source: IEA OMR; November 2013). Therefore, while the amount of oil Iran is storing offshore is concerning because it could come to market as sanctions against Iran are lifted, it does not contribute to the world’s excess oil supply since 2014.
- Most of the reported increases in floating storage have been due to short term reasons such as arbitrage plays or infrastructure issues as opposed to sustained buildups. For example:
- The seven VLCCs that Reuters reported were contracted out by traders for fuel oil storage in September 2015 were to take advantage of a short term contango in the forward curve. Most were charted for only 30 or 90 days (source: Ship & Bunker; September 22, 2015).
- A backlog of tankers trying to get into ports in China and the Gulf of Mexico in the summer/fall of 2015 happened during a period of lower demand and refinery maintenance season. There were also infrastructure issues at play.
- The backlog in China was largely due to record imports into the country and not enough capacity at the ports to handle the volume (source: Bloomberg; October 11, 2015). A port explosion at Tianjin in August 2015 was also a contributing factor (source: FT; November 11, 2015). The backlog in the Gulf of Mexico that built up in November quickly dissipated in December: “There currently are 27 vessels offshore in the U.S. Gulf Coast waiting to discharge an estimated 15 million barrels of crude, compared to 36 million barrels at the start of December, ClipperData said on Tuesday. There are typically 10 million to 12 million barrels of oil waiting to discharge at any one time, Clipper said.” (source: Reuters; January 5, 2016).
Our conclusion: Beyond the 27 million barrel increase already accounted for by the IEA, it is possible that some additional oil is at sea, however it is likely not substantial beyond some infrastructure issues and short term arbitrage volumes from time to time. It is doubtful that this explains a significant amount of the 412 million MTB barrels accumulated since 2014.
We can identify 128 million barrels (0.2 million per day on average) of inventory increases in China, India, and Singapore, leaving 284 million barrels (0.4 million per day on average) unaccounted for in the MTB category of the IEA’s supply/demand model.
The question is not if the IEA’s data is wrong but rather to what extent. Over 34 billion barrels of oil were consumed around the globe in 2015, a significant portion of which were shipped from one country to another via pipeline, sea, rail, or truck. To complicate matters, there are many different types of oil and the end market consists of many different refined products (that may be further transported before actual consumption). To make the gargantuan task of tracking all of this even more difficult, many important oil producing and consuming countries do not have reliable, timely, or transparent data available. Tabulating more than 34 billion barrels a year in such a complex global market is effectively impossible, yet the IEA is not only supposed to track all of this as historical data, but is also tasked with forecasting it for future periods.
Therefore, it is not a surprise that the IEA’s model of supply and demand is inaccurate and constantly needs revision as better information and data is attained over time. What we can do is analyse the forecaster’s historical track record to provide some insight into how accurate current estimates are likely to be.
The IEA consistently underestimates demand, even after the fact
There are two main takeaways from Figure 9. The first is that the IEA consistently underestimates demand. In 12 of the last 15 years, demand has been revised upwards by at least 0.4 million barrels per day. Since 2000, annual oil demand has been revised upwards by an average of 0.7 million barrels per day.
The second takeaway makes the first point even more important: this chart shows revisions starting after the year ended. For example, the demand revision for the year 2005 is the change in 2005 demand as of the IEA’s February 2006 report versus 2005 demand as stated in the IEA’s February 2009 report. This is not simply the case of the IEA underestimating future demand and then revising it as current data becomes available. Instead, this chart demonstrates that even with up to date information, the IEA still can’t accurately calculate demand for the most recent year even though it has already ended. The demand numbers for a year can change significantly even two or three years after the fact. 2014 demand has already been revised upwards by 0.3 million barrels per day since the start of 2015 – it will be interesting to see if this number is revised even further over time.
The IEA’s latest forecast for 2016 world oil demand is 95.8 million barrels per day, a 1.2 million barrel per day increase over 2015 demand of 94.6 million barrels per day. Given the IEA’s track record of underestimating demand, not only is the reliability of the 2015 number suspect, the accuracy of IEA’s 2016 forecast is even more questionable. Based on history, the likely direction of both of these estimates is up.
The IEA is more accurate with supply estimates, although revisions are common
Supply revisions tend to be of smaller magnitude compared to demand revisions. The average revision over the last 15 years is +0.2 million barrels per day.
Similar to the demand revisions, Figure 10 (and the ones to follow) also show revisions post year-end. It makes sense that supply numbers are more accurate and timely than demand given the available data from producing companies, exporting countries, pipeline throughput, etc.
Excess supply consistently turns out to be lower than initially stated
Figure 11 combines the supply and demand revision data above and not surprisingly reveals that for most years, excess supply turns out to be materially less than previously thought at the time (after that year was already over). In many cases excess supply is subsequently revised lower by 0.5 – 1.0 million barrels per day.
Given the IEA’s track record of underestimating demand, it is likely that 2014, 2015, and future demand estimates will continue to be revised upwards, resulting in a much tighter supply/demand picture than currently touted and explaining (erasing) a large portion of the 412 million MTB barrels accumulated over the last couple years. There is a significant difference between a market that is 1.5 million barrels per day oversupplied (the current IEA view) and a market that is 0.5 million or even 1.0 million barrels per day oversupplied.
“I once was lost but now am found” – MTB barrels (also: John Newton’s ‘Amazing Grace’)
When supply and/or demand estimates are revised, there must also be an offsetting change in the IEA’s model in order for it to balance. Not surprisingly, the Miscellaneous to Balance plug in the model is the balancing force.
Of note in Figure 12, the MTB revisions are almost identical to excess supply revisions (Figure 11) because the other two factors in the model are relatively accurate: OECD inventories and Floating Storage / OIT are known with more precision and rarely revised after the fact as shown in Figure 13 and Figure 14, respectively.
What would it mean for oil prices if the oil market is actually tighter than the official numbers?
The IEA’s consistency and magnitude of historical downward revisions to excess supply and the MTB plug are foundation for skepticism when analysing the current oversupply situation.
The only period in recent history that is comparable to the oversupply situation today is the late 1990’s when oil prices also fell by more than 60%.
There are a number of similarities between 1998 and 2015. For example, reasons for the decline in oil prices in 1997-1998 included:
- Concerns over demand in Asia
- OPEC increasing supply
- Technology reducing cost of new production
- Fear of a certain Middle East country increasing production.
Back in 1996-1998, the IEA couldn’t figure out where the excess oil supply was going and even coined the term “Missing Barrels”
Interestingly, there is another important parallel. Much like today, a significant portion of the excess supply from 1996-1998 was unaccounted for by the IEA. In fact, of the 1.1 billion barrels of cumulative excess supply from 1996-1998, only 391 million barrels went into OECD inventories or increases in Floating Storage / OIT. That means an incredible 720 million barrels were simply accounted for via a plug to balance the model.
The MTB barrels accumulated from 1996-1998 never were revised lower (actually they were revised higher to over 1.1 billion barrels), and yet according to the IEA, the world quickly went from 2.1 million barrels oversupplied in 1998 to 1.1 million barrels per day undersupplied in 1999. MTB went from 1.7 million barrels per day in 1998 to -0.4 million barrels per day in 1999, suggesting that the actual delta between supply and demand was never as large as reported by the IEA.
Does history rhyme?
The catalyst for the rapid oil price recovery in early 1999 was a coordinated production cut from OPEC and a few other nations, although OECD stocks had already started to decline in the fourth quarter of 1998 suggesting that the market was actually starting to balance earlier than the price would indicate.
What happens post-2015 will likely be very different than post-1998. With that said, we believe the parallels to the late 1990’s are worth keeping in mind when analysing the current oil market.
Figure 18 (reprise of Figure 4)
World oil production and consumption data is fuzzy at best and should not be taken as fact
In a world where more than 34 billion barrels of oil are consumed every year, it impossible to keep track of every barrel. Forecasting this complex market with a high degree of accuracy is unattainable.
There is no doubt there is currently an excess supply of oil
Some data is fairly reliable and we know that OECD inventories are increasing.
It is likely that the amount of excess supply is overstated and the market is tighter than forecasted
Half of the asserted world excess oil supply is accounted for by the IEA as Miscellaneous to Balance, which theoretically reflects inventories in non-OECD countries. In reality, this number is just a balancing plug and is at unrealistically high levels with 412 million barrels accumulated since 2014 and counting.
The IEA has a track record of underestimating demand. Supply is known with more accuracy and OECD inventory numbers are fairly reliable. Since 2000, annual oil demand has been revised upwards by an average of 0.7 million barrels per day.
Underestimated world oil demand could explain (erase) a significant portion of the MTB barrels and result in a much tighter oil market than most people have been led to believe.
Implications of a potentially tighter oil market
In 2016, demand is expected to grow 1.2 million barrels per day while non-OPEC production is expected to decline by 0.6 million barrels per day (source: IEA OMR; December 2015). OPEC has regained market share and there is room for Iran to increase production with sanctions recently lifted. The market is rebalancing of its own accord already and this is based on current estimates, which we know are subject to revision. If the actual imbalance is less than the IEA’s numbers, the rebalancing will happen sooner, which should drive oil prices higher and faster than current expectations. It would also likely be volatile as revised estimates are digested by the broader markets and the implications sink in.
What to watch for
The IEA issues its Oil Market Report monthly. In particular, look for upward revisions to world oil demand for 2014 and 2015 as this would mean the oversupply situation is not as severe as touted and could explain the large amount of MTB barrels. This would also mean that 2016 demand is currently estimated off too low of a base and would likely need to be revised higher as well.
In addition to growing world oil demand, the turning point for sustained higher oil prices will probably be one or more of the following:
- The start of OECD inventories withdrawals
- Non-OPEC production declining faster than currently forecasted
- Supply disruption
- OPEC coordinated production cut
"Forecasting future events is often like searching for a black cat in an unlit room that may not even be there."
-Steve Davidson in The Crystal Ball
For More Information:
Notice to Reader:
Unless otherwise stated, opinions expressed in this document are those of the author and are not endorsed by the author’s employer. No guarantee, either express or implied, is made that the information in this document is accurate, complete or up-to-date. The contents of this document are for informational purposes only and are not intended to provide financial, legal, accounting or tax advice and should not be relied upon in that regard. Commissions, trailing commissions, management fees and expenses all may be associated with mutual fund investments. Please read the prospectus before investing. Mutual funds are not guaranteed, their values change frequently and past performance may not be repeated.