EIA Overestimates US Production, Huge Production Drop in the Works

EIA uses a model based on rig counts and inventory levels to estimate weekly US crude oil production.  This model has some major flaws that arise when the mode of operation for US producers makes a drastic change.  In the current environment where price is pivoting around or is very near break even for producers and US producers are acting as the worlds swing producer by increasing and decreasing output based on price, these flaws become significant and lead to some wild overswings in prices as many traders have little idea of what is actually happening.

US shale wells have a decline rate of approximately 1%/week, 5%/month or 65%/year.  This means US producers must add enough wells every month to bring on 5% of the previous months output with new wells every month.  If producers bring on more or less wells than normal in a month EIA cannot compensate with its weekly model.

US producers temporarily store oil at wellheads and elsewhere.  If there is a dramatic shift in how much producers have in temporary storage, the EIA model does not compensate.  This becomes an issue in Gulf Coast states that have property/inventory taxes that are paid based upon end of year inventory.  This encourages producers to empty all storage leading up to the year end pumping oil into storage in other states.  This can be seen with increases in storage at Cushing, OK at year end and a draw down in Gulf Coast storage.

Another source of error appears to be where in the US oil inventory appears.  Indeed most traders treat Cushing, OK inventory as a proxy for US production so it is logical that the EIA weekly production model also treats Cushing inventory with heavier weight than elsewhere..  Indeed over the last several weeks US production has been shown to be rising as Cushing inventory has risen, while completely discounting large Gulf Coast inventory decreases.  Even though it is possible for inventory from Canadian and other inventory destined for the Gulf Coast to be temporarily stored in Cushing and for Gulf Coast oil to be pumped to Cushing to lower Gulf Coast inventory at year end.

The only method EIA uses to compensate for errors in its model is to report the error in Unaccounted for Oil which few track and is hardly ever mentioned in the mainstream media or by most analysts.

So this brings me to discuss the recent fall in US production that is drastically underreported by EIA data using data reported to the state of Texas as a proxy.  Data reported to the state of North Dakota confirms what is seen in the Texas reports.

Intuitively it shouldn't make sense that US production is little changed or even increasing even as rig counts and oil prices fall drastically.  Especially since price took a major downturn in August.  Granted the backlog of uncompleted wells can cloud this picture, but most US producers have been pretty vocal about not completing wells if price drops too far, and a look at producer financials indicates that most have severely negative cash flow at these prices.  Inspection shows many will have used up about 50% of their cash in the last year and at current prices will have no cash reserves in 6-12 months.  Who is going to lend them money to maintain operations with this kind of trend?

The first plot shows US production vs Unaccounted for Oil vs WTI spot price on Tuesday of the report week.  What should be noted is the odd significant increase in production + Unaccounted in October and November followed by the huge falloff in 2 out of the last 3 weeks.  Later it will be shown that the huge falloff correlates directly to production declines shown in both EIA monthly data, but more significantly in production and completion data reported to the state of Texas.


Fig 1:  Note the large volatility in US Production + Unaccounted for Oil since the beginning of 2015 as price ranged from appx $40-$65.  A significant point is that several big drops in Unaccounted for Oil led EIA estimates of falling production.  The size of the recent drops in Unaccounted for Oil are huge and just so happen to coincide with very large price drops, below the mid to low $40s which has been reported as the break even point for many US producers.

Data reported to the state of Texas for crude production as well as well completions further supports the case for recent production declines being under reported by EIA as well as projecting huge declines in the coming months.  It should be noted that there are appx. 180,000 producing oil wells in the state of Texas and analyzing the data indicates that appx. 1600 completions are required every month just to keep up with the severe declines in existing wells.

The distribution of wells shows for November 2015 shows 8132 producing greater than 100 barrels/day (BOPD), 45,259 with production greater than 10 BOPD but less than 100 BOPD and the large majority of wells, 130,491, producing less than 10 BOPD.  This means the newly completed wells most of which are producing near or above 100 BOPD due to high grading in this low price environment are responsible for a very outsized portion of total production since it only takes an addition of 1% of the total wells to compensate for a 5%+ monthly decline in the rest of the wells.  A failure to add some or all of these wells and/or a decline in productivity of these wells can lead to very large and fast production declines.


Fig. 2  Texas Oil Production Data  This chart clearly shows a major decline in Texas well completions starting in July/August as price declined.  Even the momentary return to prices in the low $50s didn't provide the incentive to increase completions to the 1600/month level required to maintain output.  Prices in the low $40s for the entire month of November caused completions to drop below 800 for the month and it is yet to be seen just how low completions were for the month of December.  EIA and Texas monthly production reports that lag two months clearly show declines matching the completion declines.

While EIA and Texas reported production levels tracked closely at the end of 2014 and early 2015 there has been a steady and wide divergence between the two data sets since March with the Texas data being much lower.  But if we project the summer/early fall declines using the EIA data set, it is not unlikely that Texas crude production has already dropped to 90MM bbl/month which is a 15% drop since mid summer.  This is important since the weekly data set only shows a 4.5% drop in production using production only and appx 7.4% drop using production+unaccounted.

The huge drop in rig counts and low prices provide little ability and no incentive for US producers to actually push production up quickly even if price were to quickly recover to over $50/bbl.  The data seems to indicate that even with prices essentially steady in the high $45-$50 range in September and October producers did not even attempt to maintain output, because only about 1100 completions were done per month in Texas while about 1600 were needed to maintain output.

Another point supporting huge drops in production coming is that in November the number of wells producing over 100 BOPD dropped by almost 1087 or almost 12%.   This is significant because for most months 1100 completions was enough to increase this count at least by a few and 1500+ completions added about 400/month to the count.  So a drop from 1138 completions in October to 776 in November caused this count to drop by almost 1100.

The EIA monthly reports for US and Texas production show the peak occurred in March, however the weekly data indicates the peak occurred in the May/June time frame.  Again this disparity shows how the weekly reporting lags reality, essentially exaggerating the increase in US production for months and delaying the reporting of significant declines for months.

Analyzing many of the largest producers latest quarterly financial reports showed their lowest hedging position at $46 which was commonly purchased put options and did not cover 100% of planned output.  This also supports the case for a big slowdown in output at current prices.

Conclusion
The conclusion is that as long as WTI remains below $46 the actual decline in US production will remain severe.  Recovery of output levels cannot be fast even with a large price increase.  After Jan 1 producers, refiners and storage players will return to a more normal mode of operation.  This means localized inventory levels could increase meaning less oil moves into reported storage making it look like production suddenly fell, Cushing inventory could fall or remain steady making it appear production dropped.  On the other side Gulf Coast inventory will spike in early January as a backlog of tankers is offloaded, of course we could see a huge drop in refined inventory in the next two weeks as that inventory is moved out of Gulf Coast storage compensating for a January crude spike.  At some point EIA will revise their data and adjust their models to compensate for recent inaccuracies.  Indeed the monthly reports will show huge declines that were not foretold in the weekly data.

As the huge decline in US production becomes obvious it could provide some incentive for OPEC producers, especially the Saudis, to be just a little less excited about dumping every barrel possible onto the US market.  This becomes especially true if they can get their total exports to the US close to 2013 levels.  More potential tail winds for prices in the coming months.  It's been my assertion that US production needs to drop to around 8MM bbl/day to allow room for Iranian oil and provide a larger market share to the Saudis and it appears that US production is likely much closer to that level than anyone realizes.

Largest 4 WTI Crude Traders Position 11-17-2015

Here is a chart showing the largest 4 WTI crude oil traders net position using my metric vs price.  You can see they are much longer (bullish) than they have been since this time last year.  Just this week the largest traders added 10% to their long positions, about 27,000 contracts while paring back their short positions by about 7%, 12,600 contracts.

So if you want to bet crude oil prices will drop a lot more, this is what you are against.  And week by week they get more and more bullish.  You can see multiple times in the last year, where the biggest traders became bullish and price followed their lead up.

Largest 4 Crude Oil Traders Net Positions vs. Price

Here is a plot of the largest 4 WTI traders net positions using my proprietary metric.  This is what I call a leading indicator.  You can see a clear build up in long positions ahead of this springs big move up as well as a move neutral to short over the summer while price was range bound and then eventually crashing.  Then you can see another long build ahead of the late August spike and a continuous long build now.  You can determine on your own what is coming next.  I'm betting on an up move.  Long USO, crude and gasoline futures and UWTI.  I'm more confident in gasoline futures in the coming days since gasoline inventories were down this week while crude was up.  Not to mention all the crude in storage, tankers and the general US overproduction.

More on Crude Futures Positions

I decided to post this graph of absolute futures positions for the physical players--Producers, Merchants and Processors--and swap dealers that primarily provide hedging to the physical players that choose not to do that in the open futures market.  What jumped out at me was just how correlated the short position of the producers and swap dealers was to price.  Also it is quite noticeable when the swap dealers hold large long positions price rises.

The short term bearish issue is the large short position held by the physical players.  You can see the build in short positions whenever price nears $50.  One must also take into account that the plot uses Tuesday's price to match up with CFTC reporting requirements, while the peak and trough prices for the week aren't shown.

Another noticeable issue is the physical players boost their long positions significantly whenever there were opportunities in the low $40s.  This plot shows that most clearly for the late August/early September time period, but if you look at day to day futures prices you will see many other low $40s opportunities were available as well.  Indeed a clear support level can be drawn in the low $40s on a daily chart, the futures data simply supports what is seen in the price action.

Big Money is Super Bullish on Crude

As of Tuesday 11/4/2015 hedge funds, swap dealers and the biggest 4 traders were super bullish on oil.  As of today, 11/12/2015, crude is about 10% lower.  Since late August the big money has been buying crude heavily in the low $40s.  What do you think will happen after today?

Sure there is still a glut remaining, but the main driving force behind price is US production.  If it comes down price spikes.  Currently US production+Unaccounted--I feel this provides a truer picture of production than production alone-- has been maintaining around 9.6MM bbl/day.  That is after a fall to below 9.5MM bbl/day in July and September.  The late summer fall in production as WTI cratered indicates to me that producers got caught off guard by the early drop in price and didn't have much in place for hedges in August.  Indeed each time price reaches the $50 region you can see a spike in short positions by the physical and swap dealer groups.

If you look at the latest data you can see the swap dealers are far more bullish on crude than ever, as are the 4 biggest traders, and managed money is becoming more bullish as well.  I expect we will see a huge spike in long contracts from the physical group in the next two weeks as refiners and storage players binge on cheap oil.  Futures spreads are now over $1 so there is a lot of incentive to buy and store whatever is available.  But producers have little incentive to bring on new wells if they can't lock in a price near $50/bbl.

Decline in US Working Age Population vs Reported Jobs

total non farm long term 2015-11.xls

Some have downplayed while others have played up the affect the demographic shift in the US has had on the economy and jobs.  So I finally decided to do a comparison of the job situation vs the 19-64 year old working age population.

Using data provided by BLS and Census Bureau I created the above plot.  The top line is the working age population which has an obviously slowing growth rate.  The bottom lines show reported jobs using an annual monthly average and an expected jobs growth rate based on the 2007 employment peak when adjusting for the aging demographic.  You can see the estimated jobs growth line, the blue line, tails off matching the working age population line.  You can also see actual jobs are now almost inline with the expected jobs total based on growth of the working age population.

This chart both supports how incredibly slow the recovery has been, but also calls into question why the Fed has been waiting so long to raise interest rates off near zero using the very poor employment situation as an excuse, meanwhile stocks are soaring because of the cheap money used to inflate valuations with buy backs and margin buys.

Going forward I believe most will be shocked by how fast the Fed will be forced to raise interest rates as we are now nearing full employment which will drive up labor costs and drive inflation.  If things play out the way I see it in the oil patch, energy prices will soon be shooting up as well as US oil and gas production falls off as hedges expire this year.  Low energy prices may be the prime factor in holding down inflation this year, but could easily turn into the prime factor driving up inflation in 2016 and 2017.

I am short treasuries via sold call options on TLT and long TBT.

S&P 500 Historical View: Bubble or Not

Here is a chart of the S&P 500 that I use to calibrate myself to current market conditions.  I have similar charts for the DOW and NASDAQ Composite.  I use a log chart and then draw a price channel to get a view of where stock prices typically run historically, and what performance to expect when certain "trend" levels are hit, especially the top and bottom trend lines of the channel.

There are a lot of arguments about whether current conditions can be defined as a bubble.  What should be clear is that in the very long view the current price level is very high.  But one might say compared to past performance this century the price isn't that high compared to the last two peaks.  But if you look at the long term history showing many quite serious corrections when price is above the 4th trend line and up against the long term trend resistance line at the top, it shouldn't give one a lot of confidence the market has a lot of room to run.

Another way to think about it is through the lens of a long term billionaire investor, or a fund manager of billions.  That kind of money needs to be calculating the expected return on investment over a five, ten or even longer time period.  One of the methods would use the expectation of a regression to mean which would be the middle trend line, along with historical performance at various price "trend" price levels.  So do you think billionaires are approaching current market conditions with extreme confidence when in long term historical terms price is near the 100th percentile when the obvious bubbles are left out.  This is also why many analysts are predicting very low stock returns for the next 5-10 years because the probability of a return to mean is very high.  The current price level is at the mean expectation for 2022, 7 years away.

Some might argue my trend channel is too conservative.  I plotted a separate channel that enclosed the 2007 peak below the top channel line.  It actually puts current conditions around the 75th percentile rather than near the 100th percentile.  But I discount this case because of the very extreme correction of 2008.  This tells me that the price levels from 2004-2007 remained extreme by historical measures and 2007 was a very extreme peak, but not as bad as 1929/1930 or 2000/2001.  I see a return to historical norms/means in stock prices, albeit a very slow one.

Generally it seems we are at a point that few recognize when the markets are expensive.  I believe we are still suffering from a hangover from the dotcom bubble, where there is a continual comparison to the dotcom bubble which was a true 1929 type of event, while the more appropriate comparison for current conditions is 2007/2008, the 1987 flash crash or conditions in the 1960's and 1970's.

2015 US Employment trend almost identical to 2014

2014-21015 Employment Trend Comparison using non-seasonally adjusted data.  2014 data was normalized so 2014 and 2015 could be laid directly on top of each other to make any trend differences easier to see.

All year BLS has been reporting that employment gains have been slowing drastically.  Seasonally adjusted data clearly indicate this.  A recent MarketWatch article--link to follow--states "the US added an average of 206,000 jobs a month this year vs. a 15-year hig of 260,000 in 2014.  Well as you can easily see in the chart above using non-seasonally adjusted data, you can see 2015 trended almost exactly with 2014.

http://www.marketwatch.com/story/new-reports-suggest-october-jobs-blowout-is-an-outlier-2015-11-09

Being a math guy I also like to crunch the numbers.  Using average monthly employment 2012 & 2013 showed a 1.7% gain, about 190k/month, 2014 showed a 1.9% gain, about 220k/month.  But if we compare the average monthly employment for the latest 12 month period vs the previous, the data shows a 2.1% gain, about 250k/month.  Comparing monthly using a year over year (YoY) comparison the same trend is seen with monthly gains around 240k-260k/month for the last year.  The YoY monthly comparisons shows a slight downtrend from 2.4% early in the year to 1.94% in the latest month.

The plots of the raw BLS data and mathematical comparison just does not support the claim that employment gains have slowed drastically in 2015 compared to 2014.

I hope everyone is expecting a big job gains number for October.

I hope everyone is expecting a big job gains number for October.  BLS has some making up to do.  If you look at the above graph you will see multiple years of raw, non-seasonally adjusted employment data plotted.  This data was downloaded from the BLS website.

Just looking at it you can see not much has changed in terms of the employment trend compared to 2014.   Yet this year as rate hikes seemed imminent because employment was strengthening suddenly the seasonally adjusted jobs gains reported by BLS nosedived.  The 2014 average reported seasonally adjusted gain was about 240k/month, while 288k/month was reported from Sept 2014 to Feb 2015.  But then suddenly March 2015 was reported at 126k, May, June and July rebounded but still all but June were below 230k.  Then August and September were 173k and 142k respectively. Since March seasonally adjusted gains have averaged just 197k/month while the comparable period in 2014 was 242k.

If you look at YoY calculations on the raw data, every month since June 2014 has been showing a gain of 1.94%-2.39% which is about 240k-260k/month.  If you average employment for the last 12 months and compare to the same period a year earlier you come up with job gains averaging 251k/month for 2015 plus the last 3 months of 2014.

Yes if looks like monthly job gains have slowed from a peak of around 260k/month to about 230k/month now.  However this slowdown has been drastically overstated by the BLS seasonally adjusted data which is showing gains well below 200k/month lately.  It appears the BLS seasonal adjusted data is 20%+ low for the year.

Another point of interest is that in 2014 BLS seasonally adjusted reports showed average monthly gains were about 240k/month while YoY was around 220k regardless if you looked at the months individually or the year as a whole.  The BLS seasonal adjustment was about 10% high on average and 20%+ high in many months.  While in 2015 suddenly the seasonal adjusted reading is on average about 20% low, and misses low by 30% in many months.

I'm not saying any of the new jobs created or good or high paying--the data clearly indicates otherwise--, just that it seems the quantity of jobs is understated by a wide margin this year.

The December crude oil contract ended Tuesday 10-27 around $43.39/bbl.  This weeks CFTC Commitment of Traders Report (COT) for 10-27 indicates that swap dealers and the largest traders continue to get more bullish, but managed money has gotten significantly less bullish as have physical traders and the non-reportable group.  I believe this reflects physical traders and producers selling above $47, the large amount of oil in storage--especially the recent surge in tanker storage-and most importantly US production that has leveled off.

Still total contract numbers show long positions increasing for every group even as short positions increase in some groups, negating any growth in long positions in those groups.

As you can see my indicator for swap dealers is now much more bullish than it was for the March and August surges.  Indeed we saw a Wednesday surge in price when the EIA report indicated that crude+refined inventory fell 3.5MM bbl.  This was mostly due to high demand and a drop in imports.  The biggest drop in imports was from Mexico.  Imports from Canada and Saudi Arabia were very strong.  I believe imports from both Canada and Saudi Arabia, especially Saudi Arabia, can move significantly higher.

All indicators show a strong cap remaining on crude oil prices, however pressure continues to build for a strong upside move.  My belief is last weeks bullish EIA inventory data is a one off event for now, this week should show a build.  Market reaction after the Genscape inventory reports this week has been generally bearish.

I am primarily short crude.  Short CLZ5, own DWTI and have sold both puts and calls on USO.  I'm medium and long term bullish on crude, but short term bearish.

Looking to rebuy UWTI in the $8.25-$9 range, buy CLZ5 in the $42.50-$43 range and sell puts on USO when it hits $13.75-$14.

Here is a plot of price as of Tuesday each week vs US Production+Unaccounted for Oil.  I add Unaccounted for Oil to US Production to get a little clearer picture, a little sooner of what US producers are doing since EIA estimates weekly production.

It gives me a little concern for the bullish case since production has risen slightly recently.  I was surprised that a move to $50 had such a dramatic and lasting affect on US production.  I was fully expecting a drop below 9MM bbl/day.  That said, given the strong floor that seems to be established in the $42.50-$44 range by large traders/professional money, it is my belief that producers likely hedged their September-November output knowing a seasonal downturn in price was likely triggered by the refinery switch over for winter.

So now the game is for US production to fall making room for Iranian oil which will hit early 2016.  It is also by belief that the Saudis would like to pick up another 500k-1MM bbl per day of exports to the US, so oil prices could be capped in the $50s for awhile until US producers make room for more OPEC oil.

A lot of hedges some placed last year in the $80s and $90s along with new ones this summer in the $60s will be rapidly expiring in coming weeks and months.  That is likely to have dramatic impact on US production.

The COT report from my previous report indicates managed money may be slightly ahead of the big move up in price, yet this divergence still indicates price is likely to make a significant upward surge.  If it becomes obvious US production is falling, that surge will materialize.

In other words this trader is very cautious about shorting.  Really I only like to short when price hits $50, but I have been known to take out shorts at lower prices.  But I am a big time buyer in the $43 area.  I've lowered by UWTI target from $10 to $9 and even was able to pick some up this week at $8.25.  I'm buying USO via put options around $14-$14.50.  But I sell on each surge.

What I was looking for was a leading indicator.  I found that in the swap dealers, the yellow line.  This means the indicator needs to move opposite of price and reach an extreme indicating a reversal in price is imminent.  Reversal of the indicator confirms the price move is real.

Some would ask why not just follow the managed money, the gray line.  That is because it slightly lags price and more or less follows price.  When the amount of bullish managed money flattened out at the top this past summer, it was an indicator that a reversal was possible as well,

Note the late August peak in Swap Dealer bullishness has been surpassed.  This was likely caused when price surged passed $50 and producers opened a lot of hedges.  Total short positions for both the Producer/Merchant/Processor and Swap Dealer groups surged.  This has also led to an increase in US Production+Unaccounted for Oil.  I'd like to see the total short positions for the physical and swap dealer groups drop back to the levels that were seen on the March and August spikes.  Until then I will be somewhat nervous that $42.50 could break, retesting $37.75.  But history indicates that if price stays in the low $40s and certainly if it moves below $40, we will see a massive falloff in US Production+Unaccounted.

Mid summer produced a 5% production+Unaccounted drop as producers hedges expired.  It is my belief we will see a repeat as the latest block of producer hedges expire.