Exclusive News & Views on Energy Trading, Risk & The Policy That Drives All of It
By scudderpubl1189766, Oct 22, 2015 11:45 AM
Natural Gas Storage Tealeaves
Your Comprehensive Analysis for
EIA Gas Storage Report
Last week's 100 Bcf build caught most of by surprise, but not everybody. Reza Haidari of ThomsonReuters is one of the few analysts we poll each week who didn't take it on the chin; Haidari was high last week at 97 Bcf - one of six forecasts we collected (out of 41 submitted) that came within 5 Bcf of the tape. We asked him for some wisdom on last week's report and here's what he said: "The shoulder season as most analysts know arrives with some vagaries which impact on the S/D balance. One of these is a slow but progressive tightening of the "visible" balance which we have monitored happening during fall seasons. Of course, regarding today's (last Thursday) EIA storage number (+100) the market consensus was much lower than the actual number so that factor can be excluded in explaining the miss. Another fundamental shift, of course, is the diminishing CDDs, and increasing HDDs occurring simultaneously. Across a continental geography with diverse changes the switchover often wreaks havoc on analysts' models. The week-on-week change most recently, for example, showed that our GW-CDDs fell from 47 to 23 while GW-HDDs jumped from 13 to 31. Our interpretation of those weather changes was that even though heating demand increased by almost 2 Bcf/d, gas-to-power demand dropped by more: 3.2 Bcf/d. It was reported by some that total GW-DDs increased to the highest level, but this misses the point above. That may explain some of the market over-estimating the heating demand impact this early in the fall season, while under-estimating the drop off in cooling demand. The drop off in production flows also could have had a marginal impact in analysts reducing their injection estimates. On the pure storage flow side of the equation, the pipeline sample was similar to the +95 Bcf week and our model actually estimated +101, so no surprise there." In case you're interested, Haidari is at 84 Bcf this week.
Analyst Callie Kolbe of PointLogic (at 98 last week) offered the following on why her models trended toward the high side of the market for last week's 100 Bcf report, when the rest of us schleps were around 91-93 Bcf: "In comparing week-on-week supply and demand fundamentals, storage builds for the week were projected to come in at least higher than the prior storage week. Despite a week-on-week 0.45 Bcf/d fall in supply, there was just over a 0.5 Bcf/d decline in total consumption led by a 2.3 Bcf/d decline in power demand. I think most interesting is that even though residential-commercial demand was estimated to grow by over 2 Bcf/d during the week, the relationship between outright heating demand and lower temperatures with storage activity typically seen during the Winter has not yet been realized and likely led the market this week to underestimate storage builds. As such, sample field activity served as the key indicator for the week's report. Fields in PointLogic's sample of the East region showed weaker week-on-week injection activity driven by Columbia Gas and Dominion, however, this decline was more than offset by significantly stronger activity in the West and producing regions where sample activity was estimated to have increased by over 54 percent and 17 percent respectively," Kolbe tells us.
Jeff Moore of Bentek (low at 88 last week), like us, didn't see this one coming. "Obviously." He says there's something to be said about the "lag" effect of heating demand. That is, people don't immediately turn on their furnace the day it gets cold. "But I personally thought the 95 from the previous week was some indication that residential-commercial demand was starting to pick up, so it's tough to hang your hat on that argument." He noted that for last week, TCO and Dominion were apparently false signals. "Even if they had each injected just 1 Bcf more, I think I could have rationalized a 51 in the East (EIA reported 51 in the East, while Bentek reported 45 in the East). The fact that they basically cut their injections in half compared to recent weeks was what drove my estimate down. It's interesting because our daily storage report, which is essentially an "early view" estimate of the number and doesn't include TCO and Dominion, was at 101 on the week. Personally I thought it had a strong model bias to the high side so I kind of discounted it. I guess I was wrong ..."
So, what up for this week's report? It's a mixed bag. Our early view report from last Friday came in at 86.8 and the median was 85. The range was 80 to 95 Bcf. This week our EMD Consensus came in at 86 while the Survey Index (average of the top 6 surveys) was a little higher at 87.8. The range this week is 80 to 96 Bcf. The street seemed comfortable with this upper 80's neighborhood and even a bit higher. While the survey average was 87.8, our group of 7 bank analysis was lower at 86.7; and lower still was our group of independent analysts and models - this category was lowest at 85.7 Bcf. The spread between the three categories we track was 2.3, well under the 3 Bcf level pointing to a surprise. Our GWDD Model came in high at 88 Bcf. At the other end of the spectrum, we have the EMD Editor as this week's LowBaller at 80 Bcf. If 85-87 Bcf is viewed as the norm for this week, a 5 Bcf true up (true-down actually) from last week may not be out of the question. 80-84 Bcf? Entirely possible. While last week storage activity fell in the Producing and West regions, Bentek and others said there were some strong injections at several of the larger facilities in the East Region. Hmm. Enough to torque it up to 87-88? We think not. Last year, same week, EIA came in at 94 Bcf and market expectations called for an average build of 98 Bcf. Lower than expected. Just saying. -the editor
Pointlogic Energy PointLogic Energy is forecasting a +86 Bcf build. "This lower week on week number reflects a somewhat significant increase in power burn of roughly 2 Bcf/d which drove Total U.S. Demand from Electric Generation above 25.5 Bcf/d on average for the seven day period. This was partially offset with increased imports from Canada which topped 5.8 Bcf/d and the balance of the new demand was met with lower storage injections in the otherwise quiet most recent flow week," PLE says. "Regionally we see a huge decline in West injections coming in 9 Bcf lower week on week at a +2." In the East, PLE is expecting a +54 build while in the producing region injections declined to only +30 Bcf this week. The Kilduff Report says that the weather outlook has turned colder two weeks out, which is helping natural gas prices rebound off of the lows from last week's inventory report. The 100 Bcf net injection caught many by surprise, and generated some terrific selling. The inventory report tomorrow should show a net injection of 96 Bcf, which will represent another blow. That will compare favorably with last year's 94 Bcf injection and the five year average of 86 Bcf. The final chips are being laid for the season ending storage levels, and it is pretty clear that a new record will be established. We are forecasting that the final tally will be 4,011 Bcf. The supply overhang will more aptly resemble an epic hangover for some. Early season cold weather of a serious degree will be needed to remedy the low price environment. The longer-term prospects for cold winter weather will likely be limited by the El Nino effect. So, the bears will continue to reign for some time, with new lows awaiting this market.
Analyst Andy Weissman, (at 84 Bcf) in his new Energy Flash Report publication (sign up for a free trial at http://ebwanalytics.com/) says that the November natural gas contract attempted to break through resistance at $2.498 for more than an hour Tuesday, but ultimately failed. "Further, traders no doubt recall last week's much-higher-than expected injection and may be reluctant to take long positions until after tomorrow's storage report. Absent a bullish forecast shift, therefore, the November contract is likely to lose ground in today's trading. Expected warm weather after the cold shot expected in the next few days fades may also tend to push prices lower," he says. "A bullish storage report may be needed tomorrow for the November contact to mount another challenge at resistance just below $2.50," Weissman adds. In related analysis, he says that since the beginning of the year, power sector consumption of natural gas has consistently exceeded year-ago levels by a wide margin, with an average increase of 4 Bcf/day (28 Bcf/week) during the first nine month of the year. This increase has been critical in keeping storage on a trajectory below 4,000 Bcf, he says. Since late September, the year-over-year increase has been near its highest level of the year, averaging 5 Bcf/day in three of the past four weeks. With cash market prices near record lows, nearly two-thirds of this increase has been due to coal displacement. Most of the remainder, however, has been due to an exceptionally high number of nuclear refueling outages. Nuclear outages will start to tail off rapidly next week. However, power sector demand is still likely to be more than 3 Bcf/day above last year.
Bentek (at 82 Bcf) says that storage activity fell within the Producing and West Regions, with injections falling significantly within the West as power burn demand increased to monthly highs. Strong injections from several of the larger facilities within the East Region give some high-side risk to this week's forecast, especially considering the large injection reported from the region last week. However, during the same week last year, Bentek' sample injections increased within the East compared to the previous week, while the EIA announced a smaller injection week-over-week, meaning an increase in sample injections doesn't necessarily translate to a stronger injection announced from the EIA.
Tim Evans of CITI says that the natural gas market is on the defensive, with the 11-15 day temperature forecasts swinging back in a more bearish direction, with warmer than normal readings limiting heating demand through the beginning of November. Consensus expectations for Thursday's DOE storage report may be slipping a bit from the 93-bcf level to something nearer the 86-bcf five-year average for the date, but that hasn't been enough to keep a floor under prices, Evans says.
The research team at Raymond James are expecting a build of 91 Bcf this week. "The market was 1.3 Bcf/d tighter last week and has averaged 1.3 Bcf/d tighter over the last four weeks. Based on our assumptions, we expect the market to run 2.0 Bcf/d tighter on a y/y basis for the week. Thus, our official forecast for this Thursday is for an injection of 91 Bcf. If our forecast is right, the year-over-year storage surplus of 394 Bcf will increase by 8 to a surplus of 402 Bcf."
UBS is forecasting an 85-95 Bcf injection for the week. "We expect the EIA to report an 85-95 Bcf injection for the week ended October 16th, compared to 2014's 94 Bcf injection and the 5-year average of an 87 Bcf injection. We estimate inventories increased to 3,823 Bcf, modestly narrowing the surplus vs. last year to 430 Bcf and expanding the surplus vs. 5-year average to 148 Bcf." The bank notes that last week's weather was 23% warmer than last year and 15% warmer than the 5-year average. "Last week's injection implies that the weather-adjusted S/D balance loosened ~0.5 Bcf/d week-on-week. And while we estimate the weather-adjusted S/D has been 0.6 Bcfd undersupplied vs last year when the industry had a record high refill season, it has been 1.2 Bcfd oversupplied vs the 5-year average. Given the sharp decline in YTD prices, we believe an increase in coal-to-natural gas fuel switching of 3.8 Bcf/d YTD has tightened the weather-adjusted S/D balance from material oversupply in the winter to 1.2 Bcfd oversupplied that has inventories on a path to start next winter at 3.98 Tcf (~180 Bcf above normal)."
Ben Smith of First Enercast Financial says that the week came in warmer than the week prior and warmer than both last year and average. "The increased CDD totals increased generation use by an estimated 13 Bcf over the previous week. The loss of HDD cut residential/commercial heating use by 2 Bcf. In addition, exports to Mexico increased by an estimated 1.5 Bcf over the previous week in response to the warmer changes and supply was up slightly on higher imports from Canada, which more than offset declining U.S. output, Smith says. "U.S. dry gas production has remained in a very tight range since June. Current maintenance on several lines and compressor stations has kept supply from increasing despite the recent increase in transport capacity added in late summer," he adds.
Charlie Fenner of Macquarie Energy says his scrape model came in at 84 Bcf and his econometric model was higher at 87 Bcf injection. His final estimate is 86 Bcf. His calcs showed nationwide sensitivity-weighted CDDs totaled 36 for the storage week, up 13 week-on-week. He says that fuel switching has recovered somewhat in October, holding better on weekends and that "production may be rocky until at least the end of the month, due to plant and pipeline maintenance." He says that Canadian net imports totaled 5.3 Bcf/d, up 0.4 Bcf/d week/week; LNG sendout totaled 0.1 Bcf/d, flat week/week; and Mexican exports totaled 3.0 Bcf/d, up 0.2 Bcf/d week/week. Fenner is still looking at 3.95 Tcf for the end of season tally.
Forecasts Courtesy of Commodity Weather Group
Today's Changes: Another day of mixed to minor changes overall. Some short-term warmer adjustments are noted in the Midwest, some cooler early shifts are seen in the West with warmer changes there toward the 6-10 day. The Midwest to East is slightly cooler toward the day's 8-12 window, but then a warmer shift stays pretty much on schedule for middle to late 11-15 day. Overnight model consensus sees some demand loss: The American and European ensembles both went warmer overnight for the 6-15 day period, while the Canadian ensemble went cooler to colder. That Canadian ensemble is actually the best match to the upper level pattern analogs on all models as they still show a Yukon high pressure ridge spike that should generate at least a transient cool to cold push for the Midwest, South, and East. The target window looks to be days 8-12 and we keep the cool risks higher. Otherwise, a warmer pattern looks to return for days 13+.
For a free trial, go to http://www.commoditywx.com/
Key Market Analysis
Forecasting End of Winter Withdrawal
By Jack Weixel, Pointlogic Energy
The following bit of analysis comes from the good folks at PointLogic Energy (PLE). The group publishes occasional market reports called, "Get the Point." These reports are free. Sign up on PLEs company website.
Go to http://www.pointlogicenergy.com/market-news.html. Due to space limitations in Tealeaves, this is a slightly edited version. The complete report with additional charts can be downloaded on the PLE website--the editor
Summer injection season is rapidly coming to a close. It's clear that storage inventories will enter the winter heating season at record levels, but the severity of how winter plays out will be one of the biggest drivers influencing summer 2016. This week's Get the Point will focus on winter withdrawal season and end of March 2016 inventory levels.
PointLogic Energy estimates that for the remainder of summer, if supply and demand keep pace at current levels, the U.S. could inject 2,613 billion cubic feet (Bcf) of natural gas, making this summer the second most prolific injection season on record. This level of injection is a mere 120 Bcf below the record setting injection level of 2,733 Bcf seen in summer 2014.
What does this mean for inventories at the end of October? The below chart plots out both supply and demand forecasted for summer 2015 versus actual data for summer 2014 (all data in billion cubic feet per day or Bcf/d).
Inventory levels at the end of October will most certainly break the record inventory levels of 3,929 Bcf seen at the end of 2012 injection season. In fact, if the 2,613 Bcf scenario plays out, the U.S. could be looking at 4,089 Bcf in storage once colder weather finally takes hold. It's also highly likely that final injection weeks could last into the first two or three weeks of November, as the level of dry gas production available to the market has swelled above 73.0 Bcf/d over the past several weeks.
Using this storage inventory level as a base, let's take a look at three scenarios for Winter 2015/2016 to examine how demand, weather, price and production levels could impact end of March carryout inventory levels. For the natural gas markets winter is defined as November through March and summer is defined as April through October.
Scenario #1: Using EIA's STEO Predictions
Each month, the Energy Information Administration (EIA) updates its Short Term Energy Outlook (STEO), which provides updated "actual" natural gas fundamentals data from its Natural Gas Monthly report and an outlook of these fundamentals through the next calendar year.
For winter months, the most significant fundamental demand factor to forecast is residential and commercial demand (res/comm). Abnormally cold or warm weather can shift the amount of gas used to heat homes and businesses by as much as 20 percent in any given winter.
For example, during the polar vortex winter of 2013/14, January res/comm demand averaged 59.9 Bcf/d compared to the more normal winter of 2012/13 where January res/comm demand averaged only 50.1 Bcf/d. This is a difference of 9.8 Bcf/d over an entire month, which equates to 304 Bcf of natural gas that would need to come from storage withdrawals.
In its most recent STEO, the EIA is expecting a decline in res/comm demand this winter compared to last winter of 2.5 Bcf/d due to warmer anticipated weather. This equates to 378 Bcf less gas withdrawn from storage on its own, so what is moving the weather dial so abruptly? The most talked about weather event since the polar vortex winter of 2013/14 is the anticipation of a strong El Nino weather pattern this winter.
El Nino is a weather condition that is realized when sea surfaces temperatures in the Pacific Ocean exceed normal levels. In its latest update, the National Oceanic and Atmospheric Administration (NOAA) is anticipating a significant El Nino event this winter which will have sweeping impacts on temperature and precipitation levels across North America. The consensus from weather forecasters and NOAA is that El Nino has not weakened going into the shoulder season and could remain as prolific as recent El Ninos experienced in 1997 and 1992.
NOAA's Climate Prediction Center anticipates that the forthcoming El Nino event will bring warm weather to the Pacific Northwest, Midcontinent market area and the Northeast. Colder weather will be prevalent in southern regions of the U.S. extending from western New Mexico eastwardly through Texas and the greater Southeast.
Because the major consumers of natural gas as a home heating fuel reside in the red, orange and yellow portions of the above map, lower overall demand from this region will be greater than any incremental demand from the regions shown in blue. The EIA also expects that overall weak demand will keep a lid on prices, which will encourage increased use of natural gas from the power and industrial sectors. Power, Industrial and Res/Comm demand are all derived from the EIA's Short Term Energy Outlook which assumes a warmer than normal winter. In addition to the three traditional sources of demand, exports to Mexico and liquefied natural gas s Sabine Pass, located in Cameron Parish, Louisiana.
All told, under this scenario total demand will average 0.2 Bcf/d higher than Winter 2014/2015 despite weakened res/comm demand precipitated by El Nino.
Scenario #2: Winter 2014/2015 Demand Side Repeat
According to weather provider WSI, the lower 48 experienced 2,707 gas-weighted heating degree days (GWHDD) during the peak winter months of December, January and February 2014/15. Regionally, last winter in the Northeast was very similar to the polar vortex winter of 2013/14 (as New Englanders can attest battling record snow and cold throughout the period). While WSI projects 2,540 December through February GWHDDs this winter, a panel of forecasters at Scudder Publishing's recent Winter Tealeavesconference in Houston leave a 30-35 percent chance of positive Artic Oscillation (AO) numbers. A positive AO is often conducive to polar vortex occurrences, particularly in the Midwest and Northeast regions of the country. Though slim, a sudden change in the weather pattern over the densely populated Northeast could shift natural gas prices upwards, leaving power burn demand slightly neutered as the cost of marginal gas-fired power plant increases. In this regard, a repeat of winter 2014/15 weather related demand would look like the chart below.
In this scenario, Mexican exports would look the same as scenario one, because much of this gas is sourced out of Texas and the Southwest. PointLogic is assuming a slower ramp up in LNG exports due to a weather influenced increase in price at Henry Hub and the fact that Cheniere recently announced that commercial deliveries of LNG would not commence until April 2016.
While it is believed that commissioning cargoes could roll off of Train 1 by late December, the average amount delivered over the course of winter 2015/16 would be slightly lower than EIA's projection of 0.5 Bcf/d. All told, eliminating the decrease in res/comm demand in this scenario would leave total demand at a 1.2 Bcf/d premium to winter 2014/15.
Scenario #3: PointLogic Expectations
PointLogic Energy's view of 2015/16 winter demand comes in somewhere in the middle of scenarios one and two. With a 65-70 percent chance of a warmer winter than 2014/15, res/comm demand is down 2.0 Bcf/d winter-on-winter, while power and industrial demand continue to climb higher due to depressed natural gas prices. Mexican exports and LNG retain their gains resulting in a total demand increase of 0.7 Bcf/d winter-on-winter.
Regarding power demand specifically, this September as temperatures waned into neutral territory, PointLogic's nomination data and estimated total deliveries did not ratchet down as expected. Total power demand remained at an average of nearly 28.0 Bcf/d, or 2.2 Bcf/d higher than September 2014 levels. The first two weeks of October have also seen elevated power burn levels averaging 1.0 Bcf/d above October 2014.
On the supply side of the equation, PointLogic's production forecast predicts a 1.8 Bcf/d winter-on-winter gain due to expanded infrastructure options for producers in West Virginia and Ohio portions of the Marcellus and Utica, along with smaller but steady dry production increases in the Northern Rockies and Permian basins out west.
To visualize the impact on storage withdrawals over the course of the winter, the below chart plots our base level storage assumption of over 4,000 Bcf of natural gas available at the end of injection season 2015 and plots the corresponding scenario inventory levels.
Using EIA's demand projections from the STEO nets a storage inventory level of 2,100 Bcf, while the lower bound is set at 1, 970 Bcf with a repeat of the winter 2014/15 winter. The PointLogic estimate comes in squarely in the middle at 2,040 Tcf. All three of these scenarios fall within the upper third portion of the range of actuals seen over the past five years, and regrettably for producers, all signal a continuation of depressed natural gas prices going into the summer of 2016.
Lower Production Impact on Storage
So what if production levels do not pan out the way PointLogic envisions? While we are confident in our forecast, activity in specific basins comes down to individual producer decisions, who have been battered by low prices, capital expenditure reductions and falling rig counts all summer long. While evidence that rigs left drilling in the field are the most efficient and best of the fleet in terms of productivity and costs, another few months of depressed prices could lead producers to reduce the pace of new drilling even further until higher prices materialize. Publications like EIA's Drilling Productivity Report have alluded to this scenario and our own analysis of incremental flows on pipeline expansion projects have seen lower fill rates than prior rushes to fill capacity. However, a large inventory of drilled but uncompleted wells may keep production levels healthy if and when producers decide to bring that production to market. Lowering the production forecast to a more pedestrian 0.8 Bcf/d increase in winter 2015/16 compared to winter 2014/15 takes an incremental 150 Bcf of storage gas out of caverns over the course of November through March and results in the below storage inventory levels detailed below.
The range of end of March 2016 storage inventory estimates remains fairly tight in all of the scenarios laid out above. In an attempt to be predictive, but pragmatic at the same time, PointLogic estimates that unless prevailing weather patterns change significantly and production slows down considerably, storage inventories at the end of March 2016 will remain near 2,000 Bcf. This level of inventory should set prices up nicely for consumers in the summer of 2016 while continuing the woes experienced by producers during the summer of 2015. Stay tuned as we update our forecast based on the facts on the ground and plot out the trajectory and final landing spot of storage inventories throughout the winter season.
Each week we poll up to 42 professional storage forecasts for our weekly Natural Gas Storage Boxscores (as seen in each bi-weekly issue of Energy Metro Desk*). This is North America's biggest and most comprehensive natural gas storage survey and report.