Today’s report (week ended 8 Mar): EIA: net change: -204 bcf, EA: -209 bcf
- Our flow model, at -206 bcf, outperformed our supply-demand model, which pointed to a -209 bcf withdrawal. To align with today’s report, we adjusted res-com demand lower.
Next Thursday’s report (week ending 15 Mar): EA preliminary: -50 bcf
- Next week’s report will show the second-to-last withdrawal of the heating season. Production will remain flat w/w, according to our flow data, with a 0.8 bcf/d step down in Canadian trade. A 16 bcf/d w/w decline in res-com demand and a 5 bcf/d drop in power burn will drive the chop in the weekly withdrawal rate.
Shoulder season is near
A change in weather forecasts to a much milder-than-normal end to March has shifted our end-of-winter-season inventory estimate to 1.08 tcf (as compared to the ICE swap at 1.12 tcf as of 13 March), versus a low projection of 950 bcf last Thursday. From here on, the storage withdrawal rate falls off of a proverbial cliff and our first forecast injection is for the week ending 29 March, a significant departure from last year, when withdrawal activity extended through the third week of April. Still, the injection season strip has remained supported despite that upward adjustment to inventory estimates and the projected abrupt end to the withdrawal season nearly a week earlier than the traditional end of season.
With the shoulder season approaching fast and a repeat of the supportive cold in April 2018 and early heat in May 2018 unlikely this year, the focus is squarely on the injection season. How tight or loose fundamentals are in the shoulder season will set the tone for near-term trading. If storage is at 1.08 tcf at end-March, there is little need for burning off excess inventories early in the injection season and consequently no major risk that Apr-19 cash decouples from the strip. Instead, the movements in balances and prices in the shoulder season should stem from a trade-off between supply recovery (and growth) and structural demand given the smaller role weather will play in this period.
Our expectations are for structural demand to carry on as it has recently. A rise in LNG feedgas demand in the shoulder season would require another train to request permission to introduce feedgas and/or for Cameron LNG T1 to actually begin to bring in feedgas consistently, which it has not done so far according to flow data. For Cameron LNG to meet a stated first export by Q2 19, feedgas intake needs to start immediately and be sustained.
Freeport LNG T1 has stated that it could begin to take feedgas in April or May, but those initial volumes are likely to be minimal and will not materially sway balances as first LNG exports are not expected until July at the earliest.
In last week’s Panorama, we discussed potential feedgas variability due to weather, especially given high stocks at Cheniere’s Sabine Pass (see Panorama: Foggy tank tops, 7 March 2019). With fog rolling in earlier this week, feedgas throttled back perceptibly once more, falling by 1.1 bcf/d on Monday, from the previous week’s peak. In addition, Cheniere mentioned in its earnings call that maintenance and turnaounds are necessary in order to pursue some ‘debottlenecking initiatives,’ which could reduce feedgas deliveries, although the timing is unclear.
According to EIA data, industrial gas demand growth appears to be softening. Our balances break out industrial heating demand from baseload demand in an effort to weather-normalise activity in the sector. According to our models, the most recent set of complete data from the EIA (December 2018) appears to have softened toward 2% y/y growth, after averaging near 4% from May-September 2018. Isolating just Texas and Louisiana, given their high concentration of petrochemical plants and other gas-intensive manufacturing, the rates of growth have started to fall. Growth in Texas averaged close to 0.3 bcf/d y/y in injection season 2018 and has fallen toward 0.1 bcf/d. Meanwhile, growth in Louisiana has fallen to be flat y/y, after growing by 0.1 bcf/d y/y in injection season 2018.
We anticipate slower demand growth in the industrial sector will continue in Q1 19, with our weekly supply-demand models corroborating the time-lagged EIA data. While we have flagged the potential for weaker growth emanating from the petrochemicals sector, steel plant capacity utilisation, as reported by the American Iron and Steel Institute, is at a decadal high. A ramp-up in ethane crackers from Shintech, Westlake and DowDuPont could still support growth.
Mexican shoulder season demand may not move from what is currently estimated to be crossing the border—4.9-5.0 bcf/d. CFE has tendered for cargoes at Altamira through the third week of April. Such a timeline hints at a mid-to-late Q2 19 start-up of Sur de Texas-Tuxpan at the earliest. In addition, protests along Fermaca’s pipeline routes in Durango state forewarn of potential delays to the Wahalajara pipeline system starting up on time for its stated May in-service date. In summary, there is a strong possibility of no major sequential step-up in pipeline flows to Mexico this shoulder season.
For Canada, the shoulder season is bringing with it NGTL maintenance and production declines coming from the prevailing low-price environment in the WCSB (see E-mail alert: Earnings point to small 2019 Canadian gas output declines, but less capitalised E&Ps may deepen losses, 14 March 2019). NGTL has already scheduled a 0.5 bcf/d restriction in capacity from Western Alberta Mainline Loop for work between 4-11 April.
US production is a major wildcard for shoulder season balances. Production in Q1 19 has been marred by operational issues and freeze-offs. Some 65 bcf in Appalachia alone was lost due to freeze-offs. How quick a recovery from freeze-offs occurs, and the degree of sequential growth beyond the recovery, will be key to how tight the market will be. In Appalachia, winter production growth sometimes slows due to competition for local pipe capacity. Rockies output has been depressed due to lingering weather-related issues and another storm in the region this week could keep volumes offline until conditions warm for the season. As we have noted (see Insight: Q4 18 results—A well-hedged year, 7 March 2019), US producers remained well-hedged and private producer activity, particularly in Appalachia, is robust.
In short, price formation will depend on how hefty injection rates are at the start of the shoulder season. Currently, our weekly balances point to a 6.5 bcf/d average monthly injection rate in April.
|Fig 1: Projected EIA weekly storage change, bcf||Fig 2: Steel plant utilisation, %|
|Source: EIA, Energy Aspects||Source: Bloomberg, Energy Aspects|