Today’s report (week ended 2 Aug): EIA net change: +55 bcf, EA: +62 bcf
- This week, the EIA report indicated less of a build than the market had expected, in contrast to the report for the week ended 26 July in which storage built by more than market expectations. In our models, this week’s miss to the downside fully reversed last week’s miss to the upside. In the process, this week’s miss may have also shown that some baseline revisions to our weekly supply-demand model following last week’s miss may need to be reversed as well. The EIA print in our view is still difficult to justify given what flow data indicate is a pronounced increased in production (+1.3 bcf/d w/w).
Next Thursday’s report (week ending 9 Aug): EA preliminary: +60 bcf
- We peg w/w production at 0.3 bcf/d higher. Gas-fired power demand, boosted by hot weather and low gas prices, will be up by 1.3 bcf/d w/w. That increase in gas burn will not be enough to compensate for the 1.6 bcf/d w/w decline in feedgas demand.
Swim or sink
Last week, we wrote about how longstanding bearish fundamentals had finally managed to weigh on Sep-19 and Oct-19 contracts (see Panorama: Summer fire sale, 1 August 2019). Since then, cash has fallen a few dimes as LNG feedgas volumes plunged. There is maintenance on trains 3 and 4 at Sabine Pass and reduced flows to Corpus Christi, which could be train 2 being offlined temporarily as part of its commissioning process. Despite that loss of feedgas demand, our end-August projected carryout, which last week stood at 2.99 tcf, is moderately lower at 2.94 tcf.
Our balances last Thursday had assumed that maintenance at the Gillis compressor station would cut back flows into Sabine Pass, so some loss in LNG feedgas demand had already been baked into our forecast. What has helped to moderate our carryout even with feedgas disruptions is a more than 2.3 bcf/d increase in our modelled August power burn w/w owing to a shift in the 15-day forecast and lower gas prices. These factors helped PJM gas-fired generation on 5 August reach its second highest all-time daily average, likely due to coal-to-gas switching hitting the western part of the grid, where coal-fired plant dispatch costs are lower. There are several components to having maintained steady gas burn y/y in the summer to date, and risks to continued robust consumption. Power demand was at least 15 GW higher y/y in July, increasing the call on thermal generation overall. With the combination of coal retirements and lower prices—gas was nearly $0.50/mmbtu lower y/y—gas burn intensity increased.
A portion of that displacement of coal-fired generation was offset by rising renewables output, including wind and solar nationwide, and hydro in the West. We had expected burn to decrease in August under normal weather and prices closer to $2.30/mmbtu. But with forecasts for hotter-than-normal weather, especially in the South Central region, and even lower prices, power consumption is now likely to be relatively flat y/y this month, more than offsetting the loss in LNG feedgas demand. The Texas power market will continue to pull on more gas, after reaching an all-time August power demand peak on 7 August and ISO forecasts for a record high on 13 August.
With cash prices plumbing as low as they are currently, we ran a scenario for gas prices at $2.00/mmbtu in September and October, versus our reference case forecast of prices at $2.20/mmbtu. We would expect an incremental 1.1 bcf/d of gas power demand in September and an incremental 0.8 bcf/d in October if cash prices hit $2.00/mmbtu rather than $2.20/mmbtu.
For end-October, our reference case outlook is still calling for a carryout north of 3.7 tcf, which is in line with our long-standing view (see Fig 1). While Cheniere’s maintenance will lower our estimate for feedgas demand in August, we had carried an assumption of maintenance at one-generic US train for the full month of October (notably not due to the tighter spreads between the US Gulf and Europe), so the loss of feedgas demand in August is not notably additive to our end-of-season forecast. We assume the Cheniere trains currently offline will be out of action for three weeks.
Offtakers at Cheniere’s trains have until 20 August to decide whether or not to load cargoes for October. While we think there still is significant downside risk to global LNG prices, especially with European storage by our estimates filling by mid-September (see North America Monthly: Global LNG trends: European storage weighs, 6 August 2019), the arb to Europe for October-loaded cargoes is still open by a wide margin, with the TTF Oct-19 trading near $4.50/mmbtu as of midday New York time on Thursday (8 August). If the M+3 contract is considered, as October would be the month for loading and November could be the month of delivery, that margin is even wider with the TTF Nov-19 trading at $5.50/mmbtu. That margin would suggest October loadings would not be cancelled by the 20 August deadline.
|Fig 1: End-October storage projection, tcf||Fig 2: Weekly EIA storage change, bcf|
|Source: EIA, Energy Aspects||Source: EIA, Energy Aspects|