Gas Balancing - Lack of Response !

This is the second of 4 articles focusing on the current Gas Balancing regime in the UK. The first two articles review recent trends and underlying causal factors at play, concluding with potential improvements to the regime. The third article will evaluate Projected Closing Linepack (PCLP) as an indicator of the Transmission System's Status, while the final article will pull together themes from the previous three articles and suggest a framework for hedging imbalance exposure.

The UK Balancing Regime and SO signals to the Shipper

The current UK Gas Balancing regime is designed to incentivize Shippers to balance their daily physical position. (The Gas-day is 05:00 to 05:00) Where Shippers, in aggregate, are out of balance then National Grid Gas, the System Operator (SO), undertakes residual balancing actions (trades) to bring the overall position back towards balance. Shippers who retain an end of day imbalance are cashed out at penal prices, which reflect the marginal price of residual balancing trades undertaken by the SO.

The Marginal System Buy Price (SMP Buy), applied to cash-out Shippers who end the day Short, is the highest price of any Buy trades undertaken by the SO. The System Marginal Sell price (SMP Sell), applied to cash-out Shippers who end the day long, is the lowest price of any Sell trades undertaken by the SO. When either marginal price is close to the System Average Price (SAP), then an adjustment is made to make sure each marginal price has a minimum differential from the SAP price. (The default differential) The default differential is also used to derive the SMP Buy, where the SO does not transact any Buy trades for the day in question, and the default differential is used to derive the SMP Sell price when the SO doesn't transact any Sell trades for that day.

During the May 2022 Gas Operational Forum, the SO suggested that it had been executing "a high volume of residual balancing trades" during 2022 and in addition there was "often a limited response by Shippers".

The SO is financially incentivized to minimize its interventions in the market, by signaling to Shippers to act to bring their own position back towards balance. The SO can signal to Shippers to balance their position in several ways.

The SO publishes hourly updates of Projected Closing Linepack (PCLP) along with the Opening Linepack level, signaling to Shippers the magnitude of the current aggregate forecast end of day imbalance. Shippers can compare the current forecast end of day imbalance, with historical data to access the likelihood of the SO taking further residual balancing trades, which might in turn lead to even more penal marginal cash-out prices being applied to any imbalance they may hold.

The SO can also signal for a Shipper response, by undertaking sufficient residual balancing trades to set the marginal price differentials above the default level. The indicative System Prices (SAP, SMP Sell and SMP Buy) are published in real time on the ICE platform. These indicative System Prices will reflect the residual balancing trades undertaken by the SO during the day, thus far. If the SO is trading and setting the indicative marginal price, then Shippers can observe this in real time on the ICE platform, by monitoring the within day marginal price differentials. This provides Shippers with an Ex-Ante price signal to balance their position.

Finally, the final marginal prices also provide an Ex-post signal, which in theory, should serve to influence future Shipper balancing behavior. (The historical Ex-post marginal price signals would also be factored in to any Shipper assessment of the possible impact of future SO actions, in response to a forecast aggregate end of day imbalance discussed above.)

Over the last 12-months the Ex-post Marginal Price differentials show the SO has provided a very strong signal to Shippers to balance their position, relative to previous years:

Source: National Grid MIPI; Authors own analysis

The large absolute marginal price differentials are partly a function of high volatile prompt prices, although even the relative marginal price differentials are high by historical standards, particularly the SMP Sell - SAP differential:

Source: National Grid MIPI; Authors own analysis

While the larger marginal price differentials suggest the SO has been sending a stronger price signal to Shippers to balance their end of day position, it is not clear to what degree Shippers are responding to this and other signals by the SO.

In order to evaluate this, the last 12-months (11/10/2021- 10/10/2022), will be compared with the previous 2-years (11/10/2019 - 10/10/2021) where marginal price differentials were typically set close to or at the default level.

This analysis will firstly review aggregate Shipper response to within day signals provided by the SO about the aggregate forecast end of day imbalance. Secondly it will compare the aggregate Shipper response on days where the SO also provides a marginal price signal, with the response when the SO does not provide a marginal price signal. Finally, a deep dive is undertaken for two days where the SO provided significant Ex-ante marginal price signals, one where Shippers were Short in aggregate and one where they were long, accessing the aggregate Shipper response in each case.

Position and response measures

To simplify the first part of the analysis, a single reference time period is chosen to evaluate the aggregate Shipper position and any subsequent Shipper response by the end of the Gas-day. The 22:00 time period is chosen as the reference time period as it is just prior to the last quarter of the gas day, when the SO undertakes most of its residual balancing trades.

With reference to the period of interest (11/10/2021-10/10/2022), the SO only traded on 27 days prior to 22:00, while subsequently setting the Ex-post marginal price differential above the default for those days. These days will be removed from any analysis about Shipper responses to a marginal price signal post 22:00.

The difference between the 22:00 PCLP and the Opening Linepack (OLP), provides a forecast aggregate end of day Shipper imbalance as of 22:00, the 22:00 aggregate forecast position. Where this metric is positive, then the aggregate end of day Shipper imbalance is forecast to be long, (Supply > Demand) and if the metric is negative the aggregate end of day Shipper imbalance is forecast to be short.

The subsequent response is evaluated in two ways: Firstly (PCLP 04:00 - OLP) - (PCLP 22:00 - OLP), which simplifies to PCLP 04:00 - PCLP 22:00. This represents the change in the forecast end of day aggregate Shipper imbalance from our reference point at 22:00 to the last published PCLP update towards the end of the gas-day.

Secondly the response is also measured by substituting the 04:00 PCLP with the actual Closing Linepack (CLP), CLP - PCLP 22:00.

Analysis

The following charts show the aggregate 22:00 position and subsequent aggregate Shipper for the last 12-months compared with the previous two years.

Source: National Grid MIPI; Authors own analysis

In both time periods, Shippers respond to a forecast aggregate imbalance at 22:00, by reducing their imbalance as the day draws towards a close. (This is indicated by the negative gradient in the above charts.) This is not surprising, as Shippers are reducing their exposure to the penal cash-out prices, and the PCLP forecast will tend to lag intended end of day Shipper positions, as this data is not provided by the Shippers themselves. (It is provided by the DN's and terminal operators.)

During the last 12-months, however, there has been a reduction in the aggregate level of Shipper response to a forecast imbalance at 22:00, compared with the previous 2-years. (The absolute gradients are lower in the bottom two charts compared with the top two charts.)

Also, any Shippers response has also been less certain during the last 12-months, compared with previous years. (This is indicated by the lower R-squared coefficients in the bottom two charts compared with the top two charts.)

Finally, it is interesting to note that both response measures in the previous 2 years, tended to be reasonably aligned. (Similar Gradients and r-squared values.) This has not been the case in the last 12-month.

From the SO's perspective, it would have been less certain what level of response it could expect from Shippers, to address a forecast aggregate imbalance during the last quarter of the gas day.

Secondly the analysis looked at the aggregate Shipper response, when the SO introduced a marginal price signal, compared with days when the SO did not provide an additional price signal over the last 12-months:

Source: MIPI, Authors own analysis; Note: The SO price signal data set, excludes any days where the SO traded prior to 22:00, while subsequently setting Ex-post marginal price differentials.

There is not conclusive evidence to suggest a change in aggregate Shipper response, when the SO introduces a marginal price signal during the last quarter of the gas day (22:00 onwards) over the last 12-months. For one response measure the gradient suggests a slightly larger response, but the other measure suggests a reduced response.

In conclusion:

The magnitude and certainty of Shipper responses to an aggregate 22:00 forecast end of day imbalance, has deteriorated over the last 12-months compared with previous years. This is likely to have been an underlying factor in the increased SO residual balancing actions during the last 12-months. (The SO took one or more residual balancing trades on 59% of days in the last quarter of the day, compared with 42% of days in the previous 2-years according to trade data published by the SO.)

This observation is consistent with the comments made by the SO in the May Gas Operational Forum about a lack of Shipper response in early 2022.

Perhaps more importantly, there does not seem to be any significant difference in the aggregate Shipper response on days where the SO sets a marginal price signal during the latter part of the gas day.

Deep dive on days which exhibit large Ex-post/Ex-Ante marginal price differentials:

30/08/2022

Source: National Grid MIPI; Authors own analysis

Note: The Ex-ante within day marginal price differentials have been estimated using hourly trade data published the SO, as these are not publicly available on the SO website after 23:00.

On the 30/08/2022 (a bank holiday weekend) PCLP moved above Opening Linepack levels in the early afternoon. The SO began to signal an Ex-ante SMP Sell price from around 20:00 hours onwards. Projected Closing Linepack levels did eventually seem to respond but the SO continued to Sell setting a marginal price differential of approximately ￡1.76/therm.

The final SAP price was ￡4.76 and the SMP marginal price approximately ￡3/therm. This was the then highest Ex-post marginal sell price differential for at least 10 years. (An even higher Ex-post marginal SMP Sell differential was subsequently set the following day around ￡2/therm)

While the PCLP data did show an aggregate Shipper response, after adjusting the Closing Linepack to include the SO sell trade volumes, (Adjusted CLP in the above chart) Shippers (in aggregate) still ended the day long by around 5mcm, thereby exposed to penal cash-out prices for this volume.

27/09/2022

On this day PCLP data indicated Shippers were short in aggregate over much of the day:

Source: National Grid MIPI; Authors own analysis

The PCLP forecast barely responded to the Ex-ante marginal price signals during the day and the final estimated aggregate Shipper short position (adjusting for SO trades) was approximately 7.5mcm short. The Ex-post SMP Buy differential, was approximately ￡1/therm, the highest recorded since the Beast from the East in 2018.

The deep dive analysis suggests that even when the SO sets very high within day Ex-ante Marginal price differentials, these are not proving sufficient to encourage Shippers to significantly reduce their exposure to penal cash-out prices. (After factoring in SO trades.)

The fact that Shippers are not responding to large within day price signals as expected, is deeply concerning. The current UK Gas Balancing Regime rests on the assumption that Shippers have a strong incentive to respond to price signals provided by the SO (Ex-ante and Ex-post) and reduce their exposure to penal cash-out prices. (Thereby supporting the SO in reducing the aggregate imbalance, and avoiding the need for the SO to take more drastic action.)

In today's market environment, it might make sense for risk adverse Shippers to retain a long position, to hedge against forecast error, but this does not explain the lack of response to the aggregate short position and large Ex-ante price signals on the 27th of September!

What can be done to improve the responses of Shippers?

Firstly, it is worth noting that not everyone in the market has access to the ICE platform, enabling them to observe the within day Ex-ante marginal price differentials. The SO could procure this data from ICE and make it available to everyone on its MIPI platform. (Alternatively, it could calculate and publish this data independently, as it is the SO's own trades which set the Ex-ante marginal prices!)

Secondly, it is interesting to note that the SO tends to transact most of its trades towards the end of the day. The SO does this partly because it is less confident of the PCLP forecasts as a predictor of end of day imbalance earlier in the Gas-day:

Source: National Grid MIPI; Authors own analysis

The SO also has two financial balancing incentives, one to limit the day-to-day absolute change in Linepack stocks below 2.8mcm if possible (5.6mcm in the shoulder months) and one related to its impact on Cash-out prices. The SO is financially incentivized to achieve a price range (Highest Trade Price - Lowest Trade price) below 1.5% of the actual SAP. If it achieves this (or lower) then the SO is renumerated up to a maximum pay out of ￡1,200 per day. However, the SO is penalized for a traded range (Highest Trade price - Lowest Trade price) above 1.5% of SAP. This can cost the SO up to a maximum pay out by the SO of ￡24,000 per day, where the range is greater than 76% of SAP)

If the prompt market is volatile (as it has been), realized SAP prices could drift significantly from the Ex-ante levels observed earlier in the day. Combine this with the lack of confidence the SO has about earlier PCLP forecasts as a predictor of the end of day aggregate Shipper imbalance and it's not surprising the SO executes most residual balancing trades in the latter quarter of the gas day. (Thereby reducing its likelihood of incurring a penalty under both balancing incentives.)

Encouraging the SO to signal to the market earlier in the day, by setting earlier Ex-ante marginal prices would be beneficial for two reasons. Firstly, there is increased market liquidity in the working hours, when many more Shippers are active and secondly there is increased within day flexibility to call upon. Both these factors are likely to increase Shipper responses to a marginal price signal by the SO.

Changes to the residual balancing price incentive could help achieve this outcome. Rather than basing the SO price incentive on (Highest Trade Price - Lowest Trade Price)/Ex-post SAP, the SO price incentive could be based on (Highest Trade Price- Lowest Trade Price)/Ex-ante SAP, where the Ex-ante SAP price was calculated closer to the period the SO actually trades. For example, the average Ex-ante SAP price could be calculated for each quarter of the Gas-day, and the trade range could be generated from the highest and lowest trade prices actioned by the SO during each quarter of the Gas-day. The highest of the four quarterly range percentages of SAP could be used to determine the final renumeration/payment by the SO each day.

Another alternative is to relax/suspend the SO marginal price incentive, allowing the SO to be more confident in taking earlier residual balancing actions. (Without incurring a subsequent penalty) A relaxation could be triggered if the actual prompt price volatility is above some predefined threshold level for example.

Finally, as I have argued in earlier blogs, it may be time for a more significant overhaul of the UK Gas Balancing Regime, perhaps introducing shorter balancing periods.

Something must be done; the SO must have the effective levers (without being penalized for using them) to invoke the necessary aggregate Shipper response it requires as early as possible during the gas day. This will mitigate the need for more drastic actions later in the day and reduce the likelihood of the SO invoking and emergency if supply margins become tight.