Forecasting Chinese LNG demand: Why a monthly update is superior to a static yearly estimate, in an increasingly volatile market

Author: Alex Siow


ICIS forecasts China to import between 4.7 million to 5.3 million tonnes of LNG per month during February to April 2021. This is on average a 12% year-on-year (yoy) increase compared to the same months last year. On an annual basis, we are forecasting China to import 76m tonnes this year, an 11% yoy increase.

But this is only a valid prediction for now, within these few weeks. By early next month, our forecast number could have been updated and changed, depending on events that might have happened between now and early next month. Or it could be maintained, also depending on events.

This highlights the dilemma in short-term forecasting – LNG is an ever-changing market, with many uncertainties. Yet, this is what the LNG market presents now, an increasingly short-term and volatile global market, with spot traders reacting to the latest events, whereas in earlier years it was a much steadier market dominated by predictable long-term flows.

Aside from covid, which some can dismiss as a once-in-a-lifetime event, we are also dealing with more extreme or unexpected weather events such as hurricane disruptions to supply, a very cold winter and power supply interruption which drove our East Asia Index spot price (EAX) to above $30/MMBtu in January, and carbon neutral policies leading to increased gas consumption.

These events, and others, are increasingly driving the industry towards a shorter-term outlook. Here at ICIS we are trying to help shape this shorter-term view through our 24-month LNG demand forecast.

The perils of a static annual forecast

China imported 68.3m tonnes in 2020, gaining 11% from the year before. But this was not obvious to anyone during the first half of last year. On the demand side, nobody expected covid, or the extremely cold winter. Many new pipelines were also completed, opening new demand centres throughout the year. On the supply side, nobody predicted the government’s swift actions in curtailing pipe imports and gas production during the oversupplied summer.

A static annual forecast made during the start of the year – perhaps updated once in June – has proven to be inadequate in addressing the fast-changing, dynamic nature of the LNG market. If you look back to any annual forecast made available in the market, or the ones made by oil and gas companies to be used internally, many were expecting China to experience negative growth in 2020, some as recently as July 2020. Had you gone into the market with this outlook for China, it would have been a huge opportunity lost.

How has ICIS’ China LNG demand forecast changed throughout 2020?

The graph shows how our forecast value for China’s total annual imports in 2020 changed across the March to November monthly updates of our forecast.

We have consistently predicted year-on-year growth for China in 2020, except during the peak of covid in March and April, when we forecast an annual decline. Our growth forecast was maintained even when the entire global economy was in turmoil. The model is guided by our unique forecast methodology where we rely on top-down econometrics to guide the total gas demand, while at the same time incorporating bottom-up data and news from each province. This sets the floor and ceiling for all our estimates: domestic production, sectoral gas consumption etc.

Our June forecast was too high, where we were too quick to assume that curtailed domestic gas production would result in increased LNG imports. Actual gas consumption data and the latest GDP forecast figure were only made available a month later. Thus in July we re-adjusted our forecast to a lower level as major sectors were reporting lower than expected gas consumption. After July, a steady increase in our total annual forecast number can be seen, as Chinese gas use data – on both a national and regional level – was reported to be rising.

How will a month-by-month demand forecast help?

ICIS’ LNG demand forecast is an event-based, 24-month forecast, which typically gets updated once a month. Occasionally, it is also updated mid-month, should an event happen that warrants an immediate change to the forecast.

We have talked about how our predicted total import for 2020 has changed throughout the year, now let us talk about the forecast accuracy for the immediate three months ahead.

ICIS’ China LNG demand forecast dated 6-Mar, 7-Apr, 6-May etc versus Actuals 3-months ahead.

The table above shows how the ICIS forecast has fared in 2020, comparing the ICIS forecast number with the actuals for that same month and two months ahead. Our predictions were off by a large margin in April and May, around the peak of the pandemic. Yet after May our forecast margin of error was for the most part kept to within the single-digit percentage that we were aiming for. That translates to a range of 0.5 to 7 cargoes for the entire country’s import per month, a strong performance.

The ICIS demand forecast will provide, in advance, a good indication of China’s LNG needs, and if we dive down deeper into its terminal-level forecasts, we can understand LNG needs from a provincial level. For example, Tangshan Caofeidian, Qingdao, Shanghai Wuhaogou and Guangdong Dapeng all have different importing patterns, serving the needs of various neighbouring provinces.

Our import forecast includes both term and spot LNG. ICIS also tracks individual ship/cargo movements across the globe and has a database of companies’ contractual obligations.

Aside from China, the ICIS LNG demand forecast coverage includes the entire Asian market including Japan, South Korea, Taiwan, India, Pakistan, Bangladesh, Malaysia, Singapore, Thailand, Vietnam, Myanmar and Philippines. We will be extending our coverage into Europe to complete our global demand coverage by mid-2021.