China's statistics are notoriously unreliable, and oil demand is no exception.

In its Oil Market Report for February, the International Energy Agency admits that measuring it is more of an art than a science, and has announced a new methodology. While this may sound a little on the wonkish side, there is an important point: China counts for around 10 per cent of the world's oil consumption and 40 per cent of global oil demand growth. Getting this bit right (or less wrong) is crucial.

So what is the deal with measuring China's oil demand?

The IEA says it used to calculate "apparent demand" (a proxy for consumption) as the sum of Chinese refinery output and net product imports. A straightforward measure, the problem was it failed to account for changes in product inventories. Fine as long as inventory swings were small – but that's no longer the case. As the IEA says, "a step change in Chinese refining capacity has dramatically increased the potential for product swings".

From the report:

China was short refining capacity and had to rely in part on product imports to meet domestic demand… That has now changed. In the last few months, Chinese companies have aggressively expanded their refining capacity, turning China into a budding product exporter – a trend that is likely to continue. Chinese refining capacity increased by 300 kb/d in 4Q12 and is forecast to expand by another 300 kb/d in 2013. Data for the last few months show that as Chinese refining capacity ran ahead of domestic demand growth, there was more room for product stock builds.

In other words: measuring oil in and out of the country doesn't tell you how much it is actually using.

So now the IEA captures reported stock changes in its demand assessments. As a result, "This new method smoothes out abnormal swings in apparent demand: thus, a slowdown in apparent demand in the first 10 months of 2012 no longer seems as pronounced and as divorced from Chinese economic activity as it did according to the IEA's previous methodology. Likewise, the 4Q12 ramp-up in demand seems less startling when adjusted for inventory builds." (See chart)

However, this is still far from perfect, as the IEA admits. Do the data include just state-owned refineries, or independent ones too? What about seaborne stockpiles? LPG, naphtha and fuel oil are missed out of the official Chinese data, which the IEA has to use as a base.

The IEA ends on a upbeat, yet humble note.

Oil statistics, like science according to Werner Heisenberg, proceeds from error to error, not from truth to truth. China itself is making great strides in improving its data collection and reporting system. This adjustment in our methodology is far from being our final word on Chinese demand.

For a country where electricity demand is often seen as a better measure of economic progress than GDP, improvement by error to error seems a good way of putting it.