How Much Oil Will China Consume?

Steven Kopits has a new guest post at Econbrowser, and once again, it’s getting some attention. Last time, he argued that the EIA had embraced peak oil. (I rebutted it here.) This time, he’s arguing that the EIA has massively (and I really mean massively) underestimated future Chinese oil demand growth. That may or may not be true. But his argument again strikes me as being badly wrong.

The gist is as follows. He starts by observing that “the bulk of oil demand growth occurs in the two decades during which societies typically acquire motor veicles”. He then notes that “per capita consumption in both Korea and Japan peaked at 1.9 U.S. gallons per day. Korean levels are unchanged; Japanese consumption has declined to 1.4 U.S. gal per person in the last decade.” He concludes that “based on the experience of Korea and Japan, China’s current population would be expected to consume approximately 55 mbpd at steady state (when per capita consumption plateaus), or nearly 2/3 of current global oil production, were the supply available.” Kopits notes that the timing of this rise can vary depending on the situation. That leads him to produce this graph:

Yikes! And yet…

The prediction is wrong. Why? Neither country ever came close to consuming 1.9 U.S. gallons of fuel per person per day in motor vehicles.

In 1973, when Kopits says Japanese per capita oil consumption peaked, daily per capita use of oil for transportation was 0.3 gallons. (Most oil was instead consumed in industry; electricity generation was a big consumer too. All my numbers here are from the IEA.) Using the same assumptions about population as Kopits does, that yields a peak of 8.7 mb/d for Chinese oil demand for transport. China would still, of course, use oil in industry and electricity generation, which is why the EIA projections are higher than this. But since substitutes for oil are generally available in those sectors in China, demand should be held in check by prices – no one will gobble up scarce oil for electricity or industry if far more abundant coal is available. (That’s why the line in the graph for Korea stays almost flat for so long - it corresponds to a period of high oil prices.) The Japanese experience thus gives no reason to believe that Chinese oil consumption will go up to Kopits’s figure of 55 mb/d. (To be fair, Japanese per capita daily oil demand for transportation didn’t peak in 1973, contrary to what Kopits claims; it stands at about 0.5 gallons today. Still, this doesn’t change the bottom line.)

How about Korea? Korean daily per capita oil consumption for transportation was slightly less than 0.5 U.S. gallons in 1996, the peak year Kopits flags; it is roughly the same today. (Once again, the bulk of oil consumption is in industry, not transport.) This translates to about 14 mb/d in China. This is a bit harder to reconcile with the EIA numbers, since there will be some additional consumption from other sectors, but it is far more compatible with them than with Kopits’s 55 mb/d projection.

There is actually one scary looking data point: the United States. Kopits observes that U.S. per capita oil consumption is 2.5 gallons per day, and notes that, geographically, China is much more like the United States. U.S. oil consumption is actually is dominated by transportation. The EIA projects that per capita Chinese GDP in 2020 will be about $17,500 (in 2008 USD).  The United States was at that level around 1985. At that time, daily U.S. oil consumption for transportation was about 1.75 U.S. gallons per person. That does indeed scale close to Kopits’s projection for China. This suggests that it will be economically unrealistic for China to follow the U.S. path – high oil prices would force them toward a model more like Korea’s or Japan’s before they could get very far.

Bottom line? If you want to project the worst-case impact of wealthier Chinese lifestyles on Chinese oil consumption, Kopits is right that you need to focus on transportation, since there are currently no substitutes available at scale. But you need to do that correctly. The other sectors – industry, electricity – generally have substitutes that will be adopted if oil becomes scarce (and hence expensive). The EIA may be underestimating Chinese oil demand growth – god knows it’s been wrong enough times before – but the experience of Japan and Korea doesn’t lead to that conclusion.