Commodity demand and the China factor

A recent presentation of the Macquarie Bank contains very interesting data charts on world commodity demand and some discussions about the impact of the China factor.
The first chart shows that the trend in metal prices finally decouple from that of OECD lead indicator in recent years. The explanation is pretty simple, as over 2 billions people joined the world labor, and hence consumer market, the indicator needs to include the urbanization of these 2 billion people.

The second chart shows that after the steel consumption (per capita) peaked in 1973, for about 25 years the growth in GDP/capita did not drive growth in steel consumption. It was only by late 1990s, that the correlation was observed again.

The third chart shows the same information for Copper and Nickel (but the axis were curiously switched). The stagnation in per capita consumption was less persistent than that for steel. One possible reason is that technology innovation (recycling and substitute materials) for steel may have played a more important role than for copper/nickel.

The fourth chart is the most interesting. There is a GDP/cap correlation line, showing China is about 1/3 below the line in terms of per capita consumption of copper.

1. There is still room to grow for China's consumption, especially as its GDP/cap grows
2. Even more room for India and other developing countries (unlabelled dots)
3. Much more interesting observation is that for Taiwan and South Korea (and Finland, Sweden, Belgium). I believe that is due to the electronic and computer related industries, which consume a lot of copper
4. An immediate corollary for this observation is that Taiwan and S Korea are not the end consumers for copper. The consumers scatter in various parts of the world, in their export destinations (mainly the OECD). I can even bet that Japan in the 1960s stands high above the line.
5. Now come back to look at China, the "world factory". It is now quite straightforward to conclude that a lot of "China's consumption" is not consumed by China, instead, it is "re-exported" as finished goods to the rest of the world. Therefore, some of China's "apparent consumption" growth are indeed relocation of processing and assembling factories from existing processors such as Korea and Taiwan, and that the end consumer are still the same people in OECD countries -- i.e., a zero-sum transfer to China.
6. Therefore, to predict the future we need to separate China's consumption into 2 portions, domestic consumption and export consumption, because the two have very different starting bases and growth trends. Projecting the per capita growth for China is a very different task compared with that for Japan's or France's, as some may have reached the saturation levels.

A per capita study on a per country basis could be misleading in today's globalized world, especial when we analyze exporters. It is important to carefully distinguish secondary industry consumption (processing and exporting) from ultimate consumption (end consumer), in order to make an accurate projection.

The fifth chart is Macquarie's projection for 2001-2010. It shows China's consumption (red) as a portion of the world, for copper, aluminum, nickel and zinc. Macquarie projected China's consumption in 2001-2010 to be from 1/3 (Nickel) to 2/3 (Zinc) that of the world! Based on our discussion above, I believe this is definitely an over-estimate. The more likely number could perhaps be between 1/3-1/2 of the world. I am also curious about their projection of steel consumption, because a lot of the steel was indeed consumed domestically (construction and builfding) and such unsegmented trend analysis may be more accurate..

Now 1/3-1/2 is still a very high number. That is because this includes end-consumption of the OECD countries and many other which import from China. One way to verify this is to trace the historical per capita consumption of OECD countries, they most likely have decreased significantly as production was shifted to China. In particular, the US consumption may begin to decrease in mid 1990s and that for EU beginning of this century.

A much more reliable benchmark for per capita consumption of a commocidty is oil. Since the re-export for oil (and derivative products such as plastics) may distort the big picture less dramatically (except for Singapore and the Netherlands, they would be the odd one out above the correlation line). China consumes about 1/4 of the oil in the world, commensurate to its population share. (slightly higher than its population share due to its high energy consumption industries, compared with that of, e.g. California or Bangalore)

To conclude,

• Is China part of the cause for recent surge in commodity demand and price? Yes, one of the main factors.
• Is this reflected by the trade figures? Not really, the amount processed to be exported as finished goods needs to be taken into account. They are not consumed by China.
• Would China's consumption grow at the current speed for long? No. Because most of the 'transfer from Korea/Taiwan' will be completed soon and then it is about the true growth of China.
• Would China continue to have a low productivity/energy ratio (measured by GDP/fuels burned)? Yes. Because China's industries are mainly manufacturing indsutries. In fact, it will continue to be lower than that of India.
• Is China consuming all the commodities (as the Macqarie projection predicts) of the world? No. Raw metals flow into China only because the first step of manufacturing occurs in China. China's role in commodities is no different from that of Holland and Singapore in oil refinery.

---

Update: My hypothesis on copper "consumption" of Taiwan, Korea and Japan is confirmed. See this chart by a BHP Billiton Report.

• Japan's line flattens below the BHP trend line since 1993 (the rightmost 10 red points). But I would actually push 6-7 dots further left and trace the decline back to 1986/87. Strong indicator that the consumption shifted into indirect, i.e. via imported parts/finished goods, mostly from China. The vertical descent between 2000-04 (rightmost 4 dots) could represent another wave of migration of factory.
• The slope for Korea (green dots) is higher than that for Japan, and that for Taiwan (blue dots) higher than Korea's. A strong evidence that slope is inversely correlated to the size of the economy. i.e. export (of copper related products) as a % of GDP is highest for Taiwan, then Korea, then Japan when they are at the same GDP/cap level
• One can view Korea and Taiwan's numbers as the sum of two numbers, the export portion, and the true domestic consumption which is comparable to that of America or Japan
• China's (per capita consumption) line, will have a slope smaller than Japan's, because its population base is larger.
• In fact, Henandez of BHP Billiton raised the same question I raised in the upper corner of his chart, "High apparent consumption for export oriented manufacturing economies?" I feel good. :)