However, like a lot of people, I am pretty uncertain about the accuracy of Chinese statistics given that the political system there seems to have a lot of incentives for local officials to distort the numbers, and lacks the free speech rights that tends to correct these kinds of distortions over time. However, the question is, are any such inaccuracies material? For my purposes, we can live with numbers that are off by 10% or 20%, as long as the growth rates are about right. But obviously if they were off by an order of magnitude, or the growth rates are wildly wrong, then we will draw seriously inaccurate conclusions from them.
One way to check the accuracy is to look for internal consistency of different kinds of numbers. I discovered that the Wikipedia entry on the Automobile Industry in China has statistics of the number of cars produced in China, as follows:
|Year ||Production (in million units) |
|1992 ||1.0 |
|1999 ||1.2 |
|2000 ||2.07 |
|2001 ||2.33 |
|2002 ||3.25 |
|2003 ||4.44 |
|2004 ||5.07 |
|2005 ||5.71 |
|2006 ||7.28 |
|2007 ||8.88 |
|2008 ||9.35 |
In addition, the text of the Wiki entry provides an estimate of 13m cars for 2009. Just to show what these numbers look like on a graph:
Clearly, they also show a very rapid growth. China has apparently built a US sized car industry in about a decade.
We can compare these numbers to the changes in the size of the Chinese car fleet. Obviously, we wouldn't expect a perfect match - some cars from the fleet are scrapped every year, some cars produced in China may be exported rather than swelling the size of the Chinese fleet, while other cars produced elsewhere could be imported. Still, we would expect at least a loose relationship, and that's what we find:
I take this as confirmation that the fleet growth numbers are, if possibly imperfect, at least in the right ballpark.