Commentary | 22 January 2016

China's hazy GDP numbers more confusion than conspiracy

China's hazy GDP numbers more confusion than conspiracy

Leon Berkelmans

Australian Financial Review

22 January 2016

Click here for the online text.

  • Leon Berkelmans

China's hazy GDP numbers more confusion than conspiracy

Leon Berkelmans

Australian Financial Review

22 January 2016

Click here for the online text.

  • Leon Berkelmans

Executive Summary

Chinese GDP for 2015 came in pretty much at the authorities' target of around 7 per cent growth. The same happened the year before, when the target was "around" 7.5 per cent. For many, this seems too good to be true. Distrust of the official figures, and the Chinese government generally, render these figures meaningless. For example, in September, 96 per cent of economists who responded to a survey in The Wall Street Journal survey said China's GDP estimates don't accurately reflect the state of the economy. Ouch.

But it is not quite that simple. Some economists have been studying the quality of Chinese economic statistics for years. One of the best is Carsten Holz, currently at the Hong Kong University of Science and Technology, who has also spent stints at Harvard, Princeton, and Stanford.

In a thorough review of Chinese data, he concluded that "the supposed evidence for GDP data falsification is not compelling". Statistical techniques designed to uncover manipulation were unable to do so. The same tests, by the way, were able to find manipulation in Greek data!

But Holz's work has its critics. Harry Wu, of The Conference Board, has published a take-down of the stats arguing that GDP growth data has been biased upwards for decades. But then there is a recent 240 page detailed tome by Dan Rosen and Beibei Bao of Rhodium Group that suggests the level of GDP is at least 10 per cent understated.

As I said, things are not that simple. My colleague at the Lowy Institute, Steve Grenville has suggested we remain agnostic about the quarter to quarter GDP numbers. For the sake of our sanity, perhaps he is right. But I'd like to slay one myth. Given Holz's work, one could make the argument that there is manipulation, but the Chinese are far too clever to be detected.

I don't buy it. First, I've been told by people who know far more than me that there is no second set of books. If the Chinese are lying to us, they are also lying to themselves.

Second, the Chinese National Bureau of Statistics is not some well-resourced agency that would be capable of pulling off a great swindle. It's a bare bones operation. NBS headquarters has around 30 staff responsible for the entire country's national accounting work. Thirty!

Evidence of this poor resourcing is everywhere, and it causes a great deal of confusion. Most recently, this week, there has been confusion about the role of consumption in China's growth. To quote the English-language release on the Chinese National Bureau of Statistic's website, "(i)n 2015, the final consumption expenditure accounted for 66.4 per cent of GDP, 15.4 percentage points higher than last year". Except that's not what they meant. Final consumption expenditure accounted for 66.4 per cent of GDP growth, not GDP.

This is a translation they get wrong every quarter. And that has caused some commentators to get the number wrong. I've seen at least one high-profile observer get tripped up this week.

Given mistakes like these, I find it hard to believe that the NBS is malevolently manipulating the data.

But that's not to say the GDP data are good. There is signal and noise involved here. Given the GDP data are released less than three weeks after the quarter has finished, and the poor resourcing of the NBS, then the noise is going to be high. So quarter-to-quarter movements in the numbers should be treated with the utmost caution.

The Chinese Premier, Li Keqiang, was right when he said GDP data were "man-made". GDP numbers are calculated by statisticians imperfectly, and there are large degrees of uncertainty around them.

Many took this to mean that the national GDP numbers were somehow manipulated, and that the only numbers that were not were Premier Li's nominated series of electricity, rail freight, and bank loans.

But Li was actually talking about GDP numbers for Liaoning, a province in China where Li was party chief. The problems with provincial level GDP numbers are well known, and the NBS discounts them when formulating national GDP.

Internationally, China receives the most ire for the quality of their GDP data, but all countries have their problems.

For example, in September's Monetary Policy Report, the Reserve Bank of India took the extraordinary step of pointing out that GDP growth in India could be substantially overstated. These problems do not arise from some intention to artificially boost growth rates, but rather arise from the difficulties in measuring real growth in the service sector. This is a challenge that is also of particular importance to China at the moment.

If China's growth rates are overstated, it would likely have to do with similar issues. But I think it's a bit unfair to single out the Chinese, while the Indians are likely just as guilty.

 

Leon Berkelmans is director of the international economy program and G20 Studies Centre at the Lowy Institute.