Although China's official estimates of GDP growth are widely believed to be exaggerated, no one knows the true extent of the overstatement. Official almanacs such as the China Statistical Yearbook routinely publish inconsistent figures, with no explanation for the discrepancy. Since 1998, when pressure to overcome the effects of the Asian crisis led to what Chinese officials have called a 'wind of falsification and embellishment' in local economic data, the National Bureau of Statistics (NBS) has not relied on provincial-level GDP figures to compile its own nationwide estimates. The alternative methods it now employs to calculate growth have never been disclosed.
In 2002, echoing the pattern of previous years, every province claimed to have grown faster than the national rate of 8.0 percent.
The China Statistical Yearbook publishes the provincial growth estimates without comment, despite the fact that their weighted average in 2001 implied a national growth rate of just under 9.5 percent, compared with a published average of 7.3 percent. The NBS has publicly acknowledged that local statistical data contains a high 'water content,' and the reduction in the implied national rate clearly reflects an attempt to wring out low-level exaggeration.
Problems like these are hardly surprising in a country like China, challenged both by administrative limitations and by intense political pressure to report progress in local development. What makes the Yearbook's GDP numbers truly puzzling, however, is that even the (presumably exaggerated) figures for provincial growth have already been modified, since they are not consistent with the figures for absolute GDP by province in 2000 and 2001.
In other words, the process of wringing 'water' from GDP growth statistics seems to involve two stages: first amending the reported growth rates of each province, and then publishing a nationwide figure that is lower than the weighted average of the amended rates. This complicates matters considerably, but some general trends are apparent.
In both 2000 and 2001, the official figure for China's GDP was lower than the sum of the GDPs reported by each province. The nationwide figures for both years are consistent with the official year-on-year growth rate of 7.3 percent, despite provincial data implying a higher weighted average. In arriving at the national figures, the NBS apparently 'discounted' the provincial sums by just over 8 percent in 2000 and just over 10 percent in 2001, implying that the provinces' absolute figures may be losing accuracy over time. This is mathematically inevitable: if GDP growth is exaggerated year after year, even from an initially accurate baseline, the overstatement of absolute GDP must increase continuously. The higher 'discount' in 2001 thus looks like an attempt to keep the national GDP figure roughly in line with reality, while provincial GDP estimates continue to diverge.
But where do the published provincial growth rates come from, since they do not even follow from the provinces' numbers for absolute GDP? One striking feature of the absolute figures is that when summed for 2000 and 2001, their ratio implies a nationwide growth rate of 9.83 percent. This is only slightly higher than the weighted average of the growth rates actually published for each province (9.46 percent). In other words, the 'wringing out' of estimated GDP growth has been applied only to produce the published nationwide figure of 7.3 percent, not the published provincial ones. The amendments to provincial growth estimates, though very large in many cases, do not substantially reduce the weighted average.
What effect do these amendments have, then? The chart plots each province's 'implied' GDP growth (reported year-on-year increase in absolute GDP) against the size of its 'growth discrepancy' (the difference between this figure and the published growth rate). It shows that the apparent regional variation in growth rates has been systematically compressed, with published rates increased for slower-growing provinces and lowered for fast-growing ones. At one extreme, Hainan province, whose absolute GDP figures implied growth of only 5.3 percent in 2001, nevertheless reported growth of 8.9 percent in the year. At the other, Tibet's absolute figures implied growth of 18.1 percent in the year, but the published figure has been reduced to 12.8 percent. The massaging of the provincial data removes almost three-fifths of the apparent variation in growth rates among provinces.
Unfortunately, there is no obvious way to determine whether these amendments actually improve the accuracy of the provincial growth estimates. There appears to be no correlation between the amended 2001 growth figures and reported growth in provincial electricity consumption.
As if this weren't enough reason for scepticism, published statistics on electricity demand also show signs of political tampering. The Yearbook asserts that China's electricity consumption grew by 9.2 percent in 2001, even though summing the provinces' absolute figures implies total growth of just 7.9 percent. This suggests that the NBS may have raised the nationwide figure to forestall criticism of its GDP numbers by economists outside China, who sometimes cite sluggish growth in energy demand as evidence that economic growth is overstated.