Overestimates of growth in 2007 and previous years would further reduce the current size of the Chinese economy.

SCMP calculations show the adjusted nominal GDP level in China is about US$11.5 trillion using current exchange rates, still more than twice the size of Japan’s economy at US$5.16 trillion, but well below the economy of the United States at US$20 trillion.

“Our estimates suggest that the extent by which local governments exaggerate local GDP accelerated after 2008, but the magnitude of the adjustment by the NBS did not change in tandem,” the authors said.

The study focuses primarily on nominal, non-inflation adjusted growth.

The paper comes at a sensitive time for Chinese policymakers, who are battling a slowing economy due to their campaign to reduce debt and risky lending as well as the effect of the trade war with the United States. The inflation-adjusted growth rate of 6.6 per cent last year was the slowest since 1990.

The paper’s four authors – Chen Wei, Chen Xilu and Michael Song from the Chinese University of Hong Kong and Chang-Tai Hsieh from the University of Chicago – used a mix of economic indicators that are less likely to have been manipulated by authorities to prove that the National Bureau of Statistics (NBS) have not done enough to correct the errors in the data collected from provincial governments over the past decade.

It has long been believed that local Chinese officials inflate figures reflecting their economic performance, which is closely tied to their opportunity for promotion. Since 2003, the NBS has produced a national gross domestic product (GDP) figure that is lower than aggregate provincial data after examining other data such as the census and land sales.

Local statistics bureaus generally overstate industrial output as a portion of overall production as well as the size of investment within overall expenditures, the two different approaches to calculating GDP, according to the paper. The methods of data collection are often the cause, for example, calculations of investment spending have been based purely on government reports on specific projects rather than on the financial statements of the investing firms involved.

One method that the authors used to probe the accuracy of the NBS’s adjustments was comparing the growth of official GDP with the growth of revenue from value-added tax (VAT), which taxes the value added to a product at each stage of production.

Local governments have fewer incentives to manipulate VAT revenue, since a large portion of it is eventually transferred to the central government, therefore overstating VAT would only increase fiscal revenue losses.

The authors found that since 2008, the official growth rate for industry and other sectors exceeded their corresponding VAT growth rate, with the gap widening over the past decade, indicating that the government was overstating official GDP.

In other words, the overstatement of official growth has worsened since 2008 and NBS’s corrections have been increasingly inadequate to offset bottom-up data exaggerations.

A similar conclusion was drawn when the authors examined and adjusted the official GDP growth data with a set of alternative indicators, including satellite images showing lights at night, national tax revenue, electricity consumption, railway cargo traffic, as well as imports and exports that are less likely to be over-reported, although these proxies did not fully capture the growing importance of the service sector in the economy in recent years.

Among all 31 Chinese provinces, Guangdong, Zhejiang, Beijing, and Shanghai appeared to have the highest data quality, based on the economists’ own calculations that were privately confirmed by NBS officials. The worst performing provinces included Tianjin, Liaoning and Inner Mongolia, all of which have been exposed as having exaggerated economic data in recent years.

The economists suggested that the problem is that much of the underlying data needed to project GDP is outside the NBS’s control, even though the agency has been trying hard to collect local data itself. At the same time, the NBS is also in a weak political position to confront local political leaders to demand better data collection.

“Although the NBS adjusts downwards local statistics, it does not report the adjusted local statistics, perhaps out of a desire to not confront powerful local leaders,” the authors said.

Since September, the NBS has named and shamed local governments on its website for manipulating data, but it remains to be seen if local governments fall in line.

In a post in January, the NBS said it had passed 14 cases of data falsification on to local governments before February 2018 but that it had not been updated even though local officials are required by law to punish those responsible for manipulating data within six months after receiving a notice of a violation.

The NBS’s ability to fix China’s GDP data problem is bound by its limited political power, the authors indicated.

“There are three problems with China’s GDP. One is that it doesn’t necessarily measure the right thing. Two is statistical bias in the way data is collected. Three is really a macro policy problem by the government which should write down all the bad debt,” said Michael Pettis, professor of finance at Peking University.

“The NBS is only trying to fix the second problem.”