CHINA will test changes to the way it calculates investment in roads, factories and real estate, as part of efforts to improve the reliability of data.

The main aspects of the fixed-asset investment trial will involve changing survey respondents to companies or legal entities from those based on particular projects and altering investment calculations to be based on money spent rather than a project's progress, Ma Jiantang, head of the National Bureau of Statistics, said in a report posted on the bureau's website yesterday.

The government is struggling to win trust from investors and economists for its data, with exports and imports the latest indicators to be scrutinized amid concerns that fake invoices are inflating the numbers.

"The biggest problem with China's investment data is double counting, which makes investment look bigger than it really is," said Lu Ting, head of China economics at Bank of America Corp in Hong Kong. "The current statistics don't really give us an accurate measure of what's really going on in the economy for the purposes of macro-economic policy."

Ma said at a recent meeting in Xi'an that "we must reform and improve the current method of calculating fixed-asset investment to more accurately reflect the investment situation and provide better support" for macro-economic controls.

Xi'an in Shaanxi Province is one of the four trial areas, along with Jincheng in Shanxi, Wuxi in Jiangsu and Qiandongnan autonomous prefecture in Guizhou, according to the NBS website. The report didn't say when the trial will start or how long it will last.

The NBS has been improving data methodology and collection to provide more reliable figures for policymakers. In February last year, it started using a unified system to directly collect output, retail sales and investment data from 700,000 companies to boost accuracy and reduce manipulation by local authorities.

The NBS releases fixed-asset investment figures excluding rural households monthly on a cumulative year-to-date basis.