You are here
Data errors skewed trade figures
ERRORS in trade data collection meant that International Enterprise (IE) Singapore wrongly reported two months of exports data, with possible implications for fourth-quarter GDP estimates.
October 2013's non-oil domestic exports (NODX) was said to have grown 2.8 per cent, when in fact it had shrunk 2.7 per cent. Data for September was also overstated - NODX was initially said to have shrunk 1.2 per cent when the actual contraction was a larger 2 per cent - due to the "multiple counting of some trade permits".
As trade data for both months have been corrected downwards, total trade and NODX for the full year 2013 will now come in lower than expected, IE said in an annex to its trade report for December, released yesterday.
IE will only announce Singapore's full-year trade data next month, but UOB economist Francis Tan estimates that full-year NODX would have dropped 5.4 per cent, taking September and October's erroneous figures, but could now fall a sharper 6 per cent. Both are worse than IE's forecast of a NODX contraction of 4 to 5 per cent for 2013, last revised down in November.
Market economists who follow Singapore's trade data closely expressed shock. "It has to be once in a blue moon that such things happen," said Mr Tan.
But "no great damage was done", said Barclays economist Joey Chew. "After all, the October red herring of a recovery was quickly refuted the very next month when November exports fell sharply, indicating that Singapore exports are clearly not yet out of the woods. The continued slump in electronics in December further confirmed that," she said.
IE confirmed that trade data for November, released in December, was accurate.
UOB's Mr Tan thinks it a "blessing in disguise" that the errors occurred close to the end of the year and only over two months. It would be quite different if the errors had occurred in January and February and were discovered in May, as that could change the trajectory and outlook for the year more significantly, he said.
Citi economists Kit Wei Zheng and Brian Tan wrote in a note that although the two months' revisions suggest on face value that Q4 GDP may also be adjusted down, exports data could and does diverge from industrial production. But downward revisions in the non-oil re-exports will probably hit the transport and storage services sector's contribution to GDP, they said.
IE said yesterday that the errors were traced back to changes to a trade declaration system known as Access, which is used by four air express companies to declare their consolidated imports and exports. In August last year, changes were made to this system to allow the companies to make amendments to their trade permit records, such as flight details.
However, all amended permits were counted as new ones when transmitted from the Access system to the Singapore Customs' Trade Statistics System, and then to IE Singapore. In nominal terms, the counting errors meant a difference between an originally tabulated NODX value of $15.599 billion for October, and a corrected value of $14.757 billion.
In response to BT's queries, IE explained that the over-reporting was not immediately apparent as the values of the individual records still fell within the expected range. "When unusually large numbers were picked up, IE Singapore worked with Singapore Customs immediately to investigate and rectify the issue," IE said.
DBS economist Irvin Seah thinks internal processes need to be tightened when it comes to collecting official data. "We have seen quite significant revisions, not just in NODX, but also in the advanced GDP estimates. Whether these are estimates or actual figures, there ought to be as little revision as possible. These numbers are important to everyone who wants a good gauge of where the economy is going, not just economists," he said.
For trade data, Singapore Customs conducts selective checks of trade permits against the commercial documents to verify the accuracy of data submitted by traders. "IE Singapore also conducts checks on a monthly basis to track trends based on the value of goods and large ticket items. Export and import categories with significant data swings will be picked up for further verification and analysis in consultation with Singapore Customs," IE said.
As for whether these errors undermine the reliability of Singapore's statistics, Credit Suisse economist Michael Wan said that he sees them as inherent to the "messy affair" of collecting data. "I don't think it raises questions about the integrity of Singapore's statistical collection fundamentally. It's always an ongoing affair to reduce the number of errors," he said.
UOB's Mr Tan said: "The good thing is that they are at least signalling that they are doing the right thing, by coming out and correcting the errors."
A couple of errors ought not to affect credibility, said Barclays' Ms Chew. "Especially if the errors are due to technological problems rather than data collection issues, or people gaming the system - for example Chinese exporters reporting fake trade."
She does have other issues to raise about Singapore's data though. "First, the timeliness. We are one of the last to report CPI (consumer price index) in the region, and I don't understand why. Also, IE Singapore does not release a lot of the export data they collect," Ms Chew said.