Statistics SA (StatsSA) underestimated the impact its change-over in data weighting would have on its software, which resulted in it delaying year-on-year producer price index (PPI) statistics.
According to Rashad Cassim, deputy director-general responsible for economic statistics at StatsSA, the linking of the organisation's SAS software gave it more headaches than anticipated.
"We put everything in place, but we underestimated the change-over," says Cassim. "Perhaps we needed a couple of months to do it instead of one, due to the complexity of the project."
He explains that the statistical body changed over from pure sales as a basis for data to include value-add. This means a differentiation has been created between the revenue a company generates from selling its product to include the cost of materials and everything else that went into producing it. StatsSA also added new weights to products, as well as changed the basket on which it calculates PPI.
Cassim says StatsSA ran on parallel SAS systems for the months of January and February, and only had a couple of days of lead time to produce the final statistics.
"When we moved to the new system, the numbers didn't look right so we had to fix the programming around it."
He says it is the first time in 10 years that the weighting system has been overhauled to this extent and admits: "We were caught off-guard."
Causing irritation
Economist Tony Twine says while he empathises with StatsSA when it comes to the sheer scale of the PPI project, the delay was frustrating for analysts.
"PPI is an indicator of general direction of the CPIX [consumer price index minus mortgages], as well as certain components in it, such as food," he explains.
"There is a gap in the analytical framework as long as it [PPI data] is not there."
According to Twine, it does happen that big, complex projects are underestimated, but "it is disturbing when it happens to a large-scale team of numerate and computer literate people".
The PPI data is now due to be released on Thursday.
Share
