THE Australian Bureau of Statistics has stopped publishing "trend estimates" in its regular economics publications, saying they would be "misleading" because of the "significant" impact of the COVID-19 pandemic.
Bruce Hockman, chief economist with the ABS statistical services group, said yesterday that other nations including New Zealand, Canada, the UK and the Netherlands were also suspending their trend data during the coronavirus epidemic.
Mr Hockman said the ABS suspended trend estimates for retail trade in 2009 because of "the disturbance to spending caused by government cash payments to households" during the global financial crisis.
This time, the impacts of COVID-19 are "much more pervasive".
Mr Hockman confirmed that notes to the recent round of March quarter data had been the only official notification of the changes but said a more detailed explanation of the situation should appear on the ABS website later in the week.
Asked whether the changes could appear to minimise the economic impact of COVID-19, and so be used politically, Mr Hockman said the ABS was "fiercely independent" and the decision to suspend the figures had been suggested internally and had not come from the government.
He said suspending trend estimates would not affect "original" data or the "seasonally adjusted" versions.
He said "time series" data had three parts: the "trend" or medium to long-term direction, the "seasonal" component with systemic, calendar-related movements, and the "irregular" or "residual" short-term component, which captured unpredictable impacts.
"The trend captures medium to long-term movements in a time series without calendar-related and irregular effects, and reflects the underlying level of the series," Mr Hockman said.
"It results from such influences as population growth, price inflation and medium to long-term economic changes."
Mr Hockman said trend data was regularly revised backwards in time when each new set of data was collected.
These "weighted" figures were compiled in Australia using a measure known as the "Henderson moving average", which gave the most "weight" or emphasis to the most recent data to "allow for the fact that the future has not occurred".
He said the duration of the pandemic was key to whether it became part of the "trend" or not.
"Seasonally adjusted series, though, contain 'trend' and 'irregular' components and so don't need a decision as to which component the COVID-19 impacts are assigned to," Mr Hockman said.
But if the impact went for longer than a year, seasonally adjusted figures might also need to be reviewed.
IN THE NEWS:
