STATSCAN WITHDRAWS INDUSTRIAL R&D SURVEY AFTER CONCERNS RAISED BY QUEBEC

Problems and delays continue to plague Statistics Canada's showcase industrial R&D survey following the decision two years ago to use Revenue Canada data for firms spending less than $1 million s year on R&D. The most recent survey was released in late December but was withdrawn last week after concerns of under reporting were raised by the Quebec government's statistical agency, the Institut de la statistique du Quebéc (ISQ).

The ISQ was alerted to possible problems when the industrial R&D survey reported 1997 industrial intramural R&D spending for the province of $2.422 billion. The figure is approximately 7% lower than the amount estimated Quebec in a previous survey on total spending of research and development in Canada, released just one month earlier. StatsCan statisticians must now re-examine all the data to determine where the problem lies and issue a new report in the coming months. StatsCan officials believe the survey did not under report the data to such an extent, but it could revised data could increase the total 1-2%, which would boost the value of industrial intramural R&D by $100-200 million for 1997.

The current version of the industrial R&D survey marks the first time StatsCan relied entirely on RevCan data for smaller R&D performers, including those in Quebec. The previous annual survey (released a full year behind schedule) used RevCan data for all provinces except Quebec, which is negotiating with the federal government to gain access to federal tax data. StatsCan continues to conduct its survey for the approximately 1,000 Canadian companies that spend in excess of $1 million on R&D annually.

"We are responding to concerns from the Province of Quebec that data for small firms may be under reported," says Dr Fred Gault, director of StatsCan's science and innovation branch. "They are legitimately concerned about this and they believe we have missed a number of small R&D performers."

The problem does not end with Quebec, however, and data for each province will have to be re-examined and likely adjusted, right down to the industry sector level. Small firms using RevCan's T-661 form have up to 18 months after their fiscal year end to file, creating a delay of 2-3 years before StatsCan has the necessary data to built its survey. As a result, data trickles into to the science and innovation branch, making it difficult to determine when set the cut-off date for the survey data.

"We have not had Revenue Canada data of a consistent quality long enough to do an empirical analysis of this," says Gault. "We need to understand the transition from full survey data to a combination of survey and federal tax data."

StatsCan may release a working paper to more fully explain the problem and its methodology for resolving it.

R$