Munir A. Sheikh is a research professor at Carleton University and a former chief statistician of Canada.
Canada has huge gaps in our data. That’s the big takeaway from The Globe and Mail’s notable series examining the state of Canadian data, which tells us that we lag behind some other countries, particularly the United States; that these gaps exist because of constrained funding and Statistics Canada’s bureaucratic, secretive mindset; and that these gaps are having a negative effect on our decision-making.
Yes, Canada has problems. But then, who doesn’t?
Citizens and governments around the world make millions of daily decisions on a vast array of issues, and each of these can potentially benefit from more data. The existence of gaps, therefore, is a virtual certainty anywhere in the world. The much-lauded U.S., for instance, does not produce detailed monthly GDP data, while Canada does, and many experts and statisticians feel that Canada’s important GDP data are of better quality than those that the U.S. does produce and require fewer revisions. Canada is also one of just a handful of countries that produced financial flow accounts, which allows policy-makers to better understand the nature and economic impacts of the 2008 financial market crash.
Canada also does an extraordinary job in producing high-quality census data at a much lower cost compared with many countries, thanks to innovations like sampling in census and being among the first to use the internet to gather citizens' responses. Canada can also boast of higher survey-response rates in many areas than the United States. And all this despite having roughly a tenth of the resources available to the U.S. federal statistical system.
That certainly doesn’t mean all is good and well here. We face serious challenges when it comes to acquiring the highest quality and most relevant data. The quality of data deteriorates automatically as the country evolves amid forces like the ongoing tech revolution (e.g. using cell phones instead of land lines) and efforts to gather survey responses suffer. Data also becomes less relevant over time as the country’s needs begin to differ from the available information. For instance, we continue to produce a disproportionately large quantity of data on manufacturing than on the services industry, even though services now represents two-thirds of the economy. And Canada’s long-form census was, for a time, replaced by a voluntary survey that produced all the information the longer census would have accumulated but with lower quality – and a higher cost, to boot.
In my view, this has produced data gaps in census information, but bad data may be more dangerous than no data at all, since giving credence to bad information can lead to bad policy. The debate around data would be most productive if it’s framed around both quantity as well as quality, which would enable policy-makers and Canadians to deal with the most pressing national issues in an informed way. On this count, Statistics Canada has struggled, as do many others.
There are three things that can be done to proactively deal with data deterioration. First, the government can increase funding for Statistics Canada to close the most important gaps that exist now, including information that measures the digital economy. Secondly, Statistics Canada should, over time, reallocate resources from less-needed data to those that are more important. Despite its efforts, the agency has not been able to establish an effective resource-reallocation mechanism, because it has had to bend many times to the users of existing data. Users of any data become vocally unhappy if theirs stops being collected. Lastly, Statistics Canada should tap new data sources and new ways of collecting information that can replace or augment existing methods.
Indeed, on that last front, Statistics Canada’s paranoia around confidentiality and privacy makes its brass gun-shy in acquiring or sharing new data with researchers. I witnessed it firsthand. Despite best efforts during my tenure as chief statistician, confidentiality concerns made it a slog to make more business-sector microdata available. But while Statistics Canada’s record of privacy-preservation and confidentiality is excellent – better than many of the most sensitive institutions in the U.S. (we have not endured crises like Wikileaks or the Pentagon Papers) – those issues have thwarted attempts to maintain data quality. Through politicians' invocation of the bogeyman of privacy to try to kill the long-form census and a Global News report that exposed its requests for Canadians' detailed financial-transaction data, Statistics Canada ironically finds itself in a lose-lose situation – criticized for its poor dissemination of data because it is so concerned about privacy, and denied access to new sources of data because privacy concerns have bred mistrust.
But the institution itself might just provide the way forward. Recent amendments to the Statistics Act established a Canadian Statistics Advisory Council to support the minister and the chief statistician, and this council should be tasked with convincing data users that certain resources should be allocated better. By playing an oversight role on privacy and confidentiality issues, too, the council can earn the trust of Canadians who, knowing that their data are safe and secure, might be more giving with their information for the national good.
There are certainly ways to improve Statistics Canada. But if collecting data is all about getting the whole picture, we can’t lose sight of what we’re already doing well.