Writing in the National Post, MLI Senior Fellow Philip Cross says Canada needs to end its all-pervasive search for numbers.
“It is a conceit of modernity to think the world can only, or is best, understood through data”, Cross writes. “Not everything important can be measured, and not everything measured is important”.
Philip Cross, June 19, 2014
There has been much wailing and gnashing of teeth that Canada suffers from a data deficit in key areas, such as job vacancies and housing. CIBC claims Canada is “flying blind” without housing data for credit scores, mortgage delinquency rates and foreign investment in housing. A bevy of analysts say we don’t have sub-provincial data to pinpoint labour shortages, although reading any newspaper tells you they’re more likely in northern Alberta than in Gaspe.
Our society strives for evidence-based policy. But what is the evidence that more data always leads to better policy analysis? Just last week, the IMF apologized for under-estimating the strength of the U.K. economy despite a policy of fiscal austerity. “We got it wrong,” admitted Christine Lagarde, head of the IMF, as “the confidence building that has resulted from the economic policies adopted by the government surprised many of us.” Add this to the long list of growth surprises during episodes of fiscal belt-tightening, including the U.S. ‘fiscal cliff’ early in 2013 and across Canada in the mid-1990s. There was lots of data for the IMF to study, just an unwillingness to accept what they revealed. More data won’t change that.
Steve Poloz, Governor of the Bank of Canada, stresses ad nauseum how the Bank’s policy “will depend on the data.” The thought makes me shiver. To start, data are imperfect, like everything else in this world. There, I said it. Policymakers know that too, even if they regularly complain that subsequent data revisions misled them to make incorrect choices. Well, that’s what happens when you’re too dependent on data and it’s the fault of policymakers, not statisticians who warn about the pitfalls of data. For a sophisticated data user to express surprise at revisions is as credible as Captain Renault feigning shock that gambling was taking place at Rick’s Café.
More fundamentally, reality cannot be seen solely through the prism of data. It is a conceit of modernity to think the world can only, or is best, understood through data. Not everything important can be measured, and not everything measured is important. Statistics reflect aspects of reality, but also obscure and distort reality because it must be defined, inevitably an imprecise exercise, before it can be measured. Who is retired and who is employed is not a counting exercise, but is the result of meeting criteria established by statisticians. Labour shortages are contingent on the wage rate. Some economic concepts, like innovation, simply cannot be observed.
Even if data were perfect, being “data dependent” would not always lead to better analysis and policy. The economist Murray Rothbard remarked, “No scientific truths are ever discovered by inchoate fact-digging…the scientist must have a pretty good idea of what to look for, and why.” Alan Greenspan’s successful first decade heading the Federal Reserve Board resulted from his ignoring conventional measures of inflation and productivity, believing the ITC revolution was altering the economy’s potential in ways not reflected in the statistics. Without a framework to assess the most important variables, it’s easy for users to get lost in the torrent of data issued by organizations inside and outside of government.
The flood of data builds every day, as organizations learn to mine their own databases. However, more data can be harmful if it increases the noise that makes it harder to extract the underlying signal. Furthermore, in the words of Nassim Taleb, more data allows cherry-picking results that meet your fancy, which has reached an industrial scale judging by the surge of scientific results than can’t be replicated.
Don’t get me wrong. Data are fundamental to understanding the economy, but are more useful for debunking than confirming ideas. If you are entirely data-centric, you become their prisoner in an “Ottawa bubble” of officially-sanctioned data, unaware of their limitations in a world of burgeoning information flows. Front page news in Ottawa recently was former Cabinet Minister David Emerson warning civil servants that relying exclusively on Statcan’s “structured” and “cleansed” data risked irrelevance when policymakers have access to “massive amounts of information” on hand-held devices. Privy Council Clerk Wayne Wouters followed up, warning senior bureaucrats that the old model of being armed with a university degree and Statcan data no longer guarantees a monopoly on policy advice when think tanks, advocacy groups and lobbyists tap a broader range of ideas and data.
More data by itself won’t solve labour shortages or cradle our housing market to a soft-landing. People know where the jobs are in Canada, but a large majority of Canadians won’t relocate for a job. That’s the real problem, and its solution is not more data but getting people to move. Similarly, sound lending policies and regulations that avoid obviously risky borrowers allow Canada to avoid a U.S.-style meltdown in the housing market.
Ultimately, the thirst for more data can never be quenched, since it’s rooted in the mistaken but entirely human belief that increased data makes the world more understandable, decisions less risky and the future less uncertain. The impossibility of being able to predict the future is why calls for more data will never cease, but should not be heeded when Big Data is mushrooming while thoughtful analysis is lacking.
Philip Cross is a Fellow at the Macdonald-Laurier Institute and the former Chief Economic Analyst at Statistics Canada
