Too much faith in data encourages overconfidence in our understanding, sometimes disastrously.
One of the public mantras of the Trudeau government is the primacy it attaches to evidence-based policymaking. Never mind that evidence for some of its core policies is non-existent, such as that middle-class incomes need buttressing or pension incomes are at risk. There are many reasons to be skeptical that policies based on evidence produce better outcomes.
Evidence is rarely unambiguous. Evidence can rule out bad ideas, but usually is unclear about affirming the correct answer. Karl Popper expressed best that rather than confirm something is true, empirical observation can only prove a theory false — famously, data showed all swans were white, until westerners arrived in Australia and discovered black swans. Science progresses funeral by funeral, by disproving, not by proving, results. Does the economy perform better under a monetary policy regime that follows rules or allows discretion? The evidence is mixed, swinging back and forth over time.
Proponents of evidence will say we lack data, that collecting a little more will solve the problem. This leads to the almost comical promise that the solution to all our problems is always just around the corner; not this corner, that one. No, not that one, the one over there. Statistical empires have been built on such seductive promises. Data are always imperfect, and a data-driven interpretation of our society woefully incomplete.
At its worse, too much faith in data encourages overconfidence in our understanding, sometimes disastrously. The 2007 global financial crisis is a perfect example, with banks assuming they properly managed risk based on supposedly sophisticated models of post-war U.S. housing data and assumptions about the distribution of risk. This mistake was compounded by central banks drinking Ben Bernanke’s Kool-Aid that a Great Moderation in economics had been achieved by their focus on inflation (narrowly defined to exclude asset prices), while ignoring financial stability. Another example of policy misled by data is the Phillip’s Curve trade-off between inflation and unemployment. Milton Friedman described in 1968 why its key empirical finding was based on a faulty view of the world, and then watched as both inflation and unemployment soared in the 1970s. Good theory and judgement is more important than good data.
Evidence does not answer moral and metaphysical policy questions. Abraham Lincoln warned against “contrivances such as groping for some middle ground between the right and wrong” when debating slavery. Great presidents like Lincoln and Ronald Reagan have an acute sense of right and wrong; bad ones like Herbert Hoover appoint endless fact-finding commissions. The eugenics movement was the result of scientism and rationalism unchecked by morality, proof in William F. Buckley Jr.’s words that “disembodied from moral precepts, thought is misleading, empty, ugly.” Judgement and values do not come from facts, but from examining one’s conscience and our collective wisdom.
Unsurprisingly, civil servants are infatuated with evidence-based policy, slavish followers of what former federal Deputy Minister Ruth Hubbard calls “the cult of quantification.” This is partly because it absolves them of responsibility for bad outcomes but also for more insidious reasons. President Dwight D. Eisenhower famously warned of a military-industrial complex; less well-known is his caution in the same speech against a “scientific-technological elite” of government-funded experts usurping power from the citizenry.
Insisting on evidence-based policy-making transfers power to those able to assemble and analyze data, effectively excluding the public from decision-making. Toronto historian Frank Underhill warned of the “new Machiavellianism” of a managerial class that treats public opinion as just another factor to be managed rather than as the underpinning of political authority, making democracy “a subject for mockery amongst the sophisticated.”
The evidence-based approach to policy-making puts process ahead of results. The goal should not be evidence-based policy, but policies that produce good outcomes. The latter can be achieved in many ways. Our society unquestionably has improved in recent years. However, this is largely due to spectacular innovations in technology and communications coming from the private sector, not better public policies.
Even if the evidence pointed unambiguously to an ideal policy, conservatives emphasize gradual implementation. Abrupt changes confuse businesses, households and even the bureaucrats who implement them. Evidence-based policy makes no allowance for transition costs. Believing large government budget deficits are a mistake does not justify trying to eliminate them overnight.
It is a fundamental tenet of conservatism that all the important truths are known to mankind, revealed by millennia of trial and error. Far from being anti-intellectual, this is the ultimate homage to human intelligence. Over centuries, we have filled a treasure trove with practical wisdom, habits, manners and practices to draw from. Results teased out of data by today’s researchers are transitory and trivial compared with long-standing truths.
President Franklin Roosevelt said that when presented with a consensus — even among his so-called Brain Trust of close advisers – one should defer a decision because people clearly are not engaged in critical thinking. We should suspend our fixation with evidence-based policy for the same reason. Endlessly repeating our slavish devotion to evidence has muffled the importance of good theory, judgement and moral values as the true basis of policy aimed at improving our lives.
Philip Cross is the former chief economic analyst at Statistics Canada.