National Post (National Edition)
Some problems with `evidence-based' policy
One of the mantras of the second Trudeau era, along with helping the middle class and everyone working hard to get into it, has been “evidence-based policy.”
No one opposes evidence-based policy. But evidence can be a cudgel when it is used to promote a specific policy and limit discussion under the assertion that “the science is settled.” This seems to have happened with Joe Biden's proposal to raise the federal minimum wage to US$15. (All subsequent dollar figures are U.S. dollars.) Paul Krugman of the New York Times claims the science on minimum wages is settled: “new evidence came in, and it refuted old conventional wisdom.”
That conventional wisdom states that when the price of anything rises, consumers react by purchasing less and substituting other things for it. When the price of hot dogs rises, we eat more hamburgers. That's the rationale behind carbon taxes: as the tax-in price of gas rises, we drive less, switch to more fuel-efficient vehicles, maybe even buy a bicycle.
The traditional view about the minimum wage was simple: increase it and employers would reduce the hours of low-wage workers and find ways to reorganize their workplaces. Minimum wage increases would therefore lead to unemployment.
What happened to economic theory so that Paul Krugman, a Nobel laureate in economics, would applaud raising the minimum wage? Well, he cites “new” evidence emerging in the form of a 1993 paper written by David Card and the late Alan Krueger that seemed to upend microeconomic theory.
That paper was at the edge of a new wave in economic policy analysis. Rather than collecting observations and subjecting them to statistical analysis, this new style of empirical microeconomics uses experimentation. Card and Krueger exploited a “natural experiment” arising from two similar jurisdictions' serendipitous adoption of different policies. In 1992, New Jersey increased its minimum wage from $4.25 to $5.05, while adjacent Pennsylvania did not. By surveying some 400 fast-food restaurants in both states before and after the increase, Card and Krueger concluded that the increase failed to reduce employment in New Jersey's fast-food restaurants. In fact, in some instances the number of jobs increased.
But science never ends and research into the effects of minimum-wage increases has continued. Recently, the National Bureau of Economic Research, the same organization that first published the original Card and Krueger study, published an analysis of Seattle's experience. In 2015, that city increased the minimum wage for businesses operating within its boundaries from $9.47 to $11 (a 16 per cent jump) and then again in 2016 from $11 to $13 (an 18 per cent increase).
The study's authors found that the first increase was associated with a modest change in employment, consistent with the earlier Card and Krueger study. But the second increase “reduced hours worked in low-wage jobs by six to seven per cent, while hourly wages in such jobs increased by three per cent.” Those who retained their jobs experienced an increase in pay, but many also lost their employment.
Why do two careful studies come to such different conclusions?
To begin with, natural experiments, such as Card and Krueger employed, rest on the assumption that the “treatment” and “control” groups (New Jersey and Pennsylvania, respectively) are “sufficiently” similar. (By contrast, a comparison of New Jersey and Ontario would be more dubious.)
Then there's the fact that the earlier study used surveys, which are subject to all manner of error, while the Seattle research used administrative payroll information collected by Washington state. Researchers usually prefer administrative data that cover everyone rather than information from small-sample surveys.
Finally, Card and Krueger reviewed changes over an eight-month period and just for the fast-food industry, while the Seattle study looked at changes over three years and for all industries.
The Congressional Budget Office (CBO), the non-partisan research arm of Congress, just completed a detailed analysis of the Biden proposal. It echoes the conclusions of the Seattle study, finding that an increase in the minimum wage to $15 would trigger a loss of 1.4 million jobs and increase the cumulative federal deficit by $54 billion over the next 10 years. On the other hand, because some households would benefit from the wage increase, the number of Americans below the poverty line would fall by 900,000. It did not compare hiking the minimum wage with other policies to reduce poverty, however.
There are at least four lessons here. First, evidence continues to accrue and policies can never rest on a single study. Second, low-wage labour markets are local. Skilled coders may be able to sell their services internationally over the internet; retail clerks must rely on the local labour market. The effects of a minimum wage increase vary widely depending on local conditions. Third, innovation is relentless. Whether a franchise fast-food operation transfers activity such as taking orders to the consumer through an app, or a local restaurant reduces the need for servers by joining a delivery service such as SkipTheDishes, competition motivates the reorganization of the workplace to manage costs.
Finally, policy-makers need to take care with their virtue signalling. Economywide policies such as increasing the minimum wage may produce unintended consequences in local labour markets, which, after all, is where the poor people live.
EVIDENCE CONTINUES TO ACCRUE AND POLICIES CAN NEVER REST ON A SINGLE STUDY.