Natural Experiments Make the Case for Liberty
One of this year’s Nobel Prize winners in economics inadvertently created a pro-liberty methodology.
Cleo McVicker made soap in Cincinnati. Then came the Great Depression. By 1933, his company reeling, McVicker pitched the Kroger grocery store chain on a new product: a cleanser designed specifically to clean wallpaper without damaging it. Kroger said yes, and McVicker produced a squishy compound of water, flour, and salt. It did the job, though it was hardly a commercial sensation.
By the 1950s, households were rapidly switching from coal furnaces to oil or gas. Sooty wallpaper was no longer a pressing concern. Desperate to salvage the family business, Cleo McVicker's son Joseph learned from a relative that schoolchildren loved playing with the squishy cleanser. He reinvented the company. Play-Doh remains a bestselling toy to this day.
Invention switcheroos are surprisingly common. Viagra first arose, so to speak, as an angina medication. Teflon was used in artillery fuses and nuclear weaponry before it made its way to the kitchen. The Internet was a conduit for military contractors and academics to share information long before it became a conduit for peddling Play-Doh, Viagra, and non-stick frying pans.
Ideas get repurposed, too. This year's Nobel Prizes in economics were awarded to David Card of Berkeley, Joshua Angrist of the Massachusetts Institute of Technology, and Guido Imbens of Stanford. The Nobel committee honored them for modeling and popularizing the use of "natural experiments" in economic analysis. Card, in particular, essentially made his reputation by co-writing a widely cited paper on the labor-market effects of the minimum wage.
The issue had, of course, been studied extensively prior to the 1994 publication of Card and Alan Krueger's paper in the American Economic Review. What made their approach distinctive was that rather than construct an elaborate national model with dozens of variables that may or may not be truly independent of each other, Card and Krueger capitalized on a natural experiment. Two contiguous states with many common characteristics, New Jersey and Pennsylvania, made different policy choices. The former raised its state-mandated minimum wage. The latter didn't. The researchers found that, contrary to conventional economic wisdom, employment at fast-food restaurants went up in New Jersey, relative to Pennsylvania, after the minimum-wage hike.
Their paper has been cited thousands of times. Its conclusion, it seems, is spurious. Using a more comprehensive set of data, two other economists, David Neumark and William Wascher, ran the same "natural experiment" and concluded that fast-food employment in New Jersey had, in fact, fallen relative to Pennsylvania after the minimum-wage hike. The debate among these scholars, and about minimum-wage effects more broadly, continues to be robust.
That doesn't mean Card and the others don't deserve praise for promoting natural experiments as a useful analytical tool. I, for one, am glad they did that. Over the ensuing decades, this tool has been used many times to make the case for limited government.
Just two years after its publication of the Card and Krueger study, for example, the American Economic Review published a paper by the University of Michigan's James Hines that exploited another natural experiment, this time regarding taxation. When companies based in other countries invested in business operations in the United States, they owed roughly the same taxes to U.S. jurisdictions as American-based companies did. But in some countries, companies received tax credits to offset their tax liabilities in America, which made these firms less price-sensitive in choosing among American states with varying tax rates on corporate income. By comparing the behavior of foreign-based companies eligible for tax credits with the behavior of all other companies, Hines was able to isolate the effects of tax policies. He concluded that corporate tax rates "significantly influence" the location of foreign direct investment.
Similarly, in a 2004 study for Public Finance Review, Florida State's Randall Holcombe and Ohio University's Donald Lacombe looked at groups of counties along state boundaries. When some states changed their tax rates and their neighbors didn't, the natural experiment played out among contiguous counties. Holcombe and Lacombe found that over a 30-year period, counties in states that raised taxes experienced slower income growth than neighboring counties that hadn't raised taxes.
More recently, Terra McKinnish of the University of Colorado returned to the minimum wage in a 2017 study for the journal Regional Science and Urban Economics. McKinnish examined the effects of a 2007–2009 increase in the federal minimum wage, from $5.15 to $7.25 an hour. Because some states set higher minimum wages than the federal one and other states don't, the federal increase had the effect of compressing differences in wage floors across states, at least for a time. Using a differences-in-differences approach, McKinnish found strong evidence that in places with substantial commuting across state borders for jobs, "low-wage workers tend to commute away from minimum wage increases rather than towards them." Progressives had long claimed otherwise, citing studies with less-compelling research designs.
I'm not so naïve as to believe contentious political debates about taxes, regulations, or other public policies can be settled by a few innovative studies in peer-reviewed journals. Political disagreements tend to originate in implicit assumptions and deeply held values, not research findings. As Oren Cass astutely observed in a 2017 essay for National Affairs, political actors who claim to practice "evidence-based policymaking" are all too often guilty of "policy-based evidence-making." They play up research conclusions that ratify their preconceived notions and play down or ignore entirely any evidence to the contrary.
But over time, the accumulated weight of empirical evidence matters. And since the early 1990s, most peer-reviewed studies of economic performance among states and localities have conclusions that support classical liberal positions. I know because my colleagues and I have counted them. So while Washington's finances continue to be a shambles, quite a few governors and legislatures have embraced fiscal restraint and adopted pro-growth tax and regulatory policies. In part this was a response to effective arguments by activists and think tanks citing the kind of research I've summarized here—research that exists at greater scale and sophistication precisely because of the kind of analysis Card and his colleagues pioneered.
That such an outcome was hardly their intention doesn't matter. Indeed, I just quoted favorably an analyst, Oren Cass, who has drifted away from free-market ideas. He's mistaken about a lot of economic issues, but that doesn't invalidate his previous insights about the perils of policymaking. And though Card and Krueger's original work was championed by progressives, their method turned out to be rather handy for the rest of us.
