Data Don’t Speak for Themselves

Check out this graph, data for which were drawn from the USDA and CDC:

With a correlation coefficient of about -0.94, these data indicate that for the decade 2000-2009 there was a strong, inverse relationship existed between per capita consumption of beef and the number of suicides by handgun. That is, as fewer pounds of beef were consumed per person over the course of the decade, the number of suicides by handgun rose at virtually the same rate.

This proves that there’s a link between an individual’s meat consumption and his likelihood to commit suicide, right?

Of course not; this correlation is spurious. These data were not the result of a study that tracked the mental health and dietary habits of individuals over a decade. Rather, the graph‘s maker, Tyler Vigen, took data from the CDC and USDA and laid them on top of one another. That they correlate for so long–and so closely–is entirely coincidental.

Absent some theoretical framework with which to interpret data–that is, if we let the data “speak for themselves”–data correlations can seem to deny true principles. For instance, roughly 87,000 flights occur daily in the United States, ostensibly flouting the laws of gravity. But the laws of gravity aren’t nullified by the flight of giant metal birds with fixed wings. Rather, it’s the interaction between airspeed, air pressure, wing shape, and direction that create the lift that allows aircraft to soar thousands of feet above the earth. The laws of gravity are entirely intact, and anyone claiming that these flights are evidence that the laws of gravity have been overturned would be ridiculed mercilessly by his peers.

This principle is just as true in economics as it is in physics. For example, economic theory tells us that demand curves are downward sloping–i.e. as price increases, fewer units of a good are consumed–but some studies purport to find that raising the cost of labor by means of a minimum wage causes no change in employers’ demand for labor. If the law of demand isn’t being disproved, what’s actually going on?

At first glance, the law of demand would seem to imply that an increase in the cost of labor would induce employers to decrease the number of workers they hire. But if the minimum wage is only binding on certain individuals, and if employers don’t hire any such individuals, then employers won’t be impacted by the change. Likewise, for those making more than a proposed minimum, a rise in the legally binding wage floor won’t constitute a “raise” as their incomes will, at best, remain unaffected.

For example, according to PayScale the estimated average yearly earnings of someone who throws freight at WinCo Foods in Boise, ID, is $17,000. Assuming full-time status (2,080 working hours annually), that amounts to a wage of roughly $8.17 per hour. Idaho’s minimum wage reflects the national minimum wage of $7.25/hour, so any increase in that minimum of up to $0.92/hour won’t be binding on any of WinCo’s Boise-based freight throwers, and, all things equal, WinCo’s employment level won’t be impacted.

But such a lack of response in employment isn’t evidence that minimum wage laws have no disemployment effects, or, by extension, that the law of demand is mythical. All it would demonstrate is that we have to be more careful in determining the impacts of minimum wage hikes. Including workers who make more than the amount of a given proposal to raise the minimum wage is, at best, distortive of the results and, at worst, highly disingenuous.

Fortunately, the minimum wage is among the most studied policies in economics, so a great deal of work on its disemployment effects has already been done with low-earners in mind. A study out of Denmark [pdf], for example, assessed the impact of minimum wage increases on teenagers, for whom the minimum wage rises nearly 40 percent at age 18, by law.

The authors found that while those who keep their jobs see a significant increase in take-home pay, labor as a proportion of total input falls by nearly half and the overall employment rate for teens falls by a third as 18-year-old workers find themselves jobless–as one might expect given the law of demand. The upshot of these effects taken together is that the level of total wages paid by employers remains virtually unchanged.

But there are other ways that employment is distorted by the institution of a new wage floor. Quite often, the law of demand is satisfied in less immediately visible ways. Recent studies from Seattle, the University of Illinois at Urbana-Champaign [pdf], and New York University [pdf] provide such examples.

In the Seattle study, workers saw declines in hours that completely offset their gains in hourly pay, leaving them with less take-home pay. The NYU study found that higher productivity workers were substituted for lower productivity workers. In addition to such findings, the UIUC study also found that in low-skilled, labor-intensive industries a 10 percent increase in the minimum wage resulted in an increase of more than 24 percent in spending on capital, in line with concerns that minimum wage hikes lead to greater automation.

Other impacts from minimum wage hikes that aren’t immediately obvious include reductions in non-wage benefits, higher consumer prices [pdf], substituting increased customer responsibility for unskilled labor [pdf], higher credential requirements for would-be employees [pdf], delayed teen and minority entry into the job market, and slower job growth for as long as eight years after the increase.

In spite of (methodologically questionable) recent challenges, the decades-old consensus remains on solid empirical ground. In a review of more than 100 studies of the minimum wage in countries across the globe, less than 8 percent of found that increasing the minimum wage had any kind of positive impact on employment. About two-thirds of the studies, by contrast, found negative employment effects.

When the authors narrowed their evaluation to the best-quality studies, 85 percent of studies found disemployment effects, while “very few–if any–cases [were found in which] a study provide[d] convincing evidence of positive employment effects of minimum wages” on those most susceptible to employment displacement.

Given the weight of evidence, we should be immediately skeptical upon hearing of studies that purport to find net-zero (or net-positive) impacts resulting from minimum wage hikes. But even if the literature were murkier on the empirical relationship between wage floors and employment, such skepticism would still be warranted.

Intuitively, we know that increasing the price of a good leads to reduced consumption of that good. While the response depends on the individual consumer’s capacity and desire to consume at a given level (that is, their demand elasticity), at some point the next increase in price causes everyone* to consume less. Interpreting data within this theoretical framework allows one to forego the embarrassment of being viciously mocked by anyone possessed of even passing familiarity with economic principles.

In the end, economist David Henderson’s First Pillar of Economic Wisdom remains a true guiding principle: there ain’t no such thing as a free lunch. And, indeed, in the immortal words of Wesley, “Anyone who says differently is selling something.”

*The exception is in goods for which demand is perfectly inelastic, but, given the ease with which unskilled laborers can be found in the labor market, one would be hard-pressed to argue convincingly that unskilled labor is such a good.