I'm a health and fitness nut, so I read a lot of health studies as well as using my own body as a guinea pig. I've discovered a lot of things - for instance, I don't know if using creatine monohydrate leads to more strength gains than using a placebo, but I do know it makes you pee every 15 minutes.
Reading health studies (or even worse, hearing reports of health studies) can be incredibly frustrating as they often appear to be contradictory. Caffeine is good for you. Caffeine is bad for you. Ephedrine and ECA stacks are dangerous. Ephedrine and ECA stacks are effective and safe to use, etc. What to believe?
There are dozens of studies that show increasing the minimum wage has an adverse effect on employment. However, Card and Kreuger (1993) and an excellent paper shared by Erin Weir, Dube, Lester, and Reich (2010) show that there is no reduction in employment when the minimum wage is raised. How can we possibly reconcile this?
There are a few ways. Back in 2006 Stephen Gordon wrote a blog piece When the minimum wage bites which looked at the work of Pierre Fortin (and others) and concluded:
So what have we learned from all this?
- When minimum wages are 'low' - say, less than 40% of the average hourly wage - then moderate increases won't have a significant short-run effect on employment.
- When minimum wages are around 45% of the average, they significantly reduce employment.
That is one explanation - that the studies that show no adverse effect had minimum wages below the 40% threshold. I believe there may be another explanation. In Three Changes Making It Difficult To Find a Low Skill Job I discuss the three substitutes for minimum wage labour:
Find a technological solution. Rather than hiring another worker, is there a piece of equipment or machinery that will do the job, or make existing workers more productive so another person is not needed? The automated checkouts in grocery stores might be a good example, though I've often wondered how much labour they truly save.
Outsource/offshore/subcontract. Instead of doing the work in-house, can we pay someone else to do it? Off-shoring has become easier than ever for even the smallest of businesses thanks to services such as eLance and Mechanical Turk.
Do without. Instead of having 8 cash-registers open, can we get away with having only 7?
Both the Card and Kreuger (1993) and the Dube, Lester, and Reich (2010) have something in common - they both consider the fast food industry. Suppose the minimum wage is raised in that industry. What substitutes do they have? They clearly can't outsource the work to India. There may be some technological solution, but it is not obvious that there are large solutions. Can we replace a cashier with a robot? The only one of the three that obviously applies is to do with less labour.
Contrast this with, say, an incoming call centre, where all 3 apply. You can outsource the entire operation to India. You can build systems where algorithms attempt to answer your questions so a human does not have to (in other words, hire Watson's kid brother), or you can just have fewer staff and have your callers on hold longer.
I cannot think of too many minimum wage industries that have less substitutes than fast-food (and restaurants in general). If you were looking to find a situation where employers had to just 'suck up' the increase, that would be a good choice of industry. Now, I don't believe that the industry was chosen by Card and Kreuger (1993) or the Dube, Lester, and Reich (2010) for this reason. A good rule of thumb in economic research - if you find an assumption that looks strange in an applied paper, it's due to data availability. If you find an assumption that looks strange in a theory paper, it's due to mathematical tractability.
Now the Dube, Lester, and Reich (2010) team do consider this issue:
Do our ﬁndings carry over to affected groups other than restaurant workers? Although we cannot address this question directly, the results in a companion paper (Allegretto, Dube,& Reich, 2008) using the CPS suggest an afﬁrmative answer. In that paper, we ﬁnd that allowing spatial trends at the census division level reduces the measured disemployment level substantially when we consider the response of teen employment to minimum wage increases. Additionally, and parallel to our ﬁndings here, we ﬁnd that the measured disemployment effects disappear once we control for state-level trends in the underlying teenage employment. This evidence suggests that our ﬁndings are relevant beyond the restaurant industry.
Unfortunately the 2008 paper is unavailable online. I'm not as confident that the results are universal as Dube, Lester and Reich (and then there's Pierre Fortin's 40% argument to consider). Since teen labour is over-represented in the fast-food industry, the two studies may be largely measuring the same thing.
I believe we can reconcile these seemingly contradictory results. I hope we one day do to everyone's satisfaction, as we will have learned a lot about the functioning of labour markets in the process.